A comparative analysis between the Dutch and German electricity distribution network industry with regard to regulation and efficiency

Transcription

1 A comparative analysis between the Dutch and German electricity distribution network industry with regard to regulation and efficiency Master-Thesis Author: Stefan Mißfeldt Academic Supervisor: Prof. Frederic Warzynski Program: MSc. International Economic Consulting Date: Aarhus School of Business, Aarhus University Department of Economics and Business

2 Abstract The electricity distribution sector is regulated to overcome market failures due to the existence of regional monopolies, imperfect information and to balance between public as well as private interests. Incentive regulation has been implemented in order to mimic market forces which can stimulate companies to become more efficient and prevent overinvestment. This study compares regulation and efficiency of electricity distribution operators in the Netherlands and Germany. After a description of the underlying theoretical frameworks and its applications, the paper examines relative technical efficiency of all 8 Dutch operators and a representative sample of 32 German grid operators using Data Envelopment Analysis. The base model uses network length as proxy for capital expenditures and labor as proxy for operating expenses. Outputs are characterized by peak load, the number of connections and the structural variable connection density. Results show, on average Dutch operators are more efficient compared to the majority of German operators. Large companies in terms of grid size and labor generally appear less efficient than small or medium sized companies. However, there is a limited amount of companies in the German sample. Consistent with peer literature, the results confirm the absence of strong scale economies in electricity distribution. I

3 Contents 1 Introduction 1 2 Regulation of electricity distribution system operators Sector description Theoretical framework within Industrial Organization Natural monopoly Performance problems in natural monopolies Purpose and goal of monopoly regulation Types of regulation Assumptions and frameworks Traditional schemes Incentive Schemes Application in the Netherlands and Germany Overview Price cap vs. Revenue cap Yardstick vs. Benchmark Design effects on company behavior Qualitative Analysis Electricity prices and network tariff Quality indicators Investment indicators Relative performance in major cost drivers Quantitative Analysis Methodology Defining efficiency Overview benchmarking approaches Data Envelopment Analysis Data Estimation results Conclusions 69 References 72 II

4 A Appendix 76 A.1 Relative performance in major cost drivers A.2 Analysis of German frontier firms relative to Dutch DSOs A.3 List of Assumptions A.4 Descriptive data for the individual country samples A.5 Technical efficiency scores A.6 Comparison Dutch DSOs technical efficiency and cost efficiency A.7 Average technical efficiency score III

5 List of Figures 1 Regional seperation of Dutch and German distribution companies Cost-based vs. Incentive-based cap regulation Evolution of regulation schemes Actual revenue vs. efficient revenue Household electricity prices Base load electricity prices Netherlands relative to Germany Change in network tariff SAIDI Germany and the Netherlands Total network length excluding high voltage lines Total sector investments Input requirements Technical and allocative inefficiency Benchmarking approaches Returns to scale Comparison technical efficiency scores Total grid length and technical efficiency List of Tables 1 Overview Dutch and German DSO characteristics Managerial incentives vs. Rent extraction Overview Dutch and German Regulators Average weighted annual network tariff for household customers Model specification List of model specifications Descriptive data for the entire sample IV

6 1 Introduction Regulators in Netherlands and Germany face similar challenges. While regulating the monopolies of electricity distribution system operators (DSO), authorities introduced incentives in order to stimulate network companies to increase efficiency. On the one hand, consumers can benefit from efficiency gains as prices decrease. On the other hand, regulators have to balance between promoting efficiency, keep high quality of networks and encourage investments in grid capacity. Integration of smart meters, decentralized generation and electric vehicles will demand grid operators to invest heavily in network assets. In order to account for that, authorities in both countries offer different instruments. Goal of this thesis is to explore the differences in regulation and calculate relative efficiency of distribution system operators. More precisely, what are the different instruments used by the regulators in the Netherlands in Germany and how is their effect on grid tariffs, quality and investments? After understanding the underlying regulatory frameworks, the second purpose is to reveal technical efficiency of Dutch and German grid operators. Based on the results, physical characteristics will be considered in interpreting efficiency scores. Is there a large difference between German and Dutch operators? The entire energy sector experienced structural reforms in the last decade such as the EU Acceleration Directive in Distribution utilities were recognized as non-contestable regional monopolies. Electricity grids serve as intermediary in the value chain and charge a tariff for connecting electricity generation and consumer. Regarding size of the industry, efficiency gains within the sector of even 1 percent in tariffs can lead to increased consumer welfare of millions of Euros. Achieving these gains can be difficult, because there are a number of issues. Operators usually control most of the information. Thus, regulators frequently use benchmarking techniques to compare utilities and retrieve information on efficiency potential for use in incentive-oriented instruments. However, there is a lack of data to measure efficiency and systematic collection has only begun in the recent years. Next to that, robust application of parametric methods requires a large set of companies and observations. With only 8 companies it is not possible for the Dutch operators to run such an analysis, whereas the large number of German operators allows for that. Benchmark design can cause compatibility issues by assuming all operators are similar, when in reality they are not. Regional differences can have impact on operators efficiency. Besides, there are various approaches to derive efficiency. Without availability of costdata, technical efficiency is frequently estimated by comparing physical characteristics between peer companies. Another approach is the use of financial data to derive relative efficiency including the cost-dimension. This additional information can reveal the degree of efficiency in the use of capital. Therefore added value is information on the current status of efficiency between Dutch 1

