A Checkpoint: From a Portable Job to a Changing City
The evidence on full-time WFH and migration now supports a layered conclusion: remote employment changes who can move across space, but the resulting city pattern depends on costs, time horizons, and the people who do not have the same option. The useful judgment is not simply that WFH raises mobility. It is that a change in work access can begin with a worker’s private relocation choice and end as a change in population, housing costs, and local economic risk.
Four kinds of evidence, one causal chain
The strongest way to read the findings is as a chain with different links, not as a pile of interchangeable statistics. The ACS shows a durable difference between WFH workers and commuters in inter-MSA migration. That difference remains after the study adjusts for observable characteristics and appears in related long-distance measures. It establishes a pattern consistent with job portability, but it cannot by itself decide whether remote work caused the pattern. A worker who selects a flexible occupation may already have the resources, preferences, or household circumstances that make moving easier.
The RPS transition comparison narrows the question. It looks at people who commuted before Covid and separates those who continued commuting from those who later switched into WFH. The switchers moved at higher rates. The comparison is more informative because it focuses on a change in work mode rather than only on differences between two kinds of workers. Yet it still leaves open the possibility that people who planned to move were more likely to seek remote permission.
The employer-policy design adds another layer of leverage. If a pre-Covid job was capable of being done at home and an employer later changed a policy affecting the worker and co-workers, that change can predict access to WFH without being entirely chosen by the individual. The reported two-stage estimates support the interpretation that gaining access to WFH increases migration. The conclusion is stronger than a simple correlation, but it remains conditional on the policy change being plausibly unrelated to a worker’s private relocation plans except through remote access. A strong first-stage test shows that the proposed predictors explain the WFH transition; it does not make the exclusion assumption disappear.
The aggregate accounting exercise answers a different question. Once WFH workers are known to be more mobile, what happens when their share of the workforce expands? The answer can be large even if each group’s own behavior changes little. Holding the earlier group shares fixed leaves much of the observed migration increase unexplained by composition; allowing the WFH share to rise statistically accounts for roughly half of the post-2019 increase. Including the higher WFH migration rate accounts for more. This is a transparent decomposition of the observed total, not a fully causal world-without-WFH experiment.
The destination results then add direction. The aggregate calculation says more mobile workers count for more of the total; the city comparisons show that those workers are not distributed randomly across destinations.