David Escamilla-Guerrero
Explaining gender differences in migrant sorting: Evidence from Canada-US migration
with Chris Minns and Miko Lepisto
Project funded by the Social Sciences and Humanities Research Council of Canada
How women select into migration and sort across destinations has profound implications for immigrant integration, the labor market outcomes of second-generation immigrants, and the effectiveness of immigration policies. But despite accounting for about half of all international migration flows, we know little about the gender differences that may characterize the sorting of migrants. Models of migrant selection focus mostly on the labor market position of migrants, with differences in earnings dispersion between source and destination economies playing a leading role in determining who migrates and where migrants go. Using these models to study the migration decision of women is challenging due to selective labor force participation and the presence of significant numbers of "tied movers" among the female migrant population. This is particularly true if the analysis is extended to periods when institutions and social norms restricted the employment of women or in contexts with low female labor force participation after marriage.
In this paper, we ask whether female and male Canadian migrants sorted differently across US destinations in the early 20th century and examine a series of factors that can account for gender differences in sorting patterns. We start by introducing a conceptual framework that extends the Grogger and Hanson (2011) model of migrant sorting. The main implication of our framework is that gender differences in migration costs or rewards to skill across destinations would induce distinctive sorting patterns between female and male migrants. To empirically test this prediction, we combine county-level data estimated from the 1910 and 1920 full-count US censuses with a sample of recently digitized Canada-Vermont border crossing records from 1896 to 1924. The census data capture factors at the destination that proxy for the benefits (absolute returns to skill) and costs (employment opportunities, presence of ethnic enclaves, and marriage markets) associated with migration.
Example of Immigration card. Canada-Vermont Border Crossing Records.
The border crossing records report rich individual data including physical stature (height) at the time of crossing, which we use to measure the sorting of migrants. Physical stature is a function of childhood conditions (nutrition less disease environment and work assignments) that are highly correlated with human capital and earnings potential. The major advantage of using height rather than conventional evidence such as wages or occupation rankings is that it allows us to characterize sorting patterns among all women and men at a time when female participation in labor markets was limited, particularly among married women in the United States and Canada.
We find that sorting across destinations was strikingly different between men and women. Absolute returns to skill dominate in explaining sorting patterns among men, while for women job search costs and access to ethnic networks were of greater importance. We document that ethnic networks played a large role in mediating the costs of finding accommodation and in accessing financial services at the destination. These barriers were particularly prominent for single women due to the discrimination faced in labor, housing, and financial markets. Our analysis of married migrants shows that the sorting of married men across destination conditions largely resembled that of their single counterparts, while the sorting of married women was much less sensitive to destination benefits and costs. This paper extends our understanding on migrant sorting across destinations through what is to our knowledge the first comparison of sorting patterns between female and male migrants.
Origin and destination locations of Canadian migrants during the early twentieth century.