Spain Targets Contract Cartels

Picture the procurement archive every competition lawyer wishes they had time to read: every public tender, every company that bid, every price put on the table, every subcontract handed out after the award. Spain’s competition authority is trying to turn that archive into an early-warning system.

The Comisión Nacional de los Mercados y la Competencia, the CNMC, is developing an AI tool called Atenea to search public procurement records for signs of bid-rigging and collusion (La Vanguardia). It is being built on a database of about six million contracts. The point is not to let a machine find guilt, but to tell human investigators where to look first.

What the Machine Looks For

Cartels in public contracts tend to leave a trail. Firms take turns winning. Supposed rivals submit bids too high to be serious. A company loses the tender, then quietly turns up later as a subcontractor. The OECD's bid-rigging guidance has described these warning signs for years. The difficulty has been scale, not theory.

A town hall may see only one tender. Atenea can look across buyers, regions, sectors and years. According to reports, it works in stages. First it connects companies that may be appearing under different names, including cases where one business bids through several subsidiaries. Then it searches for recurring patterns: the same firms bidding together, the same losers reappearing, the same market behaving in a way that ordinary competition would not suggest.

The system is said to build on an earlier, narrower CNMC screening tool. Its scope now extends into open markets and into a newer problem: algorithmic collusion. That happens when companies’ pricing software learns to follow competitors’ moves without anyone needing to make the old-fashioned cartel phone call.

Why Spain Has Reason to Hurry

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

Those are exactly the kinds of patterns a screening system is meant to catch earlier: the same companies circling the same contracts, winners and losers swapping places, and a losing bidder coming back through the side door as a subcontractor.

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 screening of procurement data (BBH Blog). Italy’s anti-corruption authority, ANAC, already publishes tender data that could feed similar analysis (ANAC).

The harder issue is what follows when an algorithm helps decide who comes under scrutiny. The Netherlands had a bruising lesson when a court struck down SyRI, a government risk-profiling system, for breaching privacy rights (Rechtspraak). The Dutch response was to create a national register where 1,495 algorithm descriptions from 515 organisations are now publicly visible (Algoritmeregister). The principle is plain enough: when a state algorithm shapes who is investigated, people should at least know the tool exists.

The research case for screening is real, but limited. A CEPR study of Swedish pharmaceutical auctions found patterns consistent with collusion and higher prices. That does not mean an AI detector will work across every market. False positives are the obvious danger. Specialist contracts often attract only a handful of bidders. Shared costs can make prices move together. Joint bidding can help smaller firms compete for work they could not take on alone.

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

Its credibility will rest less on the cleverness of the model than on the controls around it. Can a company challenge an alert? Can the logic be audited? Can investigators override the recommendation? Spain is testing whether competition enforcement can move earlier, before a whistleblower emerges or a court case takes a decade to finish. If the safeguards are serious, taxpayers and honest bidders stand to gain. If the system operates as a black box, it risks becoming the kind of unchallengeable algorithmic authority European courts have already rejected.