CAN AGENT-BASED MODELS FORECAST SPOT PRICES IN ELECTRICITY MARKETS? EVIDENCE FROM NEW ZEALAND

摘要

Modelling price formation in electricity markets is a notoriouslydifficult process, due to physical constraints on electricity generationand flow. This difficulty has inspired the recent development ofbottom-up agent-based models of electricity markets. While thesehave proven quite successful in small models, few authors haveattempted any validation of their model against real-world data in amore realistic model. In this paper, we take one of the most promisingalgorithms, the modified Roth and Erev algorithm, and apply it to a19-node simplification of the New Zealand electricity market. Oncekey variables such as water storage are accounted for, we show thatour model can mimic short-run (weekly) electricity prices at these 19key nodes quite closely.