Is immigration a good thing? Should it be restricted? Does it harm sending economies? As immigration continues to rise, public opinion is divided on these questions. Historical political economy can be deeply informative for understanding and anticipating the consequences of migration: for the receiving societies, sending societies, as well as migrants themselves.
In this post, I discuss distinct advantages of studying historical episodes of migration. For one, looking deeper into the past helps to evaluate the long-run effects of migration, which need not be the same as short-run effects. An HPE approach to migration also sheds light on cyclical patterns in the push and pull factors that affect geographic mobility. Finally, historical cases of migration are sometimes more amenable to causal inference and can offer more fine-grained measurement than contemporary cases.
The benefits of a longer temporal lens
The effects of migration typically unfold over a long time horizon and may change over time. Taking a longer perspective thus allows for a more comprehensive evaluation. A historical perspective is also helpful for tracing the intergenerational effects of migration.
This is true for both economic and political effects. For instance, Sequeira, Nunn, and Qian (2020) examine the short- and long-run effects of immigration in the US during the Age of Mass Migration (1860–1920). They find that in the short term, immigrants increased the supply of labor for industrialization and provided new skills and knowledge that increased innovation and agricultural productivity. These initial benefits have persisted over time, as evident from significantly higher incomes, educational achievement, and urbanization, as well as lower poverty and unemployment rates in counties with higher historical levels of immigration today.
Migration also affects institutions in the sending countries. Karadja and Prawitz (2019) show that Swedish municipalities that experienced more emigration to the US during the Age of Mass Migration had higher participation in labor movements and strikes, higher turnout in elections, and stronger left-wing voting preferences. Through these channels, emigration increased welfare expenditure and contributed to the introduction of representative democracy in local governments. These processes unfolded over an extended period and would be easy to miss in a study with an earlier cutoff for measuring outcomes.
The economic effects of immigration may grow or recede over time. If migrants shift the technological frontier, their presence will have greater effects on per capita income in the long run than in the immediate term. Some of the effects of migration may appear with a lag, as migrants’ citizenship status changes, or become visible only following institutional changes at the national or international level. For instance, in a 2019 article I find that the diversity of immigrant population in Western Poland contributed to higher incomes and entrepreneurship rates only following the transition to the free market, with a forty-year delay. Similarly, Burchardi and Hassan (2013) show that regions in West Germany that received more refugees from East Germany in 1949 1961 experienced growth of income per capita after the fall of the Berlin Wall, as these migrants were able to take advantage of their social ties to seize new economic opportunities in the East.
Some effects of migration are best measured in generations rather than years, which requires looking deeper in the past. For instance, immigration may be costly for parents, but their children benefit from new opportunities. This is what Abramitzky and coathors (2021) find using data on father-son pairs that starts with foreign born fathers observed in the 1880 census and are followed through subsequent censuses. They conclude that both today and in early 20th century, the children of immigrant fathers were more upwardly mobile than the children of native fathers. The mobility gap is largest for the bottom of the income distribution.
Extending the temporal lens can also be useful for understanding the cyclical patterns of migration and predicting future demographic pressures on the receiving countries. It is often debated whether providing aid to developing countries will reduce outmigration by creating more jobs at home. Looking at emigration before the introduction of various restrictions in the receiving countries can be helpful for understanding factors that affect the “emigration life cycle” (see Hatton and Williamson 2009). It appears that emigration initially rises, but then falls with development. Clemens (2020) applies a single empirical framework to all emigration worldwide 1960–2019 and emigration to the Western Hemisphere 1850–1914 to provide additional evidence for this pattern.
Understanding and addressing selection bias
Historical cases can be useful for addressing endogeneity issues. Selection bias, which can enter at different stages of migration, is one of the main concerns when estimating the economic and political consequences of in- and outmigration. It is sometimes possible to address selection by drawing on historical cases and using quasi-experimental designs. Forced migration after WWII inspired a lot of HPE research because both the process of selection into migration and the selection of destination were relatively transparent (see my earlier post).
Focusing on historical periods when migration was relatively less regulated can also illuminate the process of selection into migration itself. Whereas today the US relies on a patchwork of legal restrictions that distinguish between migrants on the basis of skills, country of origin, and family background, it maintained relatively open borders until the 1920s. This has allowed researchers to study the decisions to emigrate and to estimate economic returns to immigration (see research by Abramitzky, Boustan, and Eriksson above). For example, Connor (2019) studies selection in Ireland in the early twentieth century using data that links migrants and non-migrants back to their childhood homes. He finds that the sons of farmers and illiterate men were more likely to emigrate and that immigrant networks were important in alleviating resource constraints on emigration. These findings not only advance our understanding of the decisions to migrate during this period, but also shed more light on the mechanisms through which Irish migration affected economic outcomes in the US, where these migrants settled.
History can also serve as a repository of case studies for studying specific research questions. Studying the effects of legal restrictions on immigration adopted by the US in the 1920s can be helpful for designing contemporary immigration policy. One of many interesting findings in the research on these laws is that the quotas on immigration reduced rates of return for low-skilled immigrants (see Ward 2019). Why might this be? One possible explanation is that by reducing the inflow of new immigrants the quotas made it easier for low-skilled immigrants to succeed. Another reason is that the quotas reduced the supply of immigrants whose demographic characteristics made them more likely to return in the past.
Solving measurement problems
Importantly, historical data can be more fine-grained and detailed than contemporary data. For instance, recent data that contain sensitive information may be closed to research to protect individuals involved. Closure periods for personal data vary by country. In the US, census records are released to the public 72 years after the day of the census. This means that the 1950 (earliest) census microdata will become available only in April 2022. Microdata up to 1940, on the other hand, is freely available from IPUMS USA. In Italy, personal data that reveal racial or ethnic origin, religious and political opinions, and membership in social organizations, become accessible to the public after 40 years, while data on health and sex life after 70 years.
Access to personal data is particularly important for research on migrant selection and on the effects of migration on migrants and their children. For instance, Abramitzky, Boustan, and Eriksson (2012) use census data from Norway and US census records from the genealogy website Ancestry.com (see more on this resource here). They match Norwegian brothers by name and age, one of whom emigrated, and assign each individual the mean earnings for their occupation in either Norway or the United States. The availability of individual-level data allows them to account for migrant selection and estimate the economic returns to migration. Another study uses individual-level data to understand the consequences of the Great Migration, which brought six million African Americans from the US South to Northern, Midwestern, and Western states. Eriksson (2019) matches all Southern-born male prisoners and non-prisoners in the 1940 US Census to their childhood household in the 1920 Census to study the effects of migration on incarceration while controlling for selection at the household level. She finds that moving North has increased the probability of incarceration, particularly for recent migrants.
If you are interested in HPE of migration, don’t forget to check out these earlier posts by Emily Sellars on Mexico-US migration, Vicky Fouka on immigrant restrictions and the Great Migration in the US, and my earlier post on forced migration after WWII in Europe.