7 and German electricity networks. In particular, the anlysis compares results from qualitative and quantitative approaches. Study of relevant indicators identify, whether emphasis on efficiency has a negative impact on grid quality and capacity investments. Beyond that, the paper discusses potential scale economies in electricity distribution. So far, there is no study comparing Dutch and German DSOs in one benchmark. In order to answer the objectives mentioned above, this thesis starts out with a theoretic analysis of monopoly regulation in the context of Industrial Organization. Next, regulatory instruments and their application in the Netherlands and Germany are compared and contrasted. Further, qualitative indicators on outcomes with respect to tariffs, quality and investment are set into relation. The quantitative analysis continues with a monetary comparison of major cost-drivers between Dutch operators and German sector indicators. Public available data for the Netherlands is compared with statistics of a proprietary dataset called "Regulierungsdatenpool". Integral part of the analysis is the firm-level efficiency benchmark between all 8 Dutch DSOs and a representative sample of 32 German firms. Since financial data is unavailable, public information about physical characteristics of DSOs was collected, based on data from Data Envelopment Analysis (DEA) is used to determine technical efficiency. The regular model is adjusted, using 9 different specifications to control for changes in scores, rankings and the impact of different scale assumptions. Outcomes from the benchmark serve as base for discussion of firm characteristics on their respective efficiency. The thesis proceeds as follows. Chapter 2 provides an overview of the sector and explains the theoretical framework regarding regulation of natural monopolies. It contains a literature review on monopoly industry structures, corresponding types of regulation and application by the two national regulators. Following theory, chapter 3 observes qualitative indicators on the "markets" including a cost-level analysis of major cost-drivers. In order to produce a firm-level analysis, chapter 4 provides an introduction to the measurement of efficiency and explains the technical efficiency scores, obtained by using DEA. Chapter 5 concludes. 2

8 2 Regulation of electricity distribution system operators Chapter two provides an overview about regulation of distribution companies within the wider context of industrial organization. Moreover, its application is discussed in the Netherlands and Germany. The argumentation starts with a review of the sector, followed by a description of typical performance problems in monopolies. Then, the role of regulation is set in a context to its historic development and present application. 2.1 Sector description Historic context This section offers information about the electricity sector and the operating environment. Networks emerge in order facilitate the transport and exchange energy carriers like electricity between generators and consumers. For electricity distribution it is not economically reasonable to connect every consumer to the generator. Instead, it is more cost efficient to use one network as shared platforms to transport a standardized good such as electricity compared to individual connections. This is the reason, why today distribution networks run the transport and delivery of electricity to end-consumers within local monopolies. From a historical perspective this has not always been the case. Deployment of gas supply for city lightning in the United Kingdom started around the year 1812 and started to become a competitive industry (Jamasb & Pollitt, 2007). This development was ended in the UK in 1860 with the Metropolis Gas Act by allowing local natural monopolies. Invention of the light bulb and technologies such as electric engines gave rise to the electricity industry in the late 19th century onwards. After realizing that networks are more cost efficient over competitive connection, electricity grids appeared. This development showed later in the Netherlands and Germany at the beginning of the 20th century, as industrialization took longer to materialize. Gradually, electricity grids replaced gas as main form of city lightning. Within the next decades, electricity distribution became officially recognized as natural monopoly. In Germany, legislation pointed out that: monopolies should limit the dangerous economic effects of competition 1. Concession agreements are issued by the municipality to the network operator for the use of public areas. The incumbent monopolist was given exclusive rights to supply its area by the municipality. Further, the legal text granted exceptions from antitrust law. The law was revised several times, reflecting the changes occurring in the upcoming decades. Parties involved This part of the text introduces the main players and explains main characteristics of the electricity industry. The market for electricity distribution shows 1 German Energy Act "Energiewirtschaftsgesetz" (EnWG, 1935) 3

