Computational modelling of price formation in the electricity pool of England and Wales

This paper develops a detailed computational model of price formation in the England and Wales electricity pool, as it operated for 11 years from 1990 to 2001. It is clear that during this period, the repeated nature of the daily auction, between a small number of generators, with a substantial amount of information in common, gave rise to a continuous evolution of learning and gaming in practice with no evidence of convergence to a stationary Nash solution. In terms of representing reality, a computational approach inspired by evolutionary economics, can succeed in reflecting well the type of behaviour observed, to an extent that cannot be matched by alternative analytical models. Cycles of pricing appear in the model, apparently as they seem to do in practice, yet average behaviour has been validated against the theoretical supply function results for the more stylised circumstances where analytical results are possible. The paper therefore makes a methodological contribution in the development of a model of competitive electricity markets inspired by computational learning and gaming. It also makes an applied contribution by providing a more realistic basis for identifying whether high market prices can be ascribed to problems of market structure or market conduct.