What Remote Career Gains Actually Depend On
The career study's headline outcomes do not, by themselves, explain why remote eligibility is associated with higher wage growth and upward seniority. This chapter examines the path beneath the destination label: whether workers gain through occupation upgrading, higher-ranked roles, a wider set of feasible matches, or simply by moving to a different city or employer type. The decomposition is useful because it turns a broad association into a set of candidate channels, but it remains conditional evidence rather than a randomized mediation experiment.
The wage association is mainly a matching-and-job-content story
The preprint uses a Gelbach decomposition to examine how the estimated wage-growth association changes when post-transition variables are added to the model. In plain terms, it starts with the adjusted relationship between remote eligibility and wage growth, then asks how much of that relationship is statistically associated with particular features of the destination transition. The candidate features include occupation upgrading, upward seniority movement, cross-metropolitan change, destination-employer size and prestige, and the availability of remote opportunities in the worker's industry and metropolitan labor market.
The largest reported share is associated with occupation upgrading, which accounts for approximately 21.9% of the total wage-growth effect. Upward seniority moves account for another 17.2%. Access to remote job opportunities within the worker's industry and metropolitan labor market accounts for 12.8%. Taken together, these channels fit a matching interpretation: remote eligibility is associated with transitions into jobs that are higher in occupational or organizational rank and with a broader set of roles that a worker can realistically consider.
These percentages are shares of a model-based decomposition, not a menu of independent effects that can be added to predict one person's raise. The channels overlap conceptually: an occupation upgrade may also be a seniority move, and a broader opportunity set may be what makes either move possible. The decomposition allocates the statistical reduction in the remote-eligibility coefficient conditional on the other variables. It is most useful for identifying which observed parts of the transition are consistent with the overall pattern, not for assigning a precise personal contribution to each channel.
The word “associated” is essential. A decomposition allocates a change in a regression coefficient across observed variables under a model. It does not randomly assign a worker to an occupation upgrade, a seniority move, or a broader opportunity set. A worker who enters a higher-ranked occupation may have acquired skills that are not fully measured; the same skills may also make them more likely to obtain a remote-eligible offer. The decomposition can show which observed transition features line up with the wage association, but it cannot prove that one feature is the sole causal mediator.
Even with that limitation, the pattern is informative. If the wage result were mainly a reward for staying home, we would not necessarily expect it to be concentrated in the content and level of the destination job.