This paper evaluates a class of endogenous job destruction models based on how well they explain the observed experiences of displaced workers. We show that pure reallocation models in which relationship-specific productivity drifts downward over time are difficult to reconcile with the evidence...