Japan made cartel bids wobble by scoring more than price

Japan's MLIT promoted a comprehensive evaluation method for public works, where price was combined with performance, function, technical, quality, and life-cycle factors. That changed procurement from a single price race into a scored contest.

A 2025 NBER working paper studied subjective scoring in auctions run by MLIT. Its key finding was counterintuitive: scoring auctions should raise prices in competitive settings, but in this setting adoption significantly reduced winning bids.

The proposed reason was cartel discipline. If conspirators cannot reliably tell whether a member lost because it cheated or because evaluators scored proposals differently, punishment becomes harder. The auction did not need secret randomness; ordinary scoring noise did some of that work.

Use scored proposals to make the cartel's chosen winner uncertain

Bid-rigging depends on members knowing who was supposed to win and whether anyone defected.

Subjective scores blur that signal without abandoning public procurement rules.

The design attacked the cartel's enforcement mechanism, not just its price.

When law cannot prove a cartel, mechanism design can still make collusion harder by breaking the feedback cartel members need to police each other.

NBER's 2025 summary framed scoring as one of several auction-design tools available when suspicious bidding is not enough for a legal case. The MLIT case became evidence that procurement rules can be changed to make collusion less stable even before prosecutors can prove an agreement.