Optimal tax threshold: the consequences on efficiency of official vs. effective enforcement

Significant efficiency gains are available when there is a gap between official and effective enforcement of a tax threshold. Using a unique dataset on Ugandan firms, I show that audits for business-related taxes are effectively based on the number of employees rather than the official tax threshold, which is in terms of sales. Based on the empirical evidence, I build a model of firms growth with entry and exit. Entrepreneurs evade part of their tax liabilities and, when audited, bargain with tax officials to keep some of the surplus from evasion in exchange of a bribe. The model is calibrated using the Ugandan data and replicates well some features of the data that are not explicitly targeted. Based on a counterfactual analysis, I show that the efficiency loss associated with evasion and corruption is of the order of 45% in Uganda. There is also a non-negligible gain in productivity per worker of 16% from enforcing the official tax threshold based on the level of sales rather than the effective threshold based on the number of employees. This gain in efficiency is essentially due to the reallocation of labor across productive units.