The paper reviews current issues over anti-corruption and transparency in government procurement, and examines theoretical works focused on collusion and corruption such as McAfee and McMillan (1992), G th and Peleg (1996), Marshall et al. (1994), Rose-Ackerman (1975, 1978), Shleifer and Vishy (1993), Laffont and Tirole (1993), Bliss di Tella (1997), Laffont and N'Guessan (1999). Furthermore, based on Sajio et al. (1996), some experiments are designed, and communication among bidders (the "dango" effect) and market institutions are tested.

When "dango" is not allowed, winning prices are very close to competitive prices and winning prices tend to decrease as periods proceed. When "dango" is allowed, subjects easily formed very robust cartels and found winning bids close to the maximum price prepared by the experimenters. No significant effects, however, are observed regarding the market institutions with or without a corruptible agent who can communicate with bidders.