This note provides a narrow replication of Fisman and Miguel's (Journal of Political Economy, 2007a; 115(6): 1020-1048) original findings about estimating negative binomial count models to study corruption practices among United Nations diplomats. We present estimates based on zero-inflated count models, given the possible presence of excessive zero counts in the dependent variable of the main specifications. Our results confirm Fisman and Miguel's original findings. However, they also suggest the importance of considering distinct generating processes for zero outcomes. We cannot reject hypotheses favoring the use of zero-inflated negative binomial models over its simpler versions in this context.
