Figure 2.

(A) The distribution of groups in the cohesion-exclusivity feature space detected in a product-type and bid-degree preserving null model compared with groups detected in observed data from Georgian procurement markets from 2011–2016. We label groups of firms as suspicious if its coherence and exclusivity are in the 80th percentile of the null model outcomes of the year in which they are detected. We highlight 2016’s suspicious zone in white. (B) Distribution of average relative prices of contracts bid on by suspicious groups and ordinary groups, 2011–2016. Suspicious groups consistently win more expensive contracts.