This paper examines the effect of improved transparency in the bidder qualification process, using the experience based on a case study of municipal public work auctions. It reveals that improved transparency reduces procurement cost by a maximum of three percent. This finding is robust to the concerns of endogeneity, sample selectivity, and distributional assumptions. The bidding-function estimates, combined with features of Japanese procurement system, imply that the improved transparency limits abuse of auctioneer's discretion, and thus weakens the stability of collusion among bidders.