9 different concentration characteristics for Germany and the Netherlands. Whereas in the Dutch market we can observe currently 8 distribution network operators, Germany is much less concentrated. The total amount of German DSOs is above 800. Despite a German population about 5 times as much compared to the Netherlands, Germany displays a much higher degree of fragmentation. Notable is the large number of regional and local distributors in Germany. Both countries have established independent regulatory authorities. Until about 1990, the Dutch market consisted of a larger number of vertically integrated firms. Business included production, distribution and supply of electricity. Afterwards, significant consolidation took place, reducing the number of players. In recent years, the mandatory ownership unbundling in the energy sector was supposed to separate the grid operations from the associated parts of the value chain. However, in the Netherlands it has not been consistently implemented due to several court decisions. A final decision is still pending. Distribution system operators were established as independent entities, entirely committed to grid operations. The operation of the high voltage transmission assets was handed on to the only remaining TSO Tennet. Following that process, to date there are 8 DSOs and one TSO in the Netherlands 2. The largest electricity distribution companies in the Netherlands are Liander, Stedin and Enexis. Together they control more than 80 percent of Dutch electricity distribution in terms of connections. The sector is in public ownership as privatization is prohibited. There are about 8.1 million connections in the Netherlands 3 creating demand of 109 TWh per year. The total length of the Dutch grid was kilometer 4 in 2010 (Netbeheer Nederland, 2011a). For historic reasons, the situation in Germany is different. Industry structure supported decentralized generation favored by policy because it was less vulnerable from a defense point of view prior to the Second World War. Market forces formed four large vertical integrated enterprises since then but the large number of small and medium sized companies remained. To date there are 866 distribution network operators in Germany by About 60 are regional suppliers. A common form are "Stadtwerke" - type of firms. The word is composed of Stadt = city and Werke = plant. Historically, these local utilities were responsible for the supply of cities and the surrounding areas. Out of these, more than 700 DSOs have below customers and can be considered to operate local (Bundesnetzagentur, 2011a,c). Figure 1 illustrates the differences in DSO seperation within both countries. The operating area of the three largest Dutch operators is marked. It is visible, that German electricity distribution industry is more dispersed. According to the German Federal Cartel Office, 2 Numbers are based on Total Dutch population counts 16,6 million citizens by Excluding high voltage assets. 4

10 Figure 1: Regional seperation of Dutch and German distribution companies Source: Energie in Nederland, Nikogosian, Vigen; Velth (2011) there are 850 geographically seperated areas, which are relevant for market investigations. About 20 percent of German DSOs are legally unbundled, 75 percent are vertically integrated and 5 percent fully seperated (Nikogosian, Vigen; Velth, 2011). The largest players are the network companies of E.ON, RWE, EnBW and Vattenfall Europe, sharing about 40% of the power distribution market. The transmission business is currently being owned by four companies: Tennet, EnBW, Amprion and 50Hz Transmission after legal unbundling. Electricity distribution in Germany is in public as well as private ownership. In total, there are about 46,9 million 5 electricity customers in Germany 6, who make up market size of total of 511 TWh, the largest in Europe. The total grid length in 2009 is 1.71 million kilometer including high voltage lines (BDEW, 2010). While distribution business is in public ownership in the Netherlands, it is both public and privately owned in Germany. An overview about the distribution industry is given in table 1. Independent regulators were set up in both countries following market liberalization. After the Electricity and Gas Act in the Netherlands in 1998, DTe was appointed as regulator, today known as Energiekamer. It is part of the competition authority (Nederlandse Mededingingsautoriteit, NMa). In Germany, the Bundesnetzagentur (Federal Network Agency, FNA) was created in 2005 to regulate various network-based industries including electricity and gas. It emerged from the Federal Ministry for Mail and Telecommuni- 5 2,5 million are industrial or commercial and 44,3 million are household customers. According to FNA Monitoring Report 2011 based on numbers of the 2010 survey. 6 Total number of citizens in Germany equals 81,7 million in

