When it rains, it pours: On the spillover effects of corruption on firms’ tax evasion

Abstract

This study examines the direct and spillover effects of corruption on firms’ tax evasion in sub-Saharan Africa (SSA), a region plagued by both phenomena. Firm-level data on 17 SSA economies spanning 2008–2014 were sourced from the World Bank’s Enterprise Survey. I instrument for endogeneity of corruption using five variables, namely the number of days it takes to (i) clear exports, (ii) clear imports, (iii) obtain an operating license, (iv) time spent by senior management on regulatory compliance, and (v) the extent to which business licenses and permits are an obstacle the firm’s operation. I implemented a battery of endogeneity-correcting estimators: traditional instrumental-variable two-stage least squares (2SLS), Lewbel’s 2SLS, Oster’s bounding analysis, propensity score matching, and Kinky Least Squares. I find that corruption of tax officials (i.e., bribe expected/requested in exchange for a favourable tax inspection) significantly increases tax evasion. More critically, this study challenges prevailing assumptions from European and Central Asian contexts by demonstrating that corruption outside tax authorities—specifically bribes expected/requested in exchange for operating licenses and government contracts—also exerts substantial adverse spillover effect on tax compliance. Firms that perceived/experienced bribe requests in exchange for obtaining operating licenses evaded taxes at rates 24.2 percentage points higher than their counterparts who did not perceive/experience bribe requests in exchange for obtaining operating licenses. In contrast, a percent increase in perceived bribe rates for government contracts corresponded to a 2.1 percentage-point increase in tax evasion. Mediation analysis identifies working capital depletion as a key transmission channel. These findings underscore the need for multisectoral anti-corruption strategies to effectively combat tax evasion and enhance domestic revenue mobilisation in SSA.