Foreign-Born Labor, Technology and Growth
A small VAR asks whether growth in the foreign-born share of the labor force precedes technology production and real GDP per person.
I originally put immigration into The Wealth of Nations—Redux. It made the main essay harder to follow and weakened its long historical sample. The foreign-born labor-force series begins only in 2007. Rather than discard nearly three decades of the women’s-participation evidence, I separated the immigration question and let it stand on its own.
The test
The model contains three quarterly series: the foreign-born share of the civilian labor force, industrial production of computers and electronic products, and real GDP per person. I use quarterly log changes and two lags. Differencing removes the obvious common upward trends; the VAR allows every variable to respond to its own history and to the history of the other two.
| Target | Lagged predictor | Lag 1 | Lag 2 | Sum |
|---|---|---|---|---|
| Technology-production growth | Foreign-born labor-force-share growth | +0.061 | +0.129 | +0.191 |
| Real GDP-per-person growth | Foreign-born labor-force-share growth | +0.102 | +0.083 | +0.185 |
| Real GDP-per-person growth | Technology-production growth | +0.089 | −0.031 | +0.058 |
What I take from it
All three two-quarter sums have the predicted sign. Growth in the foreign-born labor-force share precedes growth in computer and electronic production and in real GDP per person. Technology-production growth also precedes GDP growth.
The result fits a plausible story. Migrant labor need not replace domestic labor. It can complement domestic workers, machinery, research and management. A farm worker can make the farmer’s land and equipment more productive; an engineer can make a research team more productive. The economic question is not simply how many workers arrive. It is whether their skills are matched with the people and capital already here.
But this is not proof of an immigration-policy effect. The sample is short, the technology index covers production rather than invention, and a growing economy can attract workers from abroad. The current RainbowStats coefficient table also does not report standard errors. I would describe the result as supportive timing evidence—not a causal verdict.
Replication script
The earlier combined analysis remains available. The shorter companion script is reproduced in full below.
ForeignLabor=SET_NAME(LNU01073395,"Foreign-born civilian labor force") TotalLabor=SET_NAME(CLF16OV,"Total civilian labor force") ForeignLaborData=SAME_DATE_RANGE(LIST(ForeignLabor,TotalLabor)) ForeignLabor=EX(ForeignLaborData,series:0) TotalLabor=EX(ForeignLaborData,series:1) ForeignShare=SET_NAME(100*ForeignLabor/TotalLabor,"Foreign-born share of U.S. labor force") ForeignShare=TO_QUARTERLY(ForeignShare) TechnologyIndex=SET_NAME(IPG334SQ,"Industrial production: computer and electronic products") RealGDPPerCapita=SET_NAME(A939RX0Q048SBEA,"Real GDP per capita") ForeignVARData=SAME_DATE_RANGE(LIST(ForeignShare,TechnologyIndex,RealGDPPerCapita)) ForeignVARGrowth=LOGDIFF(ForeignVARData) ForeignVAR=VAR_MODEL(ForeignVARGrowth,2) ForeignVAR
Sources
- FRED/BLS: foreign-born civilian labor force and total civilian labor force.
- Federal Reserve industrial production: computer and electronic products.
- FRED/BEA: real GDP per capita.