Productivity gains from migration: Evidence from inventors

Do migrant inventors experience an increase in productivity after moving to a new country?

Understanding the productivity patterns of high-skilled migrants, particularly inventors, is crucial for shaping policies aimed at attracting talent and enhancing innovation. Given the significant role that migrant inventors play in global patenting and innovation, insights into their productivity changes post-migration can inform national and international policies regarding immigration and human capital allocation. As countries compete for skilled labor, it is essential to recognize whether such migration leads to a 'brain gain' effect that could benefit both host and home countries.

Utilizing a unique disambiguated dataset of nearly 1 million inventors listed on international patent applications (1990–2011), we conducted an econometric analysis to assess the productivity effects of migration among 33,865 inventor migrants. A machine-learning algorithm was employed to accurately identify inventor identities across patents. A panel regression model with fixed effects was used to evaluate productivity changes post-migration, controlling for various individual and contextual factors.

We find that migrant inventors experience a productivity increase of approximately 23% following their move, measured through an increase of about 0.07 additional patents filed per year in the preferred specification after accounting for various controls. This finding is robust across multiple model specifications and remains significant even after implementing alternative definitions of immigrant status. Furthermore, our results indicate that this productivity boost is not due to inherent selection effects or simply moving to more productive environments, suggesting that migration itself plays a significant role in enhancing inventive productivity.

• These data include 871,129 inventors who filed 2,597,315 PCT applications from 1990–2011.
• These data identify 33,865 migrant inventors within the sample.

• The regression model assumes a causal effect of migration on productivity, controlling for unobservable individual fixed effects.
• Migration is defined as the change in inventor nationality and residence as recorded in patent applications.
• Patenting behavior is assumed to be random with respect to migrant status, allowing for unbiased estimates.