Empirical Analyses of Online Procurement Auctions: Business Value, Bidding Behavior, Learning and Incumbent Effect
While there is an ever increasing adoption of e-sourcing, where a buyer auctions off procurement contracts to a small group of pre-qualified suppliers, there is a lack of understanding of the impact of dynamic bidding process on procurement outcomes and bidding behavior. To extend the knowledge of this important issue, in this thesis, we explore empirically the value of online procurement auction on cost reduction, quality management, and winner selection from the buyer's perspective. We also explore how incumbent status affects the procurement outcomes. From suppliers' perspective, we characterize their bidding behavior and examine the effect of incumbent status on bidding. First, we collect detailed auction and contract awarding data for manufacturing goods during 2002-2004 from a large buyer in the high-tech industry. The rich data set enables us to apply statistical model based cluster technique to uncover heterogeneous bidding behavior of industry participants. The distribution of the bidding patterns varies between incumbent and non-incumbent suppliers. We also find that the buyer bias towards the incumbent suppliers by awarding them procurement contracts more often and with a price premium. Next, focusing on recurring auctions, we find that suppliers bid adaptively. The adaptive bidding is affected by the rank of suppliers' final bids. Finally, with field data of procurement auction for legal services, we demonstrate that service prices are on average reduced after dynamic bidding events. Most interestingly, the cost savings are achieved without the sacrifice of quality. Incumbent winners' quality is higher, on average, than the quality of buyer's supplier base before the auctions, while non-incumbent winner's quality is lower. These findings imply that the main value of online procurement auctions for business services comes from incumbents in the form of reduced price and enhanced quality. We find that after adjusting for incumbents' higher quality, incumbent bias disappears. Our results also imply that the buyer might possess important information about the incumbents, through past experiences, that cannot be easily included in the buyer's scoring function due to uncodifiability.
Advisor:Wu, D.J.; Narasimhan, Sridhar; Zhang, Han; Keskinocak, Pinar; Toktay, Beril
School:Georgia Institute of Technology
School Location:USA - Georgia
Source Type:Master's Thesis
Date of Publication:08/24/2007