What Do Mover Designs Tell Us About Place Effects?
-
Series
-
Speaker
-
FieldEmpirical Microeconomics
-
LocationErasmus University Rotterdam, Campus Woudestein, T18-25B
Rotterdam -
Date and time
November 19, 2026
12:00 - 13:00
Abstract
Mover designs separate causal place effects from sorting by tracking individuals who change locations. I show that their estimates inherently depend on which specific places movers compare. Even with identical geographic place effects and population sorting, mover coefficients diverge simply because different moving routes pair locations differently. This distinction matters for the interpretation of mover event studies and age-at-move designs. A coefficient may accurately capture the returns along the specific routes individuals actually take, yet fail to represent population-wide geographic opportunity. Applications to Dutch firms, healthcare, and childhood moves demonstrate that this discrepancy matters unevenly, depending on how heavily movers concentrate on specific composition or causal corridors.