11 Table 1: Overview Dutch and German DSO characteristics Netherlands Germany Number of DSOs Number of Connections (million) Total grid length (million km) Ownership Public Public and private Sources: Energie in Nederland, FNA Monitoring Report cation. Additionally, European institutions such as the European Parliament and Council can set supranational legislation, that have direct impact on domestic law and proceedings. Value chain The electricity value chain consists of generation of electric power, its transmission over large distances, local distribution, finally wholesale or retail segment. Distribution networks are responsible for delivery of current from the transmission lines into customers property. The customers can be either of industrial, commercial or residential type. As an intermediate, the degree of network efficiency has direct impact on retail prices for end-consumers. Thus, it is desirable to thrive for efficient networks reducing network tariffs. Distribution networks are composed of different layers, depending on the voltage level. Networks operate usually in high-, medium- and low voltage levels. Depending on operators size, they serve one or more layer. Distribution networks typically operate in low and medium layers. Transmission networks instead manage the high voltage lines and cables, that transport electricity over long distances with fewer loss compared to medium and low voltage. Today, generation and supply are regarded as competitive industries, whereas the transmission and distribution have natural monopoly characteristics, subject to regulation. A consumer can therefore choose its supplier, not the grid operator for a given location. Latter will be contracted by the supplier for delivery of electricity through its network. Electricity itself is a homogenous good, that cannot be stored in an efficient manner (at the moment). The difficulties to store electricity imply that supply and demand have to be met simultaneously. Another important aspect is the high level of capacity intensity and long economic lifetime of assets in electricity distribution. Demand is constantly growing due to increase in population and use in electricity consuming devices. Renewing the assets on a periodic base will cause investment cycles. Therefore, careful long term planning becomes essential. Unbundling A worldwide trend since the late 1980s reformed the institutional frameowrk, organizations and operating environment of the power industry. The reform packages caused a couple of changes in the industry. Competition was introduced to generation 6

12 and supply, sector regulators were set up and transport of electricity was unbundled from its generation and supply activities. Unbundling is the split of network activities from the remaining business along the value chain. To capture the reason, changes in regulation must be viewed in the context of reform in the whole sector. The EU s overall target is to establish a common market for energy (European Commission, 1996). This introduces higher degree of competition into the segment with the ultimate target of consumer benefits by offering lower prices and freedom of choice. Unsuitable for introducing competition, monopoly behavior and vertical integration of utilities were hurdles on this path. From a firm perspective, vertical integration is desirable to realize economies of scale and synergy effects. On the downside, vertical integration can also be used to cross-subsidize inefficient elements of the value chain, avoid transparency and exercise market power. Additionally, due to the network nature, the grid operators can build up significant entry barriers for third parties. In order to overcome these issues, one of the most important changes was initiated by unbundling the electricity sector. Until recent years, larger firms were integrated owning transmission and distribution networks. The Acceleration Directive of the European Union (European Commission, 2003) aims to foster the liberalization and creation of the single, common European energy market. Requirement for that is a non-discriminatory, transparent and reasonable priced network charge. As a result, transmission and distribution business was legally separated from the other parts of the value chain. This requirement was translated into national law by Dutch law and German legislation (EnWG, 2005). The process is also known as legal unbundling. Unbundling is practiced in legal, operational/functional and information/accounting dimensions. The following criteria apply: DSO/TSO may not participate in company structures of the integrated entity directly or indirectly on a day to day basis. Further, it must have capabilities to act independent from the previous vertical integrated unit and have the necessary decision making rights. A compliance program must ensure a nondiscriminatory behavior (Gómez-Acebo et al., 2005). This way, the regulator wishes to avoid abuse of market dominance gained by vertical integration. It applies for companies with over connections. In Germany, approximately 760 DSOs were exempted due to its size below the connection threshold. 2.2 Theoretical framework within Industrial Organization Natural monopoly Definition This section provides context to the wider perspective of Industrial Organization. Within the field of industrial organization, market structure and behavior of the firm are in the center of interest. In particular, the behavior of the firm within the bound- 7

