Loading Events
  • This event has passed.

Ben LOCKWOOD (Wharton) – “Diffuse Bunching with Optimization Frictions: Theory and Estimation”

May 3 @ 12:15 pm - 1:30 pm

Microeconometrics Seminar: Every Tuesday
Time: 12:15 pm – 13:30 pm
Date: 3rd of May 2022
Room : 3001

Ben LOCKWOOD (Wharton) – “Diffuse Bunching with Optimization Frictions: Theory and Estimation”
Santosh Anagol, Allan Davids, and Tarun Ramadorai (co-authors)

Abstract: This paper presents a parsimonious model in which a continuum of taxpayers face discrete income adjustment opportunities. The model produces continuous observable income densities which match key features of empirical income histograms around kinks and notches in the tax schedule, including diffuse bunching and positive mass above tax notches. We analytically characterize the income distribution distortions produced by kinks and notches as a function of the elasticity of taxable income and a single other parameter—the average distance between adjacent income adjustment opportunities—which can be jointly estimated from empirical distributions using maximum likelihood methods. This method does not require the analyst to specify bounds on a bunching window or (in the case of a notch) a dominated income region in order to implement the estimation. We apply this method using administrative tax data from South Africa, estimating moderate elasticities between 0.2 and 0.4 at higher incomes, and large elasticities in excess of one at low incomes. We find strong evidence that firms with paid tax practitioners adjust their incomes more precisely in response to tax incentives, although we do not find a consistent pattern of income elasticities across this dimension of heterogeneity. We compare our results to estimates using a conventional bunching estimator based on a model of continuous income adjustment, which suggests that accounting for adjustment frictions results in higher estimated income elasticities.

Organizers:
Benoît SCHMUTZ (Pôle d’économie du CREST)
Anthony STRITTMATTER (Pôle d’économie du CREST)

Sponsors:
CREST