An Investigation of Internet Auction Markets: Evidence from eBay

Abstract

The purpose of this paper was to examine a market with different characteristics than a typical market using high frequency continuous transactions data. The market selected for this purpose was an Internet auction market for collectibles, specifically Ty Glory BearsTM sold on the eBay, Inc. web site. This market was chosen for its relatively high activity and homogeneity. The results indicated evidence of signaling related to seller's reputation and product information, seasonality based on day-of-the-week, and limited market depth. INTRODUCTION Financial economists have a long history of examining the behavior of markets. Traditionally, the focus has been on mainstream financial markets such as equities, bonds, and derivatives. However, there has also been an interes t in examining other non-traditional markets. Gandar, et al (1988) and Gray and Gray (1997) are among the many researchers who explore the concept of efficiency in the NFL gambling markets. Other sports betting markets such as basketball (Brown and Sauer, 1993), baseball (Woodland and Woodland, 1994), and horseracing (Asch, et al, 1982; Thaler and Ziemba, 1988) have also been studied. Outside of the sports arena, Scott and Gulley (1995) and Fapacliristou and Karamanis (1998) both look at lottery markets. These investigations have led to mixed evidence on the efficiency of gambling markets.' In addition to gambling markets, there has been a body of research into the area of collectibles markets. This has covered areas such as art (Bryan, 1985; Mok, et al, 1993; Matsumoto, et al, 1994; Pompe, 1996), stamps (Taylor, 1983; Cardell, et al, 1995), antique furniture (Graeser, 1993), coins (Dickie, et al, 1994), wine (Krasker, 1979). sports cards (O'Brien and Gramling, 1995) and comic books (Ang, et al, 1983). Most of these studies examine the return structure of the particular collectible market under investigation and consider its viability as an investment alternative based on its return as well as its correlation to more traditional financial investments (stocks, bonds, gold, etc.). However, there is limited evidence on the nature of these market based on high-frequency transaction data over a condensed period of time. There are three reasons for the lack of attention to this aspect of the collectibles market. First, the collectibles market has historically been extremely segmented by regional preferences. This makes it difficult to develop a meaningful database of transactions that aren't greatly influenced by regional biases. Second, often collectibles are distinctive making it difficult to separate market influences from product differences such as quality of condition from transaction to transaction. For example, a primary objective in Taylor's (1983) examination of the stamp market is developing a methodology to separate differences in quality from differences in price. Finally, there has not been a good source of high-frequency transaction data for a continuous time collectibles market. This paper examines the auction of collectibles by using a continuous worldwide marketplace with a homogeneous product that eliminates the three problems described above. Because the auction is conducted via the Internet, participants can see all recently completed auctions to generate an expectation of a market price. One advantage of such an auction market is that with a worldwide market all participants are widely dispersed. That is, they likely share no community (e.g., geographic, union) with other participants. This reduces the likelihood of collusive behavior and, more importantly, eliminates the geographic premiums that are often apparent in collectible markets. The wealth of data available through Internet auction markets has generated a growing body of interest (McDonald and Slawson, 2002; Houser and Wooders, 2001; Bajari and Hortacsu, 2002; Katkar and Lucking-Reiley, 2001; Melnik and Aim, 2002, and Roth and Ockenfels, 2000.) These papers tend to focus on using the Internet auction data to address theoretical related to auctions. For example. Roth and Ockenfels (2000) examine the