13 ary of regulatory constraints. In general, a market shows monopoly characteristics if there is only one supplier. From an overall welfare perspective, monopolies are undesirable. A monopolist faces a downward sloping demand curve, indicating it can restrict output to gain higher prices. This is detrimental, because the monopolist is a price maker, adjusting its output by marginal costs equal to marginal revenue. Compared with a long-run equilibrium of a competitive market, the monopolist restricts output and raises price (Martin, 2001). However, for electricity DSOs economic efficiency dictates that demand is supplied by a single entity due to its network character. Costs of constructing a competing distribution network are so high, blocking entrants from joining the market. We speak of a natural monopoly. In this case, costs of supply by a single entity is always lower then by multiple firms. In other words, a natural monopoly exists if the cost function of an industry is such, that no other combination of two or more firms can produce at a given level cheaper output than a single firm (George et al., 2000). An alternative definition shows Posner, who finds a natural monopoly does not refer to the actual number of sellers in a market but to the relationship between demand and technology of supply (Posner, 1969). In the context, technology of supply refers to the single electricity grid, whereas generation capacity can be supplied by a variety of firms. Following Joskow (2007), this cost-dominance relationship must hold over the full range of market demand Q = D(p). Assuming a single, homogenous output like electricity, where k firms produce output q i with total output Q = k q i. Each firm has a cost function C(q i ). Existence of a natural monopoly can be re-drawn as: C(Q) < (q 1 ) +C(q 2 ) C(q n ) Firms with the cost functions above are said to be subadditive at output level Q. Likewise, the necessary condition for a natural monopoly to exist is that the costs of that good is subadditive (2007) at Q. Production costs increase only under-proportional. Cost function of firm i can be defined: C i = F + cq i Accordingly, firm i s average cost of production is: AC i = F (q i + c) Average costs decrease over the relevant range of q, showing economies of scale. With rising values of q, share of fixed costs F per unit decrease. In the case of natural monopolies, it might even be less costly to produce output only by one firm above the point, 8

14 where there are economies of scale compared to multiple producers. The cost function is still subadditive for certain values of q beyond economies of scale. Market demand at this point is not large enough to allow efficient production by two firms. In the utilities context, economies of scale can be replaced rather by economies of density because for grid operators economies arise from supplying customers in a geographic area. While looking on fixed costs, it is important to take the importance of sunk costs into account. Investments in electricity grids relate to that concept, as they cannot be recovered. Once paid, investments do not matter for the incumbent anymore 7. This leads to a variety of other performance problems in monopolies Performance problems in natural monopolies Markets with natural monopoly character usually display several performance problems like excessive prices, production inefficiencies, undesirable distributional impacts and poor service quality (Joskow, 2007). Lower economic efficiency or market failure can occur due to various reasons. Price signals can be distorted due to prices above marginal costs. The extent to which the firm can increase prices depends highly on price elasticity of consumers. For necessary goods such as energy elasticity is rather low. This way, the producer can increase its rent on the expense of consumers. This can lead to dead weight losses from a total welfare point of view, representing the efficiency loss for society. Another factor contributing to lower efficiency is problem of higher production costs. The firms may not be successful in minimizing production costs. Monopoly firms isolated from competition often exhibit misallocation of resources, also called X-inefficiency according to Leibenstein (1966). Often it is related to lack of managerial effort that leads to higher production costs. Non-existence of market pressure lacks necessity to reduce costs. Therefore, monopolists can enjoy higher producer rent and thus lower the welfare by consumers, because they will pay a higher tariff. If a monopolist is reimbursed all its costs by a regulator, it does not have any incentive to reduce costs in order to benefit customers. As it does not encounter any competition, it can sustain such inefficiencies at least for a longer period of time compared to the case with competition. For these reasons, distribution networks were historically kept in public ownership. This brings up the issue about information asymmetries. Due to a lack of competition, information on cost structures and managerial effort remains uncovered. For example, there is no competitive pressure that pushes prices close to marginal costs. A third party has fewer information compared to a competitive case. Even in a regulated environment 7 From a theoretical point of view, it does matter for a potential entrant into the market. It can choose between replicating the network assets or paying a tariff for the use of the monopolists grid. It gives the incumbent an advantage, that it can exploit for strategic behavior. While this applies for telecom networks, it does not for electricity networks. 9

