Evaluation of Risks for Electricity Generation Companies through Reconfiguration of Bidding Zones in Extended Central Western Europe.

• Author: Felling, Tim

1. INTRODUCTION

   When it comes to pricing in regional electricity markets, economists often refer to socalled nodal pricing as the optimal solution (cf. Schweppe et al., 1988; Hogan, 1992; Bjorndal and Jornsten, 2001; Ehrenmann and Smeers, 2005; Bjorndal and Jornsten, 2007). Yet in Central Western Europe (CWE), a zonal pricing framework has been put in place that couples bidding zones (1) whose borders usually align with national borders. Until 2015, a so-called Net Transfer Capacity (NTC) approach was applied, which in 2015 was superseded by Flow-Based Market Coupling (FLB-MC). With FLB-MC, a better representation of physics of the electricity grid is envisaged. However, the approach assumes bidding zones without internal congestions, which appears to be rather impossible at the moment. Redispatch costs and amounts increased significantly over the last years, especially in Germany (BDEW, 2017). This is mainly a consequence of the increasing infeed from renewable energy sources (RES) in combination with the massive delays in grid expansion, but it also reflects changes in regulation, e.g. the nuclear phase-out in Germany.

   This is why there are broad discussions about the future design of electricity markets in Europe. An optimized bidding zone configuration (BZC) might be one key approach in terms of mid- and long-term redispatch prevention and congestion management. A further split of the German bidding zone--in addition to the split of the former German-Austrian bidding zone into two or more bidding zones within Germany--is frequently discussed, since the massive infeed from northern RES still causes severe internal congestions and loop-fows through neighboring bidding zones.

   The current EU guideline on Capacity Allocation and Congestion Management (CACM) even envisages reviews of the BZC in regular intervals of three to five years (cf. Commission Regulation (EU) 2015/1222). (2) Also, internal congestions shall no longer reduce capacities on cross-border lines in order to prevent internal redispatch within countries (bidding zones) (COM (2016) 861 final). (3) Such a change of BZCs gives rise to additional regulatory risk (4) for generation companies (GENCOs). Their expected net present value depends on local prices, which are not only influenced by changes in primary energy and C[O.sub.2] certificate prices but also by the BZC. Ergo, a frequent change of BZCs might have a major effect on investment and disinvestment. Various recent analyses have shown that changes in primary energy prices, C[O.sub.2] certificate prices, RES expansion, and demand changes have had a major impact on wholesale electricity prices (cf. e.g. Everts, Huber and Blume-Werry, 2016; Kallabis, Pape and Weber, 2016; Bublitz, Keles and Fichtner, 2017; Hirth, 2018). So far, there is little empirical evidence regarding the impact of a reconfiguration of bidding zones, since few reconfigurations have occurred so far in continental Europe. The split of the German-Austrian bidding zone is currently expected to lead to an increase of the average (base) price by 2 EUR/MWh for Austria, (5) i.e. roughly 5% of the current price. Model-based analyses by Blume-Werry, Huber and Everts (2017) have suggested an increase by 1 EUR/MWh. Model-based analyses for other reconfigurations of bidding zones (e.g. Trepper, Bucksteeg and Weber (2015); Egerer, Weibezahn and Hermann (2016); Felling and Weber (2018)) provide mixed evidence: if only large bidding zones are considered, the effects on average prices typically remain limited, whereas with small bidding zones or nodal prices, some areas with high price changes may occur. Moreover, these studies agree that prices tend to decrease in northern Germany, in case Germany is split in several zones.

   Hence, the focus of this paper is to quantify the additional regulatory risk induced by a frequent change of bidding zones for GENCOs and investors, taking into account the other risks these stakeholders are facing. The additional risk induced by regulation is also the only one that can be influenced by policy makers. Thus, it needs to be assessed in detail and put into perspective for policy advice. Therefore, we have developed a new methodology that investigates both the BZCs and the risks for GENCOs by computing and assessing the distribution of present values of the contribution margins obtained in future operations. In principle, this methodology is also applicable to other cases of regulatory (and other) risks that may affect power plant investments in competitive markets, including jurisdictions outside the EU.

