Retail petrol is regarded as a staple good in Australia and is an important input to many sectors of the economy. Ensuring that petrol markets function efficiently is therefore of great importance to the broader economy. The objective of this thesis is to econometrically model retail petrol markets as they exist in South-East Queensland, in order to identify the nature of competition. The investigation and construction of the model will be guided by data, models and conclusions present elsewhere in the literature and the nature of the competition exhibited by this market.
The focus is on the competitive effects of vertical integration in a system of spatial competition and the underlying reasons for the outcomes witnessed. The results of the econometric modeling in this thesis allow for the theoretical effects of double marginalisation and spatial differentiation in a retail market to be examined. This is done with the intention of creating policy recommendations designed to promote efficiency in the retail petrol industry.
Due to the complexity of the models estimated and the degree of knowledge regarding the market space, this thesis represents the most comprehensive analysis of vertical integration in a spatially differentiated setting yet conducted. Previous efforts have either ignored the role of the market space in price determination, or only represented this effect through the inclusion of income data in modelling. The depth of demographic data available to this thesis makes it possible to formally represent several different aspects of the market space in models of retail price determination and test each one independently.
Estimation is performed with spatially-lagged models along with ordinary least squares (OLS). The spatial modelling techniques used are the spatial autoregressive model (SAR), spatial error model (SEM) and spatial Durbin model (SDM). By estimating price-setting in an OLS framework, a base case without spatial interaction is established. BAR estimation allow the estimation of prices in the form of a reaction function, which shows the effect of the competitive game taking place in the retail petrol markets of South-East Queensland0 SDM estimation extends upon this by more accurately representing the effect of the market space on the price of the retail firm. Finally, the results of the SEM are estimated to determine the similarity of the pricing decision across firms. In total, these models create a picture of the pricing decision of the individual firm both in terms of its responsiveness to other retail petrol outlets and to the market space, thereby providing a clear indication of the nature of competition in the retail petrol markets examined.
In all cases, the effect of vertical integration is tested for and compared to theoretical results in order to draw conclusions regarding the behaviour of the firm. Further, estimation considers both average monthly price of retail petrol and the mark up above the market-wide average monthly price to clarify seasonalities in the pricing decision of retail firms. Consistent with Hastings (2004), this thesis finds that vertical integration is associated with a one cent increase in the price of retail petrol, when other factors such as input costs and the nature of the market space are accounted for. Further, the results of modelling indicate that the nature of the market space has a significant influence on pricing, consistent with the underlying Hotelling competitive game and other findings in the literature such as Salop (1979).