Electronic Theses and Dissertations





Document Type


Degree Name

Doctor of Philosophy


Business Administration



Committee Chair

Julia A. Heath

Committee Member

William T. Smith II

Committee Member

Andrew J. Hussey

Committee Member

David C. Sharp


This paper proposes that the Iterative Cumulative Sums of Squares (ICSS) Algorithm can be applied to help better detect periods of collusive behavior in markets by analyzing changes in the variance of product prices over time. Price rigidity is a common characteristic of oligopolies and has been theoretically proven in many studies. The kinked-demand-curve explains that firms would prefer to stay at the agreed upon monopoly price rather than cutting prices to earn more market share during a single period. An infinitely repeated Bertrand game model developed by Athey, Bagwell, and Sanchirico (2004) claims that if the firms are sufficiently patient, the optimal symmetric collusive scheme can be reached when the equilibrium-path price wars are absent and the price is rigidity. Harrington and Chen (2006)'s dynamic programming framework established an optimal cartel price path which has a transition phase and a stationary phase. The stationary phase shows that price in collusive regime is much less volatile than price in competitive regime. In order to detect the existence of cartel, this paper employs the ICSSalgorithm developed by Inclan and Tiao (1994) to detect multiple changes of variance in a given time series. The flat glass antitrust litigation in early 1990s was detected by this technique and had the results that periods of December 1982 to June 1984 and November 1987 to February 1990, with the lower variance relative to the periods before and after were defined as suspected collusion periods. By applying the model for damage analysis, the price of flat glass was confirmed to be overcharged by the producers during the conspiracy periods detected by the ICSS algorithm rather than the class periods certified by the court. The steel industry in the 1920s and 1930s had market power in agreeing on price-fixing. This industry was analyzed using the ICSSalgorithm and found relatively smaller steel price variance for the periods of August 1924 to November 1931 and June 1938 to December 1939. The damage analysis has shown that customers for steel products were overcharged by steel producers during these periods rather than the periods in the literature. Based on empirical results, the ICSSalgorithm provides a fast and simple method of detecting the existence of cartels and collusive behavior. This is the first study to apply the ICSSalgorithm in forensic economics to detection of this behavior and appears to be successful in detecting periods of anticompetitive behavior. In the future, it might provide an alternative method to more easily discover and prosecute anticompetitive behavior.


Data is provided by the student.

Library Comment

Dissertation or thesis originally submitted to the local University of Memphis Electronic Theses & dissertation (ETD) Repository.