Global education trajectories and inequality: STEM workers from China to the US

Abstract

ABSTRACTThe United States has become reliant on workers from abroad to meet its demand for the knowledge-based economy. However, some migrants may face an earnings deficit relative to similar US-born workers. This paper examines the sources of the deficit and asks whether we should expect the initial deficit to disappear with education attainment and work experience in the US. They are challenging to answer as few data sources measure and track market experiences and educational trajectories of migrants over time. Migration and educational trajectories which reflect the country's source of formal educational credentials as well as other forms of capital may explain the deficit, this study applies sequence analysis to the National Science Foundation’s ‘National College Graduate Survey’ to examine earnings differences between China- and US-born STEM workers. After identifying the dominant migration-education profiles for these STEM workers, I show that the wages of migrants with exclusively China-based education are 5–25% lower than those of workers with at least some US-based education, even among workers who are otherwise similar in terms of experience, legal status, employer type, occupation, degree level and time since migration. These findings point to significant and lasting penalties due to non-US education.KEYWORDS: Migrationeducationintellectual migrationtrajectorysequence analysis Disclosure statementNo potential conflict of interest was reported by the author(s).Notes1 College educated, either with STEM-related education or work in a STEM occupation.2 NSCG takes a stratified sampling strategy to maximize its coverage of different demographic, education and occupation groups (National Center for Science and Engineering Statistics Citation2019)3 Due to the extremely small sample size, the fifth highest degree is omitted in the analysis.4 Job-Skill Match is derived from the ‘OCEDRLP’ variable in the NSCG dataset, measuring to what extent is the respondent’s work is related to their highest degree. Higher score denotes a closer match.5 Due to data compression in the public-use NSCG dataset, the geographic unit in this analysis is limited to countries and census divisions; the institution identifier is limited to the 1994 Carnegie classification codes; citizenship and visa type are restricted to citizen/non-citizen/naturalized citizen, and temporary/permanent visas.6 Age 25–65 is selected to capture working age population with college education per ACS standard; see US Census Bureau (Citation2020) for detail.7 To reduce the complication in the optimal-matching distance calculation and trajectory clustering process, this study has adopted country code as the indicator of education location, STEM/STEM-related/non-STEM broad group codes for field of study, and 1994 Carnegie code for institution identifier.8 I have tested multiple sequence analysis fit statistics such as the silhouette score and elbow method, and the results all point to the four-type solution.9 The details of the five US-born trajectories will not be discussed here since they are not the main point of this analysis. See appendix 1 for additional detail and visualization of US-born sequences.10 I use non-linear curves here because STEM worker’s wages don’t necessarily follow the traditional age-earnings profiles. See Deming and Noray (Citation2018) for details.11 The NSCG data provide a retrospective education history of the respondent that dates back from the person’s highest to fifth highest higher education degrees. The fifth highest degree is omitted in the sequence analysis due to the limited sample size. Hence, the starting point of these visualizations should be interpreted as the respondent’s fourth highest degree instead of first degree, as they may have more than four degrees.