Using Twins to Resolve the Twin Problem of Having a Bad Job and a Low Wage

We use twin data linked to register-based information on earnings to examine the long-standing puzzle of nonexistent compensating wage differentials. The use of twin data with a panel dimension allows us to alleviate the impact of otherwise unobserved productivity differences that have been the prominent reason for estimation bias in prior studies. Using a twin design, we find some evidence for positive compensation of adverse working conditions in the labor market.