The National Markets and Competition Commission (CNMC) has launched a pioneering tender for the use of Artificial Intelligence applications as the algorithm used by the agency to detect possible fraudulent practices in public tenders.
Competition has put out to tender a contract for “services for the development and maintenance of Artificial Intelligence and Business Intelligence applications”. The contract, tendered through an open procedure and with an initial period of 24 months, is valued at 2,686,363.64 euros.
Among the CNMC information systems that are going to be developed and maintained with this service, the BRAVA system, an acronym for English, is mentioned first in the specifications. Bid Rigging Algorithm for Vigilance in Antitrust (Tender Manipulation Algorithm for Antitrust Surveillance).
This is a development of the CNMC that, with AI tools and statistical techniques, helps the agency to detect anti-competitive practices in public tenders ex officio.
BRAVA was created by the Economic Intelligence Unit (UIE) of the CNMC with the support of the IT subdirectorate, after a prior proof of concept carried out by an external company. The key to this development is its algorithm based on machine learning (machine learning), which takes the public procurement database as a source of information and “has allowed the development of a pioneering tool worldwide among competition authorities”, explained the CNMC last April, after this tool was awarded by the Digital Information Society Magazine.
Then, Competition underlined that “in recent years, the volume of information and data has grown exponentially, which has posed a challenge for competition authorities, who need new analysis tools to manage all this information efficiently. And that is where the UIE plays a fundamental role.”
Two figures illustrate this exponential growth: in 2023 alone, 196,763 contracting files were tendered in Spain, 7.66% more, for an amount of 107,557.67 million euros, with an increase of 8.56%, according to the latest data from OIReScon, the office that supervises the operation of public procurement. The number of files and the economic volume of the tenders has grown by 53% since 2019 alone.
In fact, OIReScon recommends in its latest annual report the “widespread implementation of the use of AI tools as an instrument to help detect anti-competitive and collusive practices in public procurement.” And he highlights BRAVA precisely as an example in this line.
In an appearance in Congress last September, the president of the CNMC, Cani Fernández, explained that BRAVA “is already capable at this point of classifying in an automated way, with a rigor never below 90%, and normally higher, if an offer presented to the Administration, whether a tender that we observe is potentially collusive or not. “We are advancing more and more so that detection is as automatic and rapid as possible.”
He explained that when this algorithm analyzes a tender “it sees the offers that have been made and analyzing that gives us, with a degree that is increasingly closer to 99% rigor, a first view of whether the offer of that tender has or “It has no collusive elements.” To do this, the tool is trained with information such as the resolutions of the CNMC or its predecessor, the National Competition Commission (CNC), which have already been confirmed by the Supreme Court.
When a contracting authority sends the agency an accelerated procedure to analyze “if it raises problems from the start, BRAVA is very useful to us, the truth is, because it gives us a very solid comparison image,” Fernández explained. “We believe that with this we can help prevent tenders from being awarded in an anti-competitive manner,” because doing so later “is already very difficult.”
The entity has more than 3.5 million public contracts stored in its own database. Thanks to an agreement with the College of Registrars, you have real-time access to the registration data of all companies in Spain. “We have a lot of operators, a large part of the operators that participate in the tenders, who also tend to be the usual suspects,” Fernández explained.
“Now we can know that this company wins, this other loses, loses, wins; lose, lose, win; lose, lose, win,” he explained to the deputies, drawing a drawing with both hands; “There is a pattern here.”
“This allows us to begin investigating, but not only in win-lose patterns, but also in territories, also in types of price, in types of clients, with which, we are training our tools to detect different types of collusion, whether by market sharing, whether by price, or by subcontracting.”
In that appearance in the Lower House, the president of the CNMC defined the UIE as “one of the jewels in the crown, of which we are especially proud.” Created in 2018, until a few months ago this unit had the current Director of Competition of the CNMC, Susana Campuzano, as its head.
The UIE, Fernández explained, “uses various computer and digital tools that range from statistical and econometric techniques to the use of algorithms and artificial intelligence, as well as, in conjunction, citizen collaboration channels and informants who can benefit from anonymity to provide the “indications and carry out that first investigation that will allow us to advance in the detection and discovery of anti-competitive practices.”
Save days of work
In that appearance in Congress, Fernández explained that among the AI applications that they already use is also a project to gain efficiency in the organization’s internal information process or the Guíeme database, which structures the entity’s internal information. He explained that “simply by applying artificial intelligence techniques to the search for precedents, two or three days of work can be saved.”
Another application is the monitoring of official publications of the autonomous governments to detect possible obstructions to the competition of regulations or decisions in that area, “because it is not humanly possible to follow all the bulletins of all the autonomous communities plus what appears in the BOE systematically.”
The president of the CNMC reiterated that the detection of possible anti-competitive conduct in public contracts “continues to be one of our priority areas of action. In Spain, depending on the year, we can reach almost 20% of GDP that is generated directly or indirectly by public procurement. It is a very important triggering factor for the economy and we want all the actions carried out in this market to be competitive in nature.”
In 2022, there was a record number of complaints of corruption and fraud in public procurement. A few months ago the CNMC published a list of indications of possible irregular practices in which it asked officials for help: “We need your collaboration,” it indicated.
First specific contract
One of the reasons that justify this new contract related to AI is, as explained in the document, “regulatory changes that may impact data-based analysis, supervision and regulation procedures.”
“The demand for new solutions based on artificial intelligence and data analysis at the CNMC does not stop growing, so it is necessary to establish a flexible contractual mechanism that allows the necessary resources to be available to address its development with agility, maintaining the same framework.” methodological for quality management and, especially, for the management of the security of the data and information handled by these applications,” it is indicated.
Competition points out in the file documentation that “the development of Artificial Intelligence and Business Intelligence applications is relatively new at the CNMC and has never before been the subject of a specific contract.” Those used today in these two areas are limited to some work carried out by the organization’s own personnel or carried out by the consulting firm Altia in another larger contract awarded in 2022.
The CNMC has chosen to contract these services in a unified manner and carry out better monitoring and control of the projects. It will require that the supplying company have a project manager, five “Full Stack Developers specialists” [responsables de construir, testear, mantener y manejar esas herramientas en todos los aspectos] and three other “Full Stack Developers”.
Exclusive dedication will be required of all of them: “People with a contract committed to another entity, public or private, for the same period of execution of this contract cannot be included in the offer.”
#Competition #launches #pioneering #contract #detect #rigged #tenders #artificial #intelligence
