Screening for Collusion by Detecting Structural Breaks in Market Behaviour - Dynamic Pricing Model for the Icelandic Gasoline Market

Abstract

Introduction: In the past, economic analysis has not been widely used to detect cartels, according to the author’s literature review. By developing more effective methods to screen for collusion, competition authorities will be more successful in identifying collusion, which will lead to lower incentives for firms to collude. The aim of this thesis is to contribute to the literature on how economists can detect collusion by looking at structural breaks in collusive behavioural indicators. Methodology: The main subject of this thesis is to examine the Icelandic market for gasoline, on the retailer level, from March 1993 to December 2006. Following a formal investigation by the Icelandic Competition Authority (ICA), all of the three vertically integrated firms that sold gasoline in Iceland, from March 1993 to December 2001, were fined for colluding continuously during the entire period. The analysis, used in this thesis, has two dynamic pricing collusive models and econometric tests to test endogenously for structural break, or breaks in market behaviour in the Icelandic gasoline market for the aforementioned period. The dynamic pricing models predict that during collusion, the incentives for the firms in the market to raise price, and therefore the margin, is higher when either demand is expected to increase or cost to decrease. Results: One break is detected, in the second half of 1999, but no support is found for a collusive equilibrium. The coefficients for expected demand and cost are inconsistent with ex ante expectations, i.e. that during the collusive period they would be significant. A change in the coefficient of import price explains the break in the two estimated models. The coefficient for import price shows relatively larger estimates for the first period (1993:3-1999:8/10), compared to other similar empirical studies. For the second period (1999:9/11-2006:12), the estimates are more alike estimates from other similar empirical studies. Discussion: A possible explanation regarding why the market behaviour during the first period is different from the second period, and other similar empirical studies, could be collusive market behaviour. Further analysis is needed to determine why the coefficient for import cost explains the structural break, and if it can be linked to a collusive equilibrium.