Why data is Latin America’s best weapon against COVID-19 corruption
- Corruption has surged in Latin America alongside the pandemic.
- Open data and analytics have emerged as key weapons in the fight against this problem.
Latin America was suffering from its own pre-existing conditions when COVID-19 hit, notably low growth, stagnant productivity, high informality and severe inequality. Most damaging of all, however, has been the region’s endemic corruption, which has funnelled scarce resources away from health systems and social security. Corruption has undermined state capacity and fiscal space, both of which are required to respond to the crisis.
Corruption surged in the region as the pandemic was spreading. Growing demand for the medical supplies to treat COVID-19 patients, coupled with the urgency with which these goods are needed, have created irresistible opportunities for unscrupulous officials and organized crime groups to profit from the crisis. Examples of fraud include counterfeiting much-needed medicines and supplying poor quality health equipment, as well as collusion among suppliers, including overpriced body bags. This illustrates how low these corrupt networks have stooped: even death does not escape their lust for illicit gains from public funds allocated for the emergency.
Within the region, corruption can be as deadly as the pandemic itself. Corruption in the medicines supply chain and in the procurement of medical equipment are emerging throughout the region. For this reason, controlling public emergency purchases and public health expenditures has become a priority for many governments and integrity institutions, including audit agencies, civil society, and investigative media. However, unlike in previous crises and catastrophes, information has become more ubiquitous and the greatest ally of transparency through the expansion of open data and fraud analytics.
Progress with e-procurement platforms and open contracting in the region has generated greater and better data that can be mined to detect and deter irregularities. Data related to government contracts and purchases has become more open, allowing greater scrutiny by oversight agencies and the general public, including ‘civic-techs’ and ‘gov-techs’: tech-based and data-driven start-ups seeking to make a social impact and improve public management.
Open data has become a key tool against corruption and fraud during this pandemic. For example, in Paraguay, regulations require the government to report regularly on the use of funds allocated to the emergency; this includes both the purchase of medical equipment and public contracts related to the economic recovery packages. These reports must be released to the public in an open data format and are supported by a user-friendly visualization platform. In Peru, open data is allowing investigative media to look for bid rigging cases in the public procurement of medical supplies.
The massive generation of data and information on public procurement creates an invaluable opportunity for new technologies, based on data analytics, to detect and deter systemic corruption risk in government contracting, rather than pursuing isolated cases of corruption. As such, it provides a potent tool with which integrity institutions and oversight agencies, including internal control systems, can address policy failures and institutional vulnerabilities.
A first and important step is opening-up government contracting data in an open format that enables its re-use. This can be achieved by adopting the Open Contracting Data Standard (OCDS), promoted by the Open Contracting Partnership, a global advocacy network. The OCDS establishes how to publish information and documents at all stages of the public procurement process, thereby allowing for a more comprehensive analysis of datasets beyond individual transactions or critical bottleneck. This standard has been adopted by more than 30 governments around the world, such as Australia, Chile, Colombia, France, Paraguay and the United Kingdom, as well as the European Union. Some cities, like Buenos Aires, have also committed to regulate their public procurement practices under this standard.
A second and complementary option is deploying advanced analytics and artificial intelligence tools to a variety of data, structured and unstructured. Having quality, open and reusable data allows more agile deployment of advanced analytics techniques. But the absence of sufficiently structured and refined data is not an insurmountable barrier in today’s world of data analytics and machine learning. Brazil and Colombia, for example, have developed fraud analytics platforms to leverage a variety of datasets from multiple sources to red-flag corruption risks in government contracting.
In Brazil, the Analisador de Licitações e Editais (ALICE) is a tool of the external audit office to mine public procurement documents. It captures the information from the public contracting system of the Ministry of Economy (Comprasnet), downloads the texts of contracts, analyses them to generate reports on vulnerabilities and red-flags, and provides an overall assessment of the contracting processes. In Colombia, the comptroller-general’s office uses OCEANO, an analytics platform that cross-checks information from the e-procurement database to detect possible irregularities. This tool has revealed suspicious links between certain companies and some provincial governors and city mayors, which led it to identify cost overruns in health emergency-related contracts.
Data analytics allows governments and oversight agencies to move from a reactive and complaint-motivated approach to a proactive and preventive one. The power of data-processing machines and programmes also increases the speed of judicial and administrative investigations. However, its success rests largely on the availability of high-quality data. Unless artificial intelligence tools can rely on accurate data, the risk is generating inaccurate or, worse, biased results, following the ‘garbage-in, garbage-out’ principle. Indeed, most of the investment in advanced analytics and machine learning projects is spent on ‘cleaning’ datasets, allocating just 18% of the time to developing algorithms and testing models.
To respond to these challenges but also to seize those opportunities, several countries in the region are seeking to organize their data and harness its potential in the fight against corruption. For example, Ecuador is evaluating the quality and availability of 30 datasets that, when cross-referenced, enable the detection of corruption risks. In Colombia, the Transparency Secretariat of President Iván Duque seeks to develop a data-lake that gathers and cross-references datasets from different sources to identify corruption risks in real time, not only in public procurement but also in the licensing or the generation of regulations.
William Edwards Deming, an American historical statistician, used to say that “in God we trust; all others bring data”. Overcoming challenges in terms of data quality, as well as investing in computational power to process that data, would allow for more effective anti-corruption policies whose deterrent power would rest on their ability to predict and anticipate. Faced with the opportunism that corruption and organized crime agents cultivate in emergencies such as COVID-19, open data and advanced analytics provide an opportunity to expose their networks and prevent their crimes.
This article was written with Camilo Cetina and first appeared in the World Economic Forum on 28 August 2020, available here.