Data analytics against corruption: Advancing accountability in public spending

Imagine you’re a public official responsible for safeguarding public spending. You oversee numerous government procurement processes, ranging from school supplies to infrastructure projects, and conduct audits across numerous agencies. Detecting collusion or fraud often involves manually reviewing extensive documents—a challenging task even with management information systems in place. Institutional controls—such as audit offices, financial oversight units, and integrity systems—play a critical role in identifying risks. Given that procurement constitutes a large share of public spending, it often serves as the most effective starting point for establishing those safeguards. Yet, procurement systems often face significant challenges due to their large size, complex structure, and considerable discretionary authority, as well as the inherent interactions between the public and private sectors (IEG, World Bank, 2024).

Traditional approaches to identifying and preventing corruption, such as manual reviews or handling complaints, can be costly and may not be comprehensive. A conservative estimate suggests that as much as 8 percent of global procurement contract values—roughly $880 billion annually—may be lost to corruption (Bosio, 2021). And since corrupt networks typically operate on a systemic level, investigations focused solely on individual cases may overlook wider patterns. 

GRAS: Expanding the frontiers through data analytics

In our experience working on the World Bank's Governance Risk Assessment System (GRAS), first piloted in Brazil, we harnessed advanced data analytics to detect patterns of fraud, collusion, and corruption before they became major issues. Instead of relying on complaints or labor-intensive manual audits, we could proactively identify risky patterns and act quickly to safeguard public spending. (Ortega et al., 2023). GRAS combines public procurement data with sources like payroll, social benefits, beneficial ownership, georeferencing, and electoral data to generate actionable risk reports. This approach turns procurement information into a tool for oversight, benefiting countries regardless of their digital development level.

GRAS uses 60 main risk indicators, or “red flags”, organized into four categories:

1.  Procurement irregularities - Red flags include single-source tenders, unrealistic timelines, or unjustified cost increases during the execution phase.
2. Collusion indicators - These capture repeated coordinated bids or recurring “loser” companies, often revealing fake competition and contract manipulation.
3. Supplier characteristics - Unusual ownership structures or offshore registrations can indicate shell companies lacking financial credibility and potentially concealing the true owners.
4. Political connections - Links to political campaign financing or public officials can signal conflicts of interest or contracts awarded for electoral favors.

_{Image Source: World Bank, 2023 (Ortega et al., 2023).} 

From evidence to action

During the piloting of GRAS, we were able to uncover over 850 suppliers with indications of collusion, 450 likely using strawmen, and 500 cases where companies owned by public servants received contracts from their own agencies. GRAS has since been fully integrated with government platforms through a World Bank loan in Paraíba, Brazil, demonstrating its compatibility public administration infrastructure (Ortega et al., 2023). The success of this methodology has generated considerable international interest. We are currently engaged in ongoing discussions and capacity-building initiatives in over a dozen countries across various regions, interested in learning about the technical pathways for implementing GRAS, adapted to local contexts and different maturity stages of e-procurement systems. Recently, we convened data analytics and integrity officials from six African countries—South Africa, Nigeria, Rwanda, Ghana, Ethiopia, and Kenya—to discuss possible ways to adapt GRAS to their specific data environments. This collaborative effort will help pave the way to support the GRAS implementation in emerging economies.

The new normal

GRAS is evolving from an effective model developed in the context of advanced digital government and highly sophisticated open data systems to contexts with less mature data environments. 

Just as search functions have become a standard aspect of content management systems, risk-assessment tools like GRAS should become a standard feature for government e-procurement and auditing systems. GRAS provides an example of how data analytics can transform transaction-recording platforms into tools that strengthen integrity and accountability. Ultimately, detecting and addressing fraudulent practices is one component of broader efforts to restore public trust and increase government accountability.

Editor’s note:

The Global Program on GovTech and Public Sector Innovation supports the adoption of technology and use of data to enhance government effectiveness and efficiency, as well as increase citizen engagement and improve governance. Learn more about the program.

The Anticorruption for Development Program (AC4D) is focused on key areas that include reducing corruption in public procurement, strengthening beneficial ownership transparency and combating illicit financial flows, reinforcing accountability institutions and the rule of law, and innovating through the use of data and technology.