In many retail gasoline markets, prices follow a saw-toothed cycle first posited by Edgeworth (1925) and formalised by Maskin & Tirole (1988). A growing literature explores driving factors behind such cycles, most particularly in Canada and the US. This paper explores price cycles in a retail gasoline market in Australia with a unique regulatory environment that provides a census of data. We make use of a threshold regression model, and pay particular attention to local market effects and market structure. Both are novel in the study of retail petroleum prices.
*corresponding author 3Gasoline Price Cycle Drivers: An Australian Case Study
IntroductionRetail gasoline prices in many jurisdictions follow a saw-toothed pattern, shown in Figure Two and called an Edgeworth Cycle. Figure
Figure One about hereNote that prices all seem to rise together, and often to roughly the same level. Clearly, Shell head-office in Perth plays a major role in determining when and by how much each Shellbranded outlet increases its price.Wang (2009) Section Two of this paper briefly reviews the empirical price cycle literature, while SectionThree provides an explanation of threshold regression models. Section Four provides an 4 overview of the approach we take to obtain a structural measure of local market competition.Section Five provides a brief overview of the Perth data used. Section Six introduces the model and its results. Section Seven concludes.
Edgeworth Cycles and Threshold Regression ModelsPrice Cycles were first posited as an equilibrium of a dynamic game by Edgeworth (1925) and formalised by Maskin & Tirole (1988), who named the cycles after Edgeworth . Their distinctive pattern is shown in Figure Two.
Figure Two about hereMaskin & Tirole (1988) show that Edgeworth Cycles are one equilibrium of a repeated, alternate move game when symmetric duopolists produce an homogenous good and who use Markovperfect strategies to choose prices from a finite grid, provided that the discount rate is sufficiently high.. The cycles arise because, for prices above the minimum, a small reduction in price is sufficient to capture the whole market from a rival until it moves again. At the minimum, it is in the interests of both parties for prices to move up again, but not for either party to be the firstmover. They thus play a war of attrition. However, once one firm moves, since the optimal response by its rival will be a slight undercut, the initiator of the price rise has an incentive to increase price by as much as possible in order to maximise its benefits over the price cycle.The model has been extended by Eckert (2003), who allows firms to be of different sizes, by Lau do so by allowing the coefficients to vary over the four phases of the cycle they identify in their work. Noel (2007aNoel ( ,b, 2008Noel ( , 2009) uses a Markov-Switching regression, which is more computationally intensive. Here, we are able to utilise Hansen's (1996Hansen's ( , 1999Hansen's ( , 2000 threshold regression model as we have a census of data, not a...