Can a shopping agent enforce the law of one price?

Abstract

Using data gathered from an Internet shopping agent, Pricegrabber.com, a panel of price data was collected weekly over an ll-month period for a group of 26 commodities. Both price data and shipping costs were reported allowing the incorporation of transportation costs into the analysis. Various measures of price dispersion were constructed, (for example, the standard deviation of prices), for each of the commodities and the time series behavior of these measures of price dispersion were analyzed.AR(1) models were estimated and, using the estimated AR(1) coefficients, the halflives of the measures of deviation for the 26 commodities were constructed. The half-life is given by: (ln (.5)/ In pi )), where pi is the estimated AR(1) coefficient for each price dispersion time series and estimates the time it takes for half of the deviation to be eliminated. The half-hfe of these deviations provides insight to the speed with which deviations from the Law of One Price (LOP), the concept that identical goods should sell for the same price, are reversed.There are several reasons to believe, a priori, that LOP should be more hkely to hold in these markets. First, the cost of information and conducting transactions are dramatically lower for consumers using shopping agents to facihtate online purchases. With a simple mouse click consumers can compare the price of identical goods across several markets and respond quickly to any price differentials. Secondly, electronic commerce is not subject to many of the restrictions, regulations, and taxes that affect conventional retail trade. Third, the nature of the Internet makes the strategy of pricing-to-market inherently difficult to employ. Finally, some types of market frictions are less problematic with electronic commerce, such as price stickiness, due to menu costs. Menu costs are greatly reduced as price changes can be carried out by a few keystrokes.As a benchmark for comparison to the prehminary results, consider the half-byes estimated for international deviations from LOP (that is, from Purchasing Power Parity) by Frankel and Rose [JIE, February 1996] who found a half-hfe of 4.6 years, a result that has been corroborated by several authors. An example of this analysis conducted for the deviations from the LOP for retail commodities between U.S. cities [O'Connell and Wei, JIE, January 2002] found the half-lives of deviations were in the neighborhood of two quarters. The results of the current study suggest much shorter half-lives of deviations for the goods in the sample. Specifically, the average half-life was found to be approximately 3.99 weeks. There is heterogeneity of half-lives across the goods in the sample with the shortest half-life of 0.49 weeks and the longest of 14.93.These results are at least suggestive that information costs and sticky prices are a major contributor to the failure of LOP. Forthcoming work includes further exploration of the differences in half-lives across the goods in the sample, looking to product type, availability of the commodity, non...