A Transaction-Efficiency Analysis of an Internet Retailing Supply Chain in the Music CD Industry*

Decision Sciences, Winter 2003 by Rabinovich, Elliot, Bailey, Joseph P, Carter, Craig R

HYPOTHESES AND EMPIRICAL METHODOLOGY

The measurement of transaction costs-given by the disparity between the seller's (i.e., producer, wholesaler, Internet retailer) marginal cost, MC^sub s^, and the price, P^sub b^, paid by the buyer-rests on two key assumptions. First, prices quoted to consumers are based on sellers' profit-maximization incentives. Second, prices in buyer-seller exchanges significantly depend on transaction costs. These assumptions are addressed in the empirical analysis by studying music compact disc (CD) Internet-retailing operations owned by publicly traded firms.

The profit maximization incentive assumption is supported by limiting the study to publicly owned Internet sites, with the highest levels of historical profitability in the e-commerce industry and with profit maximization incentives promoted by stockholders inside (Monsen & Downs, 1965) and outside the firm (Hindley, 1970).

The second assumption is supported by three product and market conditions for CD retailing over the Internet. First, CDs are commodity items. They are standardized, durable, and low-priced goods. Furthermore, the attributes that generate consumer utility for an individual CD title remain unchanged irrespective of where, when, how, or by whom the title is sold (Mas-Colell, Whinston, & Green, 1995). Also, pricing actions of a single CD seller are unlikely to alter market prices (Brynjolfsson & Smith, 2000). Therefore, CD sellers are likely to be price takers as opposed to price setters, to compete under Bertrand-price competition conditions, and not to pursue monopolistic pricing policies for each CD title.

Second, because of its nascent conditions, the Internet market for CDs is highly competitive. These conditions lead to a close association between transaction costs and the disparity between marginal costs and prices charged to consumers. Third, the Internet market for CDs is subject to low search costs, causing prices to converge to marginal costs. Seller attempts to reap market profits and drive retail prices above marginal costs can easily be detected by all market participants.

Internet-based CD markets also exhibit properties that facilitate the empirical measurement of transaction costs. First, Internet CD sellers' marginal cost structures are reasonably uniform, since they share common product suppliers and a common, interoperable, and open-standard information interface with consumers. This makes available to all sellers almost identical cost-effective levels of communication with consumers. Second, the study of Internet CD markets can effectively account for any additional marginal-cost heterogeneity determinants pertaining to Internet-retailing site, products, and channel-structure attributes, as defined by the pure-play and bricks-and-clicks Internet retailer typology.

In line with the previous considerations, transaction costs result from the gap between the actual price paid by the buyer, P^sub b^, and a maximum price that the seller could possibly charge for the product, P^sub s^. Thus, transaction costs are not measured in terms of how far P^sub b^ is above MC^sub s^ across different dyadic-exchange settings. Instead, transaction costs are measured by how far below P^sub b^ is from P^sub s^. From Figure 1 and the properties exhibited by Internet-based CD markets, both of these measurement approaches result in equivalent transaction cost measures. In this case, an increase in the disparity between P^sub b^ and P^sub s^ is reflective of lower transaction costs and higher efficiencies in the seller-buyer transaction (Figure 1). This transaction measurement approach is consistent with economics literature on intermediation (Demsetz, 1968), where market transaction costs are measured as the difference between the transacted products' upper-bound prices and mediated prices.