Using Divide and Conquer to Improve Tax Collection: Theory and Laboratory Evidence

Working Paper: NBER ID: w28042

Authors: Sylvain Chassang; Lucia Del Carpio; Samuel Kapon

Abstract: We consider a government collecting taxes from a large number of tax-payers using limited enforcement capacity. Under random enforcement, limited capacity results in multiple equilibria: if most agents comply, limited enforcement is sufficient to dissuade individual misbehavior; if most agents do not comply, enforcement capacity is over-stretched and fails to dissuade misbehavior. In settings without behavioral frictions, prioritized enforcement strategies can implement high collection as the unique rationalizable outcome. We investigate both theoretically and experimentally the extent to which this insight extends to environments with incomplete information and bounded rationality.

Keywords: tax collection; government capacity; divide and conquer

JEL Codes: C72; C73; C92; D73; D82; D86; H26


Causal Claims Network Graph

Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.


Causal Claims

CauseEffect
random enforcement (K40)multiple equilibria can exist (C62)
taxpayer behavior (H26)multiple equilibria can exist under random enforcement (C62)
enforcement capacity (P14)taxpayer compliance (H26)
taxpayer compliance (H26)high or low compliance rates (I18)
prioritized enforcement strategies (K42)unique high collection equilibrium (C62)
strong enough incentives to comply early (H26)high collection (A30)
prioritized enforcement strategies (K42)tax collection rates (H26)
provision of timely information (D83)compliance rates (H26)
strategic ordering of enforcement actions (K40)compliance rates (H26)

Back to index