From Monitoring to Evaluation: Using Red Flag Indicators to Assess Systemic Change in Public Procurement

Anti-corruption reforms in public procurement are typically grounded in the assumption that regulatory change can reshape incentives, constrain discretion, and improve institutional performance. Yet, establishing whether and through which mechanisms these reforms translate into observable changes in administrative behaviour remains a central challenge for policy evaluation. This difficulty is particularly pronounced in the case of universal regulatory interventions and in domains such as corruption, where outcomes are structurally latent and multidimensional. This paper addresses these challenges by evaluating the 2023 Italian Public Procurement Code (Legislative Decree No. 36/2023), a comprehensive reform that redefined procedures, thresholds, and degrees of administrative discretion across the entire procurement system. Rather than attempting to measure corruption directly, the study adopts a risk-based perspective grounded in a Theory of Change framework. The analysis starts from the reform’s own causal logic, focusing on its expected effects on key integrity channels - such as transparency, competition, bounded discretion, accountability, and efficiency - and on the procedural conditions under which procurement decisions are made. The empirical strategy relies on administrative data from the Italian National Database of Public Contracts, which records all public procurement procedures carried out by contracting authorities. The study focuses on a set of red flag indicators commonly used in preventive anti-corruption monitoring, selected to capture different stages of the procurement cycle and to reflect institutional vulnerabilities relevant to the reform’s declared objectives. In this context, red flags are not treated merely as indicators of wrongdoing, but as systematic signals of how administrative practices and risk exposure adjust over time following regulatory change. Methodologically, the paper employs a multilevel Interrupted Time Series design implemented through Generalized Linear Mixed Models. This approach is particularly suited to evaluating universal reforms in the absence of external control groups, as it allows each contracting authority to be compared to itself over time while accounting for structural heterogeneity and differential responses to the reform. By combining temporal dynamics with cross-authority variation, the framework captures gradual adjustments, behavioural inertia, and heterogeneous reform effects that would be obscured by aggregate analyses or purely cross sectional designs. Beyond the Italian case, the paper contributes to broader debates in anti-corruption research and public policy evaluation by demonstrating how risk-based indicators and administrative microdata can be repurposed to assess systemic institutional change. In doing so, it advances a methodological approach that bridges preventive monitoring and policy evaluation.