Modelling dynamic interactions and adaptive behavior in Balancing Mechanism offers

In competitive, bilateral markets, repeated interaction amongst power plants with the same economic characteristics would be expected to result in convergence of their offer prices. If they do not, it raises interesting issues related to the sustainability of heterogeneous competitive strategies. In this paper, we analyze the offer prices submitted to the UK Balancing Mechanism in 2008 by four coal-fired power stations, separately owned (by British Energy, Electricité de France, E. On and Drax), approximately of the same age, size and efficiency, all LCPD compliant, and located in the same congestion zone. We find evidence of offer price dispersion and heterogeneous offer strategies despite the repetitive and transparent nature of the market. First, differences emerge across companies concerning both the speed of mean reversion, the volatility persistence, and the frequency and intensity of spikes. Second, while Drax, BE and E. On tie their offers to coal and gas prices, BE offers track the APX day-ahead price index quite closely. Finally, vector autoregressive (VAR) and vector error-correction (VEC) model estimates show that offer prices are co-integrated with coal prices, and that EdF acts as the leader in peak-load periods.