We use data for Ontario workers with permanent impairments resulting from work‐related injuries to investigate the complex relationships among post‐injury work outcomes: wages, accommodations, returning to the same or different employer, and duration of work absence. We argue the different aspects of post‐injury work experience may be jointly determined, making post‐injury job characteristics endogenous in a duration model. To explore the endogeneity issues we instrument post‐injury job variables from first‐stage equations and compare results from this “informed” model to a “naive” model that treats the variables as exogenous. We find that returning to one's pre‐injury employer is associated with more favorable post‐injury work outcomes, including higher wages, greater likelihood of job accommodations, and shorter durations of work absence relative to workers who change employers. We also find substantial differences between the naive and informed models, with accommodations having the predicted negative effect on duration only after we control for endogeneity.
