Silvia Vannutelli

How does the quality of information affect the allocation of regulatory enforcement? We study this question in the context of the Italian Court of Auditors, a large bureaucracy that oversees municipal finances through a hierarchical structure. Local auditors gather information on municipalities and report to judges, who decide whether to issue enforcement deliberations. We exploit a reform that randomized auditor assignments, increasing auditor independence without changing judicial rules. Combining novel administrative records with a machine-learning measure of predicted municipal default, we find that judicial enforcement increased overall, with the largest gains in high-risk municipalities. This targeting improvement arises through two channels. First, auditors report more financial irregularities after random assignment, especially in high-risk municipalities and where pre-reform local ties were stronger. Second, experienced judges use these improved signals to focus deliberations on high-risk cases, highlighting the complementarity between information quality and decision-makers' expertise.