Confronting the challenge of hybrid modeling: Using discrete choice models to inform the behavioural parameters of a hybrid model

The development of hybrid models represents an important step in energy economy modeling, as hybrid models embody the most useful features of both top-down and bottomup models. Hybrid models explicitly represent technologies in a similar way to bottom-up models, thus enabling policy makers to understand the effects of technology-specific policies on energy consumption and the economy. However, to simulate consumer choice between alternative technologies and processes, hybrid models diverge from bottom-up models to use the behaviourally realistic approach of top-down models, resulting in a simulation model rather than an optimization model, which is ultimately of more use to policy makers. The primary challenge in developing a representative hybrid model lies in specifying the algorithm to simulate consumer choice between alternative technologies. The algorithm needs to capture the fundamentals of consumer behaviour at a disaggregate level. This essentially involves estimating the empirically derived aggregate parameters used in topdown modeling at a technology specific level. Discrete choice models (DCMs) are well suited to provide this information. DCMs use empirical data from historical or hypothetical technology choices by consumers to determine the relative importance of technology attributes to adoption of that technology. However, where DCMs have been used for predicting consumer behaviour relating to energy use decisions, the models have been run independent of a whole economy model, meaning that important system feedbacks have been ignored. By using empirically derived DCMs to inform the behavioural parameters of a hybrid model, the behavioural realism of the DCM can be coupled with the system feedbacks of the energy economy hybrid model to produce an integrated model that is both behaviourally realistic and technologically explicit.