Technology bias in technology-neutral renewable energy auctions: Scenario analysis conducted through LCOE simulation

Abstract

This report identifies technology bias between renewable power plants in technology neutral auctions, caused by different auction design elements. We evaluate our technologies (PV, onshore wind, offshore wind, and biomass) using a quantitative model, with which we determine LCOEs, bid prices, and social values of the technologies are calculated. Based on these calculations, we calculate bias between two technologies, by taking the difference of differences between the bid prices of the auction, and the unit social values of the power plants, which are the total social revenue minus the total social cost in per unit terms. With respect to bias, two efficiency concepts are used in the report: allocative and general efficiency. An auction is allocatively efficient if the technology with the highest associated unit social value, would emerge victorious from a multi-technology auction. The concept of general efficiency in terms of bias, is based on the average bias between all technologies, which we determine as the unweighted average of pairwise biases. Comparing two designs, we deem the design that leads to lower average bias as more efficient. In the model, inputs from different sources are used for the technologies and other important parameters. As a result, the cases are not a representation of a given country or energy system, instead they serve as basis for scenario comparison in archetypical technology/market situations. The main approach of the analysis is to compare various cases with differing input parameters and auction design elements. Seven different auction design elements are evaluated, which are remuneration scheme, support period, granted realisation period, timing of the auction within a year, balancing cost payment responsibility, grid integration cost compensation payment and environmental harm compensation payment (i.e. internalisation of environmental effects into the auction). In the model, three main scenarios are defined. In the “laboratory scenario”, all technologies face equal market prices and no externalities are considered. The “no externality scenario” is very similar, but in this case, price cannibalisation is introduced, so renewable deployment influences the achieved market prices for all technologies. Finally, in the “baseline scenario”, grid integration costs and environmental harm are introduced – here, they are considered as influencing the societal value of the technology, but are still regarded as externalities for producers. The separation of these three cases is important, because the different setups lead to quite different results even when base cases are considered. In the “laboratory scenario”, PV proved to be socially the most beneficial, while in the “no externality” and “baseline” cases, it was onshore wind. The laboratory scenario is allocatively efficient for sliding and fixed premium remuneration schemes when the same auction rules are implemented for all technologies, while the “no externality” and “baseline”, this is only the case for fixed premiums. Average bias between all technologies is lower in fixed premium schemes in the “laboratory” and “no externality” setups, while in the “baseline scenario”, sliding premium schemes lead to lower bias, because the hierarchy between the technologies is completely different in terms of bias compared to the other two cases. Because of these differences, all three main scenarios are being used by the evaluation of the different auction design elements. With respect to the different design elements, it is impossible to formulate a rule of thumb type policy conclusions. The main reason behind this is that that when a different setup is assumed, it can change the hierarchy between renewable technologies in terms of bias. This can result in the fact that the same change in the design (for example increasing support period from 15 years to 20), may increase the average bias in one setup and decrease it in another. We term this as “starting point effect”, as the effect of a design element change is heavily influenced by the extent of biases present in the originally assumed setup. Therefore, caseby-case analysis is required to determine the effect of a design element change, on bias. Our model results enable us to formulate conclusions relating to the extent different design elements influence technology bias. The outcomes show that while change of support period, or the introduction of grid integration costs and environmental harm compensation may heavily influence average bias between technologies, the effects are more moderate when changes in granted realisation period or in balancing payment responsibility are applied, and almost negligible if changes in timing of the auction within a year occur. Remuneration scheme design is a very important determinant as well, but there is no clear hierarchy identifiable comparing two-sided sliding premiums and fixed premiums. Both schemes are though clearly leading to lower risk of technology bias than one-sided sliding premiums, as in several setups where a technology is mature enough to survive without support, one-sided premiums may result in very high biases. An additional very important conclusion of the report is that allocative and general efficiency do not necessarily occur simultaneously. By comparing two designs, it is often the case that a given setup results in allocative efficiency, but in terms of general efficiency it fares worse than another allocatively inefficient auction setup. The reason behind this phenomenon is that allocative efficiency in practice mainly consider the differences between the most mature technologies (and check which of them is being awarded), in our case (based on the assumed input values), these technologies are typically PV and onshore wind, while general efficiency takes into account all the biases, that is including toward biomass and offshore wind, which are the technologies that have smaller chances of winning in the auction due to their currently higher technology cost. To conclude, this report provides a useful theoretical concept and practical tool to analyse the effect of design element changes on technology bias, which can be used by policymakers or other interested actors. The quantification of technology bias enables policy makers to make informed decision about implementing multi-technology auctions, about which technologies to include, about if differentiation of rules may be an appropriate decision, and it so, at what level potential bonus or malus should be set.