The Role of Immigration in Automation-Induced Displacement and Re-employment in the AI Era, 2014-2025

• In Event: Mathematical Sociology Section Flash Talks
  In Refereed Roundtable (60 minute): Table 12. The Role of Immigration in Automation-Induced Displacement and Re-employment in the AI Era, 2014-2025

Abstract

The rapidly increasing use of Artificial intelligence (AI) in the US economy predicts massive labor displacement. At the same time, immigration faces contentious policies and sentiments. Beyond the shrunk immigration inflows are their uncertain skill compositions. Current and future immigration complicates the changing relationships among growing tech capital, large labor displacement, and prolonged new task content creation in the AI era. This paper integrates theories on international immigration to the US with a new framework about automation-induced labor displacement, productivity growth, and labor reinstatement. Using Survey of Income and Program Participation (SIPP) 2014-2025, we analyze three consecutive 48-month panels to test whether exposures to automation will increase the probability of job loss, lengthen unemployment spells till finding a new job in non-automated industries. Furthermore, we test how immigration modifies these relationships through both individual and industrial level differences from those of natives. The proposed analysis of the recent years will feel the pulse of the current AI era, deepening our understanding on whether immigration is vital for the US future workforce.
Our preliminary analysis shows that workers in high-automation industries consistently experience more layoff days than those in low-automation industries. We also observe that immigrants face greater risks of being laid off than natives. The survivor curves for the time from being laid off to re-employment reveal that immigrants, particularly those in high-automation industries, reach re-employment faster than natives in either type of industries.
These descriptive patterns will be clarified in our ongoing research. Using event history methods and models, we will employ the hazard model for being laid off and the cox model for the waiting time before a new job entry. Both modeling types are multi-level with person-months nested in persons, which in turn are nested in industries from which the individuals are laid off.

Authors

• Lingxin Hao, Johns Hopkins University

• Linxuan Yang, Johns Hopkins University