15 this inefficiencies can persist, at least until detected by authorities such as the regulator. By looking at quality of service and related issues, a third factor becomes evident. Within energy networks, quality refers to the security of supply, outages of the network and other factors related to quality (e.g. hotline services). Generally, there is bias in quality related issues when a monopoly is present (Joskow, 2007). The difference becomes apparent, when contrasting welfare. A simple profit maximising monopoly looks at the willingness to pay for quality of the marginal consumer. On contrary, social welfare is maximized by focusing on the surplus achieved by the average consumer (Spence, 1975). Additionally, high level of sunk costs amplify the matter of break-even constraints in a regulatory context with price and revenue caps. Private firms will only supply goods and services, if they can at least recover the cost for providing them. If private firms will not expect to earn enough revenue to cover its costs, it will most likely not provide the services voluntary. For example, in absence of regulation and legislation, suppliers will then most likely not provide its goods and services to remote areas at a reasonable price. Investments in future capacity requirements will most likely not be made. Next to static inefficiencies, the natural monopoly also tends to show dynamic welfare losses with same reasons as static inefficiencies. According to Mueller et al. (2010), dynamic inefficiencies can lead to a distorted and delayed adaption of a changing economic environment as well as a lower willingness for technological progress. Public interest plays an important role, therefore considerations about political economy cannot be omitted. Interest groups, whose welfare is changed because they are exposed to certain effects of regulation, can put pressure on or alter existing regulation schemes. Imperfect conditions for both quantities and prices stimulate intervention to address the monopolist from taking advantage Purpose and goal of monopoly regulation Purpose of regulation is to limit impact of performance problems or even to remove them. The overall goal is therefore to maximize social welfare, limiting rents transferred due to given market imperfections. Itself, the regulator should be independent, offer transparent processes, review financial and investment plans as well as monitor terms of service. Yet, it is a fact that regulation itself is imperfect and might lead to higher costs in the market and unanticipated responses by firms subject to regulation. Institutions, who are about to implement regulation have to consider, whether net effects of regulation as response to imperfection improve performance in contrast to unregulated industry. Therefore, regulation has to be careful designed in order to leave the market better off than imperfect unregulated markets. Ideally, regulated firms can be incentivized to apply necessary changes by themselves. As a first step, regulators have to assess the nature and magnitude of performance prob- 10

16 lems. Then check, which instruments can be employed to approach the problem and its potential strong- and weak points. In the next step, authorities must find out the impact of proposed regulated environment. Regulatory goals, with regard to electricity distribution, can be summed up in three main dimensions: efficient costs and prices to support retail market adequate grid quality sufficient amount of investments over time to ensure capacity Efficient costs and prices Consumer welfare will be lower in case of excessive monopoly prices, regulators goal to restrict monopoly pricing is to reduce costs to obtain efficient costs. Within the literature following Farrell (1957), efficiency of the firm can be split in two categories: technical efficiency and allocative efficiency (or pricing efficiency). Technical efficiency or productive efficiency is achieved by realizing a maximum of output with a minimum of input (Mueller et al., 2010). Allocative efficiency is the ability to select optimal input proportions with given input prices, ideally equal (or close) to marginal costs (p=mc). A firm can be productive efficient without being allocative efficient, where social welfare is maximized. The relationship can be established as: Cost efficiency = technical efficiency + allocative efficiency (Fried, Harold O.; Knox Lovell, CA.; Schmidt, 2008). Constraints and imperfections discussed above will in case of monopolies cause prices to deviate from that. Hence, second-best pricing is often referred as as regulatory goal. The existence of allocative efficiency does not imply that costs are efficient. Reasoning behind the cost efficiency goal is that lower costs will reflect in lower consumer prices. In other words, technical efficiency reflects efficiency based on its physical in- and outputs, rather than cost-indexed in- and outputs as measured by allocative efficiency. Not having any competition, firms can exploit full economies of scale and scope. At the same time, there are no incentives by lack of competition to decrease costs. Instead, regulators must create an environment where regulated firms minimize costs by adjusting inputs to reflect relative input price and show the necessary managerial effort to control costs and other sources of X-inefficiency (Joskow, 2007). Adequate quality Another important goal is adequate quality of service. In the context of distribution networks, quality refers to the reliability of electricity grids and its fair distribution. A grid is reliable, when it does not experience frequent outages below a certain amount of duration. Also authorities want to make sure, that quality is achieved on a certain level, because technical specifications do not allow it to change by individual level, rather on the array of neighborhoods (2007). At last, regulatory goal is providing 11

