The Costs of Mismatch

I calibrate a model of vertical skill mismatch to US data and demonstrate that both high-skilled workers and low-skilled workers prefer mismatch to segmentation in a decentralised environment. Using this framework, I provide estimates of the output costs of skill-mismatch with reference to a natural benchmark - a labour market in which search is fully segmented by skill. Surprisingly, I find that, despite misallocation (due to highskilled workers undertaking low-complexity tasks), mismatch raises net output by around 7.5%. I show that mismatch is particularly beneficial for low-skilled workers and argue that a key mechanism underlying these net benefits is the (endogenous) response of job creation to the expanded pool of searchers under mismatch. These results call into question the view that mismatch should be seen as a form of misallocation that has deleterious effects on productivity.