Deviation signs-based bunching methods

Abstract

Bunching caused by discontinuous policy has received increased research attention over the last decade. In practice, buncher adjustment costs frequently result in bunching in an unknown region rather than at a specific point (e.g., a kink or notch). To investigate this phenomenon, researchers often rely on visual determination or regional knowledge, the selection of estimated regions, the choice of k in k-fold cross-validation, and the assumption of a conditional lognormal distribution. This study introduces deviation sign-based bunching methods for kink and notch settings with frictions, eliminating the need for these prerequisites. Using conservative non-bunching regions and deviation sign-based indicators, we can calculate the data-driven bunching region and counterfactual distribution. Subsequently, the elasticity of taxable income, a key parameter in bunching research, can be estimated. Simulations validate the efficacy of our methods. The elasticity of Chinese small and micro enterprises is calculated to be 1.267.