What Can Bunching Teach Us About Tagging?

• In Event: Tax System Design and Taxpayer Compliance

Abstract

Many papers in the literature on bunching at kinks and notches provide informal analysis of how agent choices of variables other than income (or the running variable in question) appear to be distorted in the vicinity of the bunching region. This paper provides a formal interpretation of these observations as well as clear guidance on what parameters can be identified from these patterns. In the case of bunching at kinks, I show that by comparing average values of the consumption of some untaxed good below and above the kink point, we can point identify sufficient statistics needed to implement the optimal differential commodity tax formulas of Ferey, Lockwood, & Taubinsky (2024) which provide guidance on when and how tagging can justify such differentiation. This results holds in a model with multidimensional heterogeneity, and some of the necessary sufficient statistics are even point identified under assumptions where the ETI is not identified. When the ETI is known, stronger results hold. In the case of notches, related results are derived that have wide applicability.

Author

• Dylan Moore, dtmoore@hawaii.edu