Empirical analysis of price data in the delineation of the relevant geographical market in competition analysis

This paper reviews a number of modern as well as classical econometric techniques suitable for empirically determining whether commodities in physically separated markets belong to the same geographical market. Even though the tools presented generalize to the delineation of the relevant product market our main focus is on the geographical delineation of markets. The analyses rely entirely on the use of price data of different types in an attempt to operationalize the so-called SSNIP methodology for price comparisons. In particular, the stationarity versus non-stationarity of price data appears important because otherwise spurious results can potentially occur. Both bivariate and multivariate price comparisons will be discussed. We also consider situations with data observations covering a relatively long period and where it is likely that structural changes have occurred in the sample period whereby the degree of market integration is likely to have changed. New techniques to deal with such recursive features will be suggested. For the methods presented a discussion of the practical problems and concerns facing the model builder are addressed.