Abstract: TABLE OF CONTENTS I. INTRODUCTION II. THE MECHANICS OF DATA MINING III. THE ECONOMICS OF DATA MINING FOR PRICE DISCRIMINATION IV. ANTITRUST DOCTRINE AND POLICY PERTAINING TO PRICE DISCRIMINATION A. The Robinson-Patman Act B. The Sherman Act 1. Policy a. Allocative Efficiency b. Consumer Welfare Maximization c. Business Fairness d. Special Interest Favoritism 2. Monopoly Doctrine 3. Oligopoly Doctrine V. CONCLUSION I. INTRODUCTION In 2000, customers of Amazon.com discovered that the online retailer was varying the prices charged for DVDs depending on the identity of the purchaser. (1) Although Amazon discontinued what it described as a test (2) after public outcry, Amazon's brief foray into first-degree price discrimination stands as a noteworthy example of the possibilities for price discrimination using aggregated data. In its price test, Amazon sought to use information it already had about its customers to predict higher prices that the customers would still be likely to pay. (3) Nearly a decade later, brokers of consumer information now sell terabytes of data for the purposes of market segmentation and other consumer analytics. (4) Such aggregations of consumer data are ripe for applications of data mining technologies. (5) These technologies enable producers to recover more of the economic surplus created by a transaction with a particular consumer by facilitating the development of first-degree price discrimination schemes. (6) In contrast to a perfectly competitive market where producers capture only their marginal costs and consumers capture the entire economic surplus of a transaction, price discrimination allows producers to recover some or all of the economic surplus. Thus, effective first-degree price discrimination reduces the welfare of consumers compared to a competitive market. In addition to effecting a redistribution of wealth, price discrimination incentivizes consumers to engage in aftermarket arbitrage. It also incentivizes producers to develop mechanisms to prevent such arbitrage and to invest in more effective price discrimination schemes. These changes in behavior waste resources that would otherwise be efficiently allocated in a competitive market. Furthermore, perfect price discrimination is impossible in real markets. Consequently, imperfect price discrimination imposes deadweight losses that would not occur in a competitive market. (7) The policy rationales advanced to justify antitrust doctrine recognize that each of these results is an economic loss. However, although the policies behind antitrust law tend to disfavor price discrimination, the doctrines do not typically proscribe such discriminatory conduct. Part II of this Note examines the mechanics of current data mining technologies and distinguishes between uses that promote price discrimination and uses that serve other ends. Part III considers the economic effects of data mining technologies used to facilitate price discrimination. Part IV examines the policies and doctrines underlying the Robinson-Patman Act and the Sherman Act and argues that the policies that justify the Sherman Act are consistent with enforcement against data-mining-based price discrimination, although it is not available under present doctrine. Even if this conduct is not proscribed, the presence of data-mining-based price discrimination is indicative of the presence of other harms that are proscribed by the doctrine. Part V concludes that current antitrust policy and doctrine are mismatched and that, without legislative or judicial augmentation of the doctrine, data mining technology will likely pose a greater risk of future economic loss. II. THE MECHANICS OF DATA MINING Data mining refers to the process of extracting patterns from data. (8) For instance, a credit card issuer may mine transaction data to detect suspicious transactions to reduce credit card fraud; (9) astrophysicists may mine telescope data to select regions in space for more careful investigation; (10) or advertisers may mine consumer data to tailor advertising to particular demographics. …
Publication Year: 2009
Publication Date: 2009-03-22
Language: en
Type: article
Access and Citation
Cited By Count: 3
AI Researcher Chatbot
Get quick answers to your questions about the article from our AI researcher chatbot