Curbing grand corruption in the contracting out of public services lessons from a pilot study of the school meals program in Colombia

Do targeted transparency interventions reduce corrupt behavior when corrupt actors are few and politically influential; their behavior imposes small costs on numerous individuals; and corrupt behavior is difficult to observe? Results from a study of informal audits and text messages to parents, meant to curb corruption in the School Meals Program of Colombia, suggests that they can. Theory is pessimistic that transparency interventions can change the behavior of actors who exert significant influence over supervisory authorities. Moreover, inherent methodological obstacles impede the identification of treatment effects. Results substantiate the presence of these obstacles, especially considerable spillovers from treated to control groups. Despite spillovers, we find that parental and operator behavior are significantly different between treatment and control groups. Additional evidence explains why operator behavior changed: out of concern that systematic evidence of corrupt behavior would trigger enforcement actions by high-level enforcement agencies outside of the political jurisdictions where they are most influential.