Measurement error in minimum wage evaluations using survey data

Abstract

"We assess the role of measurement error in minimum wage evaluations when the treatment variable - the bite - is inferred from a survey wage distribution. We conduct Monte Carlo experiments on both simulated and empirical distributions of measurement error derived from a record linkage of survey wages and administrative data. On the individual level treatment effects are downward biased by more than 30 percent. Aggregation of the treatment information at the household, firm or region level does not fully alleviate the bias. In fact, the magnitude and direction of the bias depend on the size of the aggregation units and the allocation of treated individuals to such units. In cases of a strongly segregated allocation, measurement error can cause upward biased treatment effects. Besides aggregation, we discuss two possible remedies: the use of a continuous treatment variable and dropping observations close to the minimum wage threshold." (Author's abstract, IAB-Doku) ((en))

Cite article

Bossler, M. & Westermeier, C. (2020): Measurement error in minimum wage evaluations using survey data. (IAB-Discussion Paper 11/2020), Nürnberg, 49 p.