Work Package 4: Technology diffusion model

Description

In WP4, we aim at developing a model that allows the analysis of historical diffusion dynamics as well as predictions of diffusion of residential solar PV systems in Austria. To this end, we will employ a two-staged modelling approach that combines regression analysis with the Bass Model (Bass, 1969) estimated by using non-linear least squares (NLS) and applied to publicly available data. We will build upon previous studies on diffusion modelling as well as the more recent approach by Lee et al. (2014), and Liu and Madlener (2019), and establish first a temporal diffusion model that depicts the relation between solar PV characteristics and their respective diffusion patterns, accounting for innovation (internal) and imitation (external) factors, which can then be applied to solar PV systems with known product-level parameters in order to predict their future diffusion. The application of the model will be both on the national and the regional scale, thus enabling to identify heterogeneity in the diffusion dynamics amongst the regions and federal provinces in Austria.

In a second step, the epidemic diffusion model is extended in direction of additionally accounting for the spatial dimension. This allows for the analysis of spatial heterogeneity (e.g. geographic clustering, spatial spill-overs, neighbourhood effects, local advocacy effects, urban–rural divide/agglomeration effects) as well, which can then at least to some extent also be included in the macroeconomic model DYNK. Since the application of ordinary least squares must be expected to yield biased estimates (due to spatial correlation), spatial econometric modelling has to be applied.