Inside the Tender Trap: How Custom Scoring Rules Predetermine Auction Winners

The study shows that corruption in China’s open scoring procurement auctions is widespread, with officials manipulating tender rules to preselect winners while using zombie bidders to mimic competition. Using large-scale data, expert audits, and structural modeling, it finds that about two-thirds of auctions show signs of manipulation and that meaningful reform requires targeting rule-design, not just investigation or enforcement.

Corruption in public procurement continues to undermine governance across developing economies, and new research from the World Bank’s Development Economics Vice Presidency, its Development Impact Group, and the Research Support Team reveals how deeply entrenched these practices remain inside China’s open scoring auction system. Drawing on more than 300,000 procurement auctions from 2006 to 2021, the study exposes a pattern of manipulation hidden not in the evaluation stage but at the very start, inside the writing of the scoring rules themselves. Procurement officials, often working with preselected firms, quietly insert selective requirements that only one bidder can satisfy. Other firms, sometimes willingly, sometimes unknowingly, are brought in to meet the legal minimum of three bidders, creating what the study calls “zombie bids” that simulate competition without offering real rivalry.

The Red Flags Hidden in Plain Sight

Two revealing patterns emerge immediately from the data. First, three-bidder auctions dominate, accounting for more than two-thirds of all open scoring tenders. Because three is the legal minimum to validate an auction, this concentration signals routine engineering of participation. Second, the winning margins are unusually large: winners often receive near-perfect scores, while losing firms score so low that they appear not to have attempted genuine competition. One case illustrates this perfectly. In an organic fertilizer procurement, the three firms submitted similar price bids, yet two firms earned drastically lower quality scores. Later court testimony revealed the truth: the winner had promised a 5% kickback to the procurement official, while the other firms joined solely to provide the appearance of competition. Such cases confirm what the score patterns suggest: rules were crafted to make the winner inevitable.

Measuring Corruption Through Auction Theory

To systematically detect these patterns, the study introduces an innovative statistical tool grounded in auction theory. It defines a measure called the “pseudotype,” reflecting the highest score a rational bidder could earn without losing money. Under normal competition, a winner’s score should be aligned with the expected pseudotype of the strongest rival. But in manipulated auctions, losing bidders’ pseudotypes collapse while the winner’s score inflates far beyond competitive levels. Using nonparametric maximum likelihood estimation, the researcher finds that about 65% of all auctions show evidence consistent with corruption, rising to 71% in three-bidder cases. At the individual-auction level, more than half exhibit the unmistakable signature of manipulated scoring rules, downgraded rivals, and inflated winner scores.

Experts Confirm What the Numbers Reveal

To verify that the algorithm captures real manipulation, the researcher conducted an expert audit with five procurement specialists from the province’s formal evaluation pool. Given 500 tender documents, without knowing bidders or results, experts were asked to identify unnecessary or suspicious scoring rules. Their assessments closely matched the model’s predictions. In auctions flagged by the statistical test, experts independently identified questionable rules 91% of the time, producing an overall consistency rate above 80 percent. The experts repeatedly pointed to familiar tricks: obscure certifications that only one local supplier possesses, overly detailed personnel requirements matching a single firm’s profile, or brand-specific technical parameters disguised within general specifications. The patterns indicate that rule design, not evaluator misconduct, is where corruption is strategically embedded.

Policy Lessons: Temporary Fixes vs Real Reform

The study also examines the impact of China’s sweeping anti-corruption campaign launched in 2014. Using an event-study framework, it finds that corruption investigations do temporarily increase the use of open auctions, encourage more bidder participation, and reduce corruption indicators. But these improvements fade within months, especially when investigations target low-level officials. Durable change occurs only when high-ranking officials are scrutinized, underscoring how corruption networks are structured around senior influence rather than isolated misconduct.

Finally, the researcher uses a structural auction model to test reforms. The most effective solution is an anonymous expert pre-review of tender documents, which removes unnecessary or biased scoring rules before bidding begins. According to the simulations, this reform would reduce prices by about 18%, increase quality by 3%, and improve overall welfare by 11%. Roughly 70% of the gains come from eliminating rule manipulation, and the rest from increased bidder entry. By contrast, commonly proposed tweaks, such as raising price weights, deliver negligible improvements and risk lowering quality.

The study reveals a procurement ecosystem where corruption thrives not by breaking rules, but by writing them. Without reforms that target the tender-design stage, open scoring auctions may continue to mimic fairness while quietly steering contracts toward favored firms.