17 incentives that keeps firms focus on cost efficiency while continuously providing constant technological foundation or quality levels. Yet, there are difficulties to set the appropriate level or value of quality. Sufficient investments In order to ensure reliability targets and connection of additional customers, networks capacity has to be expanded step by step. Regulators goal is therefore to ensure, the grid operator makes the necessary investments to expand the grid capacity accordingly to ensure a reliable network. Especially the integration of renewable and decentralized generation require investments to the grid. Not just extension of the grid, also technological advancements such as smart meters show a need for large capital expenditures. However, the efficiency goal is measured in the short run over a regulatory period and investments in grids are long-term oriented. Firms must be efficient by cutting costs and at the same time be able to maintain and expand their networks. There is a trade-off between allocative and productive efficiency 8. In order to become efficient in the short run, the firm can become productive efficient while ignoring capacity investments for the future. In the long run, this is not efficient, because future capacity shortages can lead to severe losses to consumers. After all, firms do not take losses to society into account for their investment decisions. In the real world, constraints cannot achieve this target because of constraints and rent seeking behavior. However, in practice regulators aim to approach to aim at a balance between productive and allocative efficiency. This way, firms providing networks should be able to cover its costs without making excess profits, but high enough to induce the firm to make the necessary investments. Dynamic inefficiency Existence of allocative and productive efficiency are both static efficiency measures. Alone, they only look at one moment in time. Dynamic efficiency potentials are instead by determined by investments in new technologies and processes. Total sector welfare is maximized, if efficiency over time is at maximum. Because adaption to future demand environment needs investments, temporary inefficiency can exist (Mueller et al., 2010). Regulation must therefore focus on static and dynamic efficiency. In other words: dynamic efficiency can lead to temporary static inefficiencies due to investments and investment budgets, which are sunk costs most of the time. In order to incentivize a firm to do so, it has to be convinced that investment is profitable ex ante and it can recover its current costs. For regulation to be sustainable in the long run, It gets evident, a right balance between efficiency targets on the one hand and room for expansion to ensure sufficient capacity. 8 A more detailed discussion will follow in a later section. 12

18 Regulation must not be too tight to prevent the firms from investing. Likewise it cannot be too lax, lowering effort by the firm to reduce the cost base. Operators are caught in a target conflict between: efficiency targets, structural changes and technological progress while maximizing their profits under regulatory constraint. Prime paradigm for design of regulation is therefore to influence management of firms to act in all shown dimensions. Characteristics of well-performing networks Assessing the performance of networks is difficult. This section lines out, which characteristics literature regards as important for the performance of electricity networks. It is best answered from a consumer-orientation. A network fulfills its purpose in a satisfactory way, if its output fully satisfies the users. Main characteristics of well-managed networks include affordable pricing for network access, a high degree of reliability and a secure supply. Efficient grids operate with minimum of expenses, allowing low charges for its use. This way, efficient networks do not inflate the electricity price more than its necessary part. Along with affordable prices, electricity has to be delivered over the grid in a timely manner. For that, capacity must be available both in sufficient amount and at any time. From a perspective over the whole value chain, capacity shortages would lead to higher prices, decreasing welfare. Higher prices can occur because a lack of capacity will lead to bottlenecks in the grid, causing congestion. Generated electricity cannot be delivered to the location of demand anymore in a timely manner. This way, producers of electricity located geographically close to the demand gain an advantage because they are still able to satisfy the demand. Local pockets of market power can emerge, who can charge a "congestion" - premium. For an efficient grid, distortions in delivery due to capacity constraints do not occur. At the same time, grid operators must be adequately compensated for their effort. From a technical efficiency point of view, optimal networks supply a maximum of electricity output and capacity to customers, given a minimum of physical input. In other words, well managed grids require less labor and capital input to deliver a certain amount of electricity to a maximum amount of connections. By minimizing inputs that it pays for and maximizing output which it can charge, a network company can achieve higher values of technical efficiency. Attributes of technical efficiency will be discussed in following chapters. High levels of quality are sign of a desirable network. An increasing amount of outages lowers the perceived quality of networks by the consumer. A well managed grid shows a low number and duration of outages per customer. Quality goals depend on preferences: quality can also be too high so that the consumer pays for an excess of capacity, which is not even needed in the future. Additionally, there is growing demand in society to have energy generated by clean green energy. Integration of renewable generation means large investments in the grid capacities to maintain reliability. Thus, a good network an- 13

