Spain Targets Tender Cartels

Imagine a system that remembers every public tender ever issued: who bid, at what price, who lost, and who later turned up as a subcontractor. Now imagine it checking those records for the same warning signs a competition investigator would look for, if time were no constraint. That is what Spain’s competition authority is trying to build.

The Comisión Nacional de los Mercados y la Competencia (CNMC) is developing an AI tool called Atenea to scan public procurement data for possible bid-rigging and collusion (La Vanguardia). It draws on a database of roughly six million contracts. Its job is to flag suspicious tenders for human investigators, not to decide guilt by itself.

What the machine looks for

Cartels in public procurement tend to leave traces. Firms take turns winning. Losing bidders submit prices that are too neat, too high, or too consistent. A company that keeps losing later receives part of the work as a subcontractor. The OECD’s bid-rigging guidance has listed these signals for years. The hard part is not knowing what to look for. It is seeing the pattern across thousands of tenders.

That matters for Malta too. Public procurement here is not an abstract Brussels file; it is how roads, hospitals, IT systems, waste contracts and local council works get delivered. In a small country, one tender can reshape a market. A pattern across tenders can tell you much more than any single award notice.

A local official sees one procurement process. Atenea can compare behaviour across buyers, regions, sectors and years. It reportedly works in layers. First, it links companies that may appear under different names, as when the same corporate group bids through several subsidiaries. Then it groups recurring signals: do the same two firms keep appearing together? Finally, it checks whether a market behaves differently from what real competition would suggest.

The system is reported to build on an earlier, narrower CNMC screening tool, extending it into open markets and into a newer risk: algorithmic collusion. That is when companies’ pricing software learns to follow competitors’ moves without a human agreement being made, much as shops in the same street might quietly stop undercutting each other.

Why Spain has reason to hurry

Spain is not solving a theoretical problem. Last month, its Supreme Court confirmed a €13.5 million fine against Indra for taking part in a cartel that rigged public IT-services tenders, with conduct running from 2005 to 2015 (Cinco Días). The practices included cover bids, competitors staying out of tenders, and advance knowledge of tender details through insider contacts. Indra was attributed more than €324 million in contracts awarded through the affected arrangements (elDiario.es).

Those are exactly the patterns a machine could, in principle, detect earlier than a decade-long investigation: the same firms appearing together, winners and losers switching roles, and a losing bidder later reappearing as a subcontractor.

For Maltese readers, the point is familiar. In a micro-state, procurement problems are rarely just about price. They are about who has access, who is close enough to the decision-makers, and whether honest firms believe it is worth bidding at all. A tool that catches suspicious patterns earlier could protect taxpayers. Used badly, it could also become another opaque layer between citizens and accountability.

The accountability test

Spain is part of a wider European move towards machine-assisted enforcement. Germany’s proposed 12th amendment to its competition law would allow systematic procurement screening (BBH Blog). Italy’s anti-corruption authority ANAC already publishes tender data that creates raw material for similar analysis (ANAC).

The difficult question is what happens when an algorithm helps decide who gets investigated. The Netherlands learned this the hard way when a court struck down SyRI, a government risk-profiling tool, for violating privacy rights (Rechtspraak). The Dutch response was to build a national register where 1,495 algorithm descriptions from 515 organisations are now publicly visible (Algoritmeregister). The principle is simple: if a state algorithm affects who comes under scrutiny, the public should know it exists.

Academic evidence supports screening, but with limits. A CEPR study of Swedish pharmaceutical auctions found patterns consistent with collusion and higher prices. But such findings are market-specific. They do not prove that universal AI detectors can work everywhere.

False positives remain the central risk. Specialist contracts naturally attract few bidders. Common input costs can make prices move together. Joint bidding can allow smaller firms to access work they could not handle alone. In Malta, where many sectors have a limited number of serious operators, that distinction matters.

Atenea is reported to be still under internal development, with deployment expected in the coming months. No public technical specification has surfaced. The CNMC has not published the system’s architecture, error rates or governance rules.

The tool’s legitimacy will depend less on how clever the algorithm is than on whether its alerts can be challenged, its logic audited, and its recommendations overruled. Spain is testing whether enforcement can move earlier, spotting collusion before a whistleblower appears or a court case drags on for a decade. With proper safeguards, taxpayers and honest firms gain. Without them, the risk is a black-box authority of the sort European courts have already rejected elsewhere.