Endogenous technological change in power markets

Decarbonization requires the transformation of power markets towards renewable energies and investment costs are decisive for the deployed technologies. Exogenous cost assumptions cannot fully reflect the underlying dynamics of technological change. We implement divergent learning-by-doing specifications in a multi-region power market model by means of mixed-integer programming to approximate non-linear investment costs. We consider European learning, regional learning, and three different ways to depreciate experience stocks within the European learning metric: perfect recall, continuous forgetting, and lifetime forgetting. Learning generally yields earlier investments. European learning fosters the deployment of solar PV and wind onshore, whereas regional learning leads to more wind offshore deployment in regions with high wind offshore quality. Perfect recall fosters solar PV and wind onshore expansion, whereas lifetime forgetting fosters wind offshore usage. Results for continuous forgetting are in between those of perfect recall and lifetime forgetting. Generally, learning leads to the earlier deployment of learning technologies but regional patterns are different across learning specifications and also deviate significantly from this general pattern of preponing investments.