19 ticipates the efforts necessary to adapt for such changes in the future by undertaking the means. How different regulatory schemes try to reach good network characteristics is described below. 2.3 Types of regulation Assumptions and frameworks This section explains popular regulation schemes and its attributes. Discussion will start out with a short description of pricing under a fully informed regulator in order to understand the reference scenarios. This (unrealistic) assumption will be relaxed and lead to modern incentive regulation schemes taking into account information asymmetries between regulator and firms. Contents on the following paragraphs are based on (Joskow, 2007). Pricing under full information Assuming the regulator is fully informed about costs of regulated companies: the most simple version of regulation is a linear pricing scheme. In order to break even (firm viability constraint), prices must be higher than marginal costs. A simple version in a single product case is to set price per unit equal to average cost. This scheme where each sold unit has an equal mark-up to all consumers can be improved from a welfare perspective by introducing third degree price discrimination. That is, to charge different prices for different types of consumers with varying degree of willingness to pay. Given this variation, so-called Ramsey-Boiteux pricing offers an optimal condition maximizing social welfare with respect to the break-even constraint. In this setting, prices are set in a way that the difference between a product s price and its marginal costs varies inversely with the elasticity of demand for the product (Joskow, 2007). As a result, tariffs are between marginal cost prices and pure monopoly prices, engaging in third degree price discrimination. Still, Ramsey-Boiteux pricing is only a second best solution because it still results in higher than marginal costs, so there is still a deadweight loss. By further relaxing restrictions, a better outcome can be achieved. The idea behind that is: splitting the tariff T in a two parts, a fixed access charge F and a per-unit price p depending on quantity q consumed: T i = F + pq i Compared to the previous scheme, this is a non-linear price due to decreasing average expenditures per unit for the customer. Non-linear pricing rules can be improved by setting the fixed charge payment equal to firms average fixed costs. Assume the regulated 14

20 firms cost function is C = f + cq, fixed costs f and marginal costs c. Each of the N customers faces fixed charge A = f /N and a unit charge is set at marginal costs (p = c). A tariff can then be written as: T i = A + pq i = f N + cq i In this scenario, consumers pay the fixed fee to cover the firms total costs and consume at an efficient level, towards allocative efficiency. This an optimal solution in the case, when consumers are identical. As consumers are not similar in reality, have different demand and A may vary largely. Then it would be optimal to tailor tariffs for each consumer, if the regulator would be fully informed. Yet, in the real world regulators do not know each consumers demand or willingness to pay but can observe some demand patterns. However, regulator can divide between a high demand and low demand case. One way to model different demand is to provide a menu of two-part tariffs: T 1 = A 1 + p 1 q 1 T 2 = A 2 + p 2 q 2 When A1 < A2 and p1 > p2 c, consumers with a low demand choose T 1 and high demand consumers choose T 2. This way, optimal non-linear prices reduce distortions due to differences in demand while maintaining firms budget constraints. The next step is to apply the considerations to industry requirements. Electricity cannot be stored economically but at the same time demand can vary significantly due to weather, time of the day or season. Therefore, capacity must be expanded to meet maximum demand, including equipment outages. Situations of maximum demand is called peak load, otherwise base load satisfies average demand. This expansion factor between low and high demand can vary by the factor of three (2007). Capacity investments are sunk costs. Long run marginal costs of peak capacity include (1) the costs of building and (2) operating an incremental share of the peak capacity. Efficient prices must therefore include both components to compensate the operator for production of peak load capacity as well. That is why social costs of peak load in a dynamic context are much higher than static marginal costs. Efficient prices must signal this to consumers by setting p peak > p base. In an optimal case, additional price during a peak period reflects the sum of marginal operating and capacity costs, whereas off-peak period pricing (base load) only marginal operating costs. Difference in price does not equal price discrimination, rather than represents two different products each consumed in a different period of time still embody shared costs (for base load). Ideally, prices can be varied by more than just peak and offpeak but that would require regulator to know each individual demand. Metering stