Renewable Energy Adoption in Germany - Drivers, Barriers and Implications

This thesis deals with renewable energy adoption in Germany. We exploit a unique dataset which includes the location, date of installation and size of all photovoltaic systems, wind power plants and biomass plants for generating electricity installed in Germany through 2011. Importantly, a strong federal subsidy scheme has fostered the adoption of the three technologies since 2000. Panel data analyses on different levels of geographical aggregation allow us to identify drivers, barriers and implications of renewable energy adoption in Germany. We start our analysis by motivating the topic. Then, we review literature on technology adoption. Numerous studies confirm that technology diffusion follows an S-shaped, logistic pattern. A description of the institutional context, aggregate trends and regional differences in renewable energy adoption in Germany follows. The purpose of the subsequent section is to illuminate the spatio-temporal diffusion of photovoltaic installations in Germany quantitatively and to test whether imitation drives photovoltaics adoption. We choose an aggregate approach and employ an epidemic diffusion model which includes a spatial dimension. According to our results, imitative adoption behavior is highly localized and an important factor for the adoption of photovoltaic systems. In the following section, we change our focus on spatio-temporal variation of peer effects, i.e., imitation, in photovoltaics adoption. We add detailed locational data on potential adopters. This data allows us to construct an individual measure of peer effects for each potential adopter. Based on a discrete choice model, we confirm again that peer effects are mostly localized. They generally occur within a radius of 500 meters. We also find that the peer effect's impact on the decision to adopt decreases over time. The next section makes use of the well-studied logistic shape of technology diffusion. The common diffusion path allows us to test whether the adoption rate of renewable energy plants differs between German NUTS-3 regions (`Landkreise und Kreisfreie Städte') in which a successful referendum against a single plant was organized and the remaining regions. We exploit the fact that referenda are mainly organized on the municipal district (`Gemeinde') level against a single plant or building area. Our analysis reveals that the adoption rate (i.e., the first difference in the diffusion level) is indeed lower in NUTS-3 regions where a referendum took place. This finding holds true for wind power and large biomass plants which are both industrial. In contrast, we do not find the same for photovoltaic installations which are mainly private, household installations. We interpret this as evidence that potential investors in wind power and large biomass plants not only avoid the municipal district where a referendum against the specific technology was organized but stay away from the whole NUTS-3 region. Finally, we turn to implications from renewable energy adoption. We estimate the effect of the diffusion of photovoltaic systems on the fraction of votes obtained by Germany's Green Party in federal elections. We take first differences and instrument adoption rates by lagged diffusion levels. We predict the diffusion levels with a logistic diffusion curve. The existing rationales for non-linearities in diffusion and the ubiquity of logistic curves ensure that our predicted instrument is orthogonal to variables that directly affect voting patterns. We find that the diffusion of domestic photovoltaic systems caused a quarter of the increment in green votes between 1998 and 2009. We confirm our findings with survey data from the German Socio-Economic Panel.