Inferring Fine-Grained Migration Patterns

Migration data illuminate important demographic, environmental, and health phenomena. People move in response to conflicts and natural disasters, looking for economic opportunity, in search of education, and more. However, migration datasets within the United States obscure some important trends due to their granularity. Publicly available Census data are neither spatially nor temporally granular, and proprietary data have higher resolution but are prone to demographic and other biases.

To address these limitations, Assistant Professors Nikhil Garg and Emma Pierson have developed a scalable iterative-proportional-fitting based method that reconciles high-resolution but biased proprietary data with low-resolution but more reliable Census data. They apply this method to produce MIGRATE, a dataset of annual migration matrices from 2010 - 2019 that captures flows between 47.4 billion pairs of Census Block Groups (CBGs) — about four thousand times more granular than publicly available data. These estimates are highly correlated with external ground-truth datasets, and improve accuracy and reduce bias relative to raw proprietary data.

The team used MIGRATE to analyze both national and local migration patterns. Nationally, they observed temporal and demographic variation in homophily, upward mobility, and moving distance: for example, they found that people are increasingly likely to move to top-income-quartile CBGs and identify racial disparities in upward mobility. They also showed that MIGRATE can illuminate important local migration patterns, including out-migration in response to California wildfires and the impact of public housing policy, that are invisible in coarser previous datasets.

In the time since the data release, the team has received data requests from dozens of institutions all over the world, including multiple universities, the United Nations Development Programme, journalists, and many others. These researchers are using the data to study a wide range of topics, including residential segregation, climate-induced displacement, labor market mobility, health equity, and the impact of teleworking.

Access to the data can be requested at https://migrate.tech.cornell.edu.