8 th Workshop on eBusiness, Phoenix 2009 REAL-TIME SUPPORT FOR AUCTIONEERS TO DETERMINE OPTIMAL CLOCK START FOR MULTI-UNIT SEQUENTIAL DUTCH AUCTIONS Alok Gupta, Wolfgang Ketter*, Eric van Heck*, Meditya Wasesa* Carlson School of Management Minnesota, *Rotterdam School of Management alok@umn.edu , wketter@rsm.nl , evanheck@rsm.nl , mwasesa@rsm.nl Abstract In this paper we explore the issue of starting price in subauctions and provide theoretically motivated guidance to provide online real-time support for auctioneers within the Dutch Flower Auction (DFA), a well established busines-to business (B2B) auction platform. A key contribution of this study is the design of starting price increments mechanism such that the equilibrium points with higher revenue structures become identifiable and feasible. In our analysis, we first empirically show that the relationship of clock starting price and the winning bid in a previous auction is indeed an important factor in determining future prices. Then we develop an analytical model for optimizing the sublot starting point from the perspective of maximizing revenue to generate normative insights. We then use the data to demonstrate how the auctioneers’ choice of using a static starting point may result in loss of revenue. Our analysis indicates that the existing rule to determining the starting price may be adequate for auctions that generate small prices but the rule may generate too small revenue for auctions with higher prices. These findings lead to the need of online real-time support that is able to do dynamic starting price setup in correspondence with the auction characteristics. Keywords: Auction theory, decision support system, Dutch auction, multi-item auctions 1. Introduction In this research we focus on providing technology based real-time decision support mechanism for fast paced Dutch Auctions. Our work is in the tradition of developing decision support systems for knowledge workers, for example, by automating trivial tasks like data entry businesses can decrease the corresponding error rates (Taylor & Raden, 2007). Advances in technology provides many opportunities to support organizational information needs by utilizing systems that provide more accurate and more timely information to aid managerial decision making (Eckerson, 2006). The context for our research is the Dutch flower auctions (DFA) (Kambil & Van Heck, 1998). The DFA play a vital role in maintaining the Netherlands’s leadership in this industry; they serve as efficient centers for exchange of flower products between suppliers and buyers. The suppliers in the DFA are the growers (often family businesses), while the buyers include both large and small wholesalers and retailers. The auctioneers act on behalf of growers, while the bidders act on behalf of the buyers. In DFA auctions, the auctioneers set initial prices, reserve prices, minimum lot sizes, and clock speeds; while the bidders decide which and how many flowers to bid on and at what price. In 2008, FloraHolland – a cooperative of growers that operate six auction houses strategically located throughout the country – reported trades of over 44.8 million cut flowers and 5.1 million potted plants per day, generating over 4 billion Euros in sales per annum 1 . The DFA are held on weekday mornings between the hours of 6 and 10. Up to 40 auctions can occur simultaneously, with each lasting 3 to 5 seconds on average. In total, roughly 125,000 transactions occur daily. 1 http://www.flora.nl/en/AboutFloraHolland/Cooperative/Documents/Kengetallen%20EN.pdf