Graduate Research and Discovery Symposium (GRADS)

Document Type


Publication Date

Spring 2015


Workers that lose their jobs because their employer closed a plant or division, moved or abolished their position, or simply had insufficient work for them are reported to experience huge losses in earnings, post-displacement. Empirical studies using ordinary regression techniques have estimated these losses to average between 10% - 40%. However, around this mean loss is a distribution with considerable variation, variation that for the most part has been unremarked on. Using Displaced Workers Survey data from 1994-2010, I find that the mean loss in weekly earnings of displaced workers stands at 18%, while the median loss is only 6.5%. At the 25th percentile of the earnings-change distribution, the loss amounts to 35% and at the 75th percentile there’s a gain of around 9.5%. In light of such variation in losses, I argue that classical regression models with their estimates of conditional-mean loss fail to provide a complete picture of the post-displacement earnings experience of workers. I apply quantile regression technique to this problem and use regression curves corresponding to various points on the earnings-change distribution to provide more insight into these losses.