Diffuse Bunching with Frictions: Theory and Estimation



We incorporate a model of frictions into the bunching-based elasticity estimator to rationalize diffuse bunching around kinks and mass above notches in empirical distributions. Model agents draw a sparse set of opportunities from a Poisson process, approximating a broad class of frictions including search costs, inattention, and lumpy adjustment; the predicted density depends on the standard structural elasticity and a money-metric “lumpiness parameter.” We estimate the model using administrative tax data on South African small businesses, recovering moderate elasticities of taxable income between 0.2 and 0.3 at higher incomes, and larger elasticities at low incomes. Firms appear to treat the bottom kink as a notch, and firms with paid tax practitioners exhibit sharper bunching, driven primarily by lower frictions rather than a higher elasticity.



Santosh Anagol (Wharton School, University of Pennsylvania), Allan Davids (University of Cape Town), Benjamin B. Lockwood (Wharton School, University of Pennsylvania), Tarun Ramadorai (Imperial College and CEPR)


Please note that although this paper will not be published as an SA-TIED paper, it is presented here to offer a showcase of the research conducted using the data available at the National Treasury Secure Data Lab facility, providing a glimpse into the wealth of insights derived from the analysis of administrative tax data.