Does regional variation in wage levels identify the effects of a national minimum wage?

TL;DR

This paper critically examines the use of regional wage variation to estimate the effects of a national minimum wage, highlighting biases and proposing diagnostic tools through simulations and a case study.

Contribution

It reveals biases in common estimators due to measurement error and model misspecification, and evaluates solutions using simulations and empirical data.

Findings

Estimators are biased by measurement error and functional form issues.

Small deviations from assumptions can greatly increase bias.

Diagnostic tools can help identify and mitigate biases.

Abstract

This paper evaluates the validity of estimators that exploit regional wage differences to study the effects of a national minimum wage. It shows that variations of the ``fraction affected'' and ``effective minimum wage'' designs are vulnerable to bias from measurement error and functional form misspecification, even when standard identification assumptions hold, and that small deviations from these assumptions can substantially amplify the biases. Using simulation exercises and a case study of Brazil's minimum wage increase, the paper illustrates the practical relevance of these issues and assesses the performance of potential solutions and diagnostic tools.