   The remainder of this article first gives a brief overview of the relevant literature, with a focus on the two streams of "bidding zone configurations" and "investment decisions and risks." Section 3 develops the different steps of the methodology, and section 4 presents the application of the methodology on a grid- and market-model of extended Central Western Europe CWE+ (Austria, Belgium, France, Germany, Luxembourg, Netherlands and Switzerland). Section 5 analyzes the resulting BZCs and the risks for investments in or continued operation of power plants by investigating the uncertainty in present values of the future operation margins. The article ends with a summary of the main findings and outlook on future research.

2. LITERATURE REVIEW

   To our knowledge, there are no papers that directly address the question of quantifying regulatory risk for power plant investments in the context of applied regulatory decision-making, such as the definition of bidding zones. Yet a considerable amount of related literature exists that may be broadly divided into the two major streams, "bidding zone configurations" and "investment decisions and risks."

   The first stream deals with the ongoing discussion about BZCs in Central Europe. As mentioned earlier, the combination of increasing RES infeed, a delay in grid enforcement, and the current European market design--where bidding zones mostly coincide with national borders, neglecting the physical constraints of the transmission grid--has caused a significant increase in so-called redispatch (BDEW, 2017). The reconfiguration of bidding zones is one approach for midand long-term congestion management to reduce redispatch amounts and costs (cf. Commission Regulation (EU) 2015/1222 ). According to mainstream economic theory, however, the first and best answer to these (and other) congestion management problems is nodal pricing, since nodal prices reflect the marginal generation costs as well as the transmission constraints in the market-clearing algorithm (cf. Schweppe et al., 1988; Hogan, 1992; Ehrenmann and Smeers, 2005; Green, 2007). Nodal pricing is mostly implemented in North American competitive markets, such as PJM Interconnection, (6) where the Independent System Operator (ISO) is responsible for grid operation and power plant scheduling. Yet the implementation of an ISO across Europe is not expected for years to come; hence, a more efficient method of congestion management in Europe will have to rely on zonal pricing, at least in the medium term. In that setting, cluster algorithms are applied in order to identify improved BZCs. Those algorithms cluster the nodes of a system into bidding zones. Although various methods have been used, clustering of locational marginal prices (LMPs) and clustering of power transfer distribution factors (PTDFs) are the two major approaches. Clustering of PTDFs is done, for example, by Klos et al. (2014) and Kang et al. (2013). Bergh et al. (2016) also cluster PTDFs. They present a case study for CWE and try to quantify the impact of the number of bidding zones on market outcome. The different bidding zones are obtained by clustering nodes with similar nodal PTDF-values for congested lines. Clustering of LMPs has been applied by (amongst others) Imran and Bialek (2007), Burstedde (2012), Wawrzyniak et al. (2013), Breuer, Seeger and Moser (2013) and Breuer and Moser (2014). Burstedde (2012), for example, clusters nodes to zones based on similarity of nodal prices using a hierarchical cluster algorithm following Ward's criterion, whereas Breuer and Moser (2014) apply a genetic algorithm developed in Breuer, Seeger and Moser (2013) that is based on LMPs to a large-scale model of the European transmission system. They assess the impacts of the optimized bidding zones on operational costs, network security, and market efficiency. Additionally, the authors of this paper have developed an enhanced approach to LMP-based clustering that weights nodes according to their relevance for an electrical network (cf. Felling and Weber, 2016) and which is able to compute robust configurations against a given set of uncertainties (cf. Felling and Weber, 2018).

   In contrast to that, there are several studies that evaluate not endogenously determined, but exogenously given bidding zones. Trepper, Bucksteeg and Weber (2015) investigate a split of Germany into two bidding zones and analyze both the occurring price differences and the distributional effects. Egerer, Weibezahn and Hermann (2016) analyze the implications for the German power market if it were to be divided into two or four bidding zones. One of their results is that price differences between zones could be very high, but could occur in a restricted number of hours. They conclude, however, that the results are dependent on the BZC. Blume-Werry, Huber, and Everts (2017) use the fundamental model Green-X in order to analyze the impact of splitting the common German-Austrian bidding zone. One of their key results is that the split has nearly no effects on the security of supply of both countries.

   The second relevant literature stream, as mentioned previously, relates to "investment decisions and risks." Practitioners most commonly assess large investment projects based on the net present value, which is determined using the discounted cash fow method (cf. Brealey, Myers and Allen...