Abramitzky Boustan and Eriksson 2019 — Do Immigrants Assimilate More Slowly Today Than in the Past

Summary

Abramitzky, Boustan, and Eriksson compare cultural assimilation rates between two Ages of Mass Migration — European immigrants 1850–1913 and Asian/Latin American immigrants 1965–present — using children's given names as a behavioral measure of cultural identity. They find that assimilation rates are virtually identical across the two periods: immigrants reduce the naming gap with natives by roughly one-half after twenty years in the US, both in the 1920 complete-count Census and in California birth certificates from 1989–2015. The groups most accused of failing to assimilate — Southern/Eastern Europeans historically, Mexican and Vietnamese immigrants today — actually exhibit the fastest assimilation rates.

Key Claims

• Immigrants erase $\approx 50\%$ of the naming gap with native-born mothers after 20 years in the US, in both the historical and modern periods ($13$ points in 1920 Census data; $9.8$ points in California birth records).
• The "Foreignness Index" (F-index) — the relative probability a name belongs to a foreign-born vs. native-born person — declines linearly with mother's time in the US, with essentially no slowdown.
• Country-of-origin variation: immigrants who start with the most foreign names assimilate the fastest. In the past: Finnish, Swiss, Portuguese, Austrian mothers. Today: Mexican immigrants ($-0.51$ F-index points/year), Vietnamese ($-0.34$ points/year); Chinese and Filipino immigrants shift more slowly.
• Cultural assimilation is robust across all education levels, including high school dropouts. If anything, less-educated Mexican and Vietnamese mothers assimilate slightly faster (statistically indistinguishable from more-educated cohorts).
• Cultural and economic assimilation are nearly orthogonal: cross-country correlation between naming assimilation rates and occupational earnings assimilation rates is $r = 0.04$.
• Native-born mothers shift slightly toward more foreign-sounding names with age ($1.4$–$1.5$ F-index points in 20 years), consistent with cosmopolitan exposure effects — the opposite of immigrant mothers' pattern, reinforcing the causal cultural-assimilation interpretation.
• In modern California, Anglicizing Spanish names (Maria → Mary, Ana → Anna) is a key mechanism; controlling for phonetic equivalence cuts estimated assimilation in half.

Concepts Introduced or Extended

• Cultural Assimilation — operationalized via the Foreignness Index (F-index) of children's names; first rigorous past/present comparison; shows assimilation rates are stable across eras and groups

Entities Mentioned

• Ran Abramitzky — lead author; Stanford economist
• Leah Boustan — Princeton economist; co-author
• Katherine Eriksson — UC Davis economist; co-author

Quotes

"Immigrants reduce the naming gap with natives by about one-half after spending twenty years in the US, both historically and today."

"The groups most often accused of a lack of assimilation actually assimilate most rapidly."

"There is little correlation between cultural and economic assimilation at the country level."

"It would be a mistake to determine immigration policy based on the belief that immigrants will remain foreigners, preserving their old ways of life and keeping themselves at arm's length from the dominant culture."

My Take

The paper is a tour de force of historical data assembly and a clean methodological contribution. The mother fixed-effects design is elegant: by comparing siblings, they control for all time-invariant immigrant characteristics, including age at arrival. The F-index is also well-validated against alternative assimilation measures (intermarriage, English fluency, citizenship applications) in Appendix Table 5.

The California-only limitation for modern data is non-trivial — Mexico-born mothers are overrepresented relative to the national immigrant population, and the authors' own reweighting exercise reduces the estimated 20-year gap closure from $9.8$ to $8.4$ points. The conclusion that rates are "similar" across eras holds, but the modern sample is not representative of the US.

The zero correlation between cultural and economic assimilation is striking and underappreciated. It suggests the two processes respond to different incentives — cultural assimilation may be more about identity commitment and time exposure, while economic assimilation is more about skill transferability and labor market discrimination. The identity economics framework (Akerlof-Kranton) provides a coherent theoretical underpinning for why full first-generation assimilation wouldn't be expected even without constraints.