Does Immigration Increase Local Innovation and Economic Growth?

The study asks whether immigration increases local innovation and wage growth in U.S. counties. Using census, ACS, patent records, and wage data from 1975–2010, the authors link immigration shocks to changes in patenting and earnings. They find that an influx of 10,000 immigrants raises patenting by about 1.22 patents per 100,000 residents over five years (a 25 percent increase) and boosts annual wages by roughly $150 per capita (about 8 percent higher wage growth). More educated immigrants generate even larger gains. Overall, the study concludes that immigration meaningfully stimulates both innovation and economic growth.

The Policy Scientist's Perspective

This article examines a question made more salient by rising immigration restrictions across much of the world: whether inflows of migrants materially influence innovation and wage growth. The broader relevance is clear, as many countries face aging populations and slowing productivity. This article's approach improves on earlier research by using a more credible identification strategy and unusually comprehensive census, ACS, patent, and wage data. Their causal inference strategy is considerably stronger than standard regressions. The results are likely to extend to other advanced economies with comparable institutional settings.

Full Citation and Link to Article

Burchardi, K. B., Chaney, T., Hassan, T. A., & Terry, S. J. (2021). Immigration, innovation and growth. American Economic Review, 111(11), 3786–3836. https://doi.org/10.1257/aer.20211601

Central Research Question

The authors ask whether immigration has a causal impact on local innovation and wage growth in the United States, and if so, through what mechanisms these effects unfold. They are interested not only in the short-run labor-supply implications of immigration but also in the dynamic effects on idea production, productivity, and regional economic performance. Their central goal is to determine whether immigration stimulates or depresses local economic activity once endogeneity in migrant settlement patterns is addressed. Because migrants tend to choose locations with favorable economic conditions, disentangling cause from correlation is essential. Thus, the study asks whether an exogenous increase in immigration raises patenting and wages and whether these impacts are large enough to influence long-run growth trajectories.

Previous Literature

The paper builds on several strands of research. A long line of work in labor economics, exemplified by Card, Borjas, Ottaviano and Peri, and others, has examined the wage effects of immigration. These studies often rely on shift-share instruments based on past settlement patterns. However, concerns have been raised about the endogeneity of these historical shares, particularly their correlation with persistent local productivity shocks. A second literature studies the contribution of immigrants—especially high-skilled immigrants—to innovation, patenting, and scientific progress. This includes work showing large effects of foreign-born scientists, STEM workers, and inventors on U.S. technological output. A third tradition, in endogenous growth theory, models how population growth and the stock of ideas interact to determine long-run economic outcomes. Finally, recent advances in econometrics have clarified the limitations of conventional shift-share designs and the conditions under which they generate biased or over-rejected estimates. This article incorporates insights from all these areas but distinguishes itself by proposing a substantially more credible identification strategy that exploits 130 years of ancestry-based migration dynamics to isolate quasi-random components of immigrant inflows.

Data

The authors assemble an unusually rich dataset combining historical census microdata, American Community Survey samples, patent microdata from the U.S. Patent and Trademark Office, and county-level wage data from the Quarterly Census of Employment and Wages. The migration and ancestry data span from 1880 to 2010 and cover more than 3,000 U.S. counties and nearly 200 origin countries. These files allow the authors to reconstruct both historical settlement patterns and contemporaneous inflows of migrants. Patent data, available annually from 1975 to 2010, are assigned to counties based on corporate assignee addresses. Wage data include average county-level earnings for workers and can be further disaggregated for education-based analyses. Together, these datasets allow for fine-grained measurement of immigration shocks, innovation output, and wage dynamics across space and time. The breadth and historical depth of the dataset are critical for building the instrument and for tracing out both short-term and longer-horizon effects.

Methods

The empirical strategy consists of two major components: (1) construction of an immigration instrument using predicted ancestry shares, and (2) instrumental-variables estimation of the causal effect of immigration on innovation and wages. The key innovation is the ancestry-prediction procedure. Instead of using realized ancestry shares—likely correlated with persistent local productivity shocks—the authors construct instruments for each county’s ancestry composition using migration flows from 130 years of U.S. history. Specifically, they interact historical “push” shocks (country-specific migration to the U.S. outside the focal region) with “pull” shocks (the share of European migrants choosing each county). This generates a predicted ancestry distribution plausibly orthogonal to county-level shocks. The second step applies the standard shift-share logic: predicted ancestry is interacted with contemporary national inflows from each origin country to form immigration shocks.

These immigration shocks serve as instruments in regressions of changes in patenting and wages on changes in immigrant inflows. The main models use five-year differences and include time and state fixed effects, with some specifications adding county fixed effects to absorb location-specific growth trends. The authors further disaggregate effects by education level and test for spatial spillovers. They also build and estimate a structural model of endogenous growth and migration to quantify the elasticity of idea production to research labor and to evaluate the aggregate effects of immigration counterfactually. The design emphasizes causal inference and explicitly addresses limitations of earlier shift-share methods.

Findings/Size Effects

The authors find strong evidence that immigration increases both innovation and wages at the county level. Using the preferred instrumental-variables specification, a county receiving 10,000 additional immigrants (roughly one standard deviation) experiences an increase of about 1.22 patents per 100,000 residents over five years—approximately a 25 percent increase relative to the mean of 4.61. The elasticity-based specification shows similarly large effects. These increases are not driven primarily by immigrant inventors: roughly 80 percent of the rise in patenting occurs through increased patenting by U.S.-born workers, suggesting complementarity in idea production.

For wages, the estimated effect is also positive. An arrival of 10,000 adult immigrants increases average annual wages per capita by roughly $150 over five years, equivalent to about an 8 percent increase in wage growth relative to the sample mean. These results reflect both the direct labor-supply impact and the induced productivity gains from greater innovation.

The study documents substantial heterogeneity. High-skilled immigrants generate much larger increases in both patenting and wages. For example, 10,000 immigrants with one standard deviation (3.7 years) higher schooling than the average produce roughly five times the increase in local patenting compared with otherwise similar immigrants. Wage effects also increase sharply with the education level of native workers: high school dropouts experience near-zero gains, while college-educated and graduate-degree workers see the largest increases. The spatial analysis shows that immigration creates spillovers up to about 100 kilometers, raising innovation and wages in nearby counties.

The structural model yields complementary findings. The estimated elasticity of new ideas with respect to research labor is high (approximately 0.8), consistent with sizeable local scale effects. Counterfactual simulations suggest that without the rise in immigration after the 1965 Immigration and Nationality Act, U.S. patents and income would be about 5 percent lower today.

Conclusion

The study concludes that immigration has a meaningful and positive causal effect on local innovation and wage growth in the United States. By constructing a more credible instrument for immigration and drawing on a uniquely deep historical dataset, the authors overcome limitations that affected earlier research relying on realized ancestry shares. The findings underscore that immigration’s dynamic effects on idea production and productivity can outweigh its short-run labor-supply impacts. The results carry implications for understanding long-run growth patterns, the distribution of gains across skill groups, and the role of migration policy in shaping regional and national economic performance. The strength of the results, the methodological rigor, and the consistency between reduced-form and structural evidence position this article as a substantial contribution to the empirical growth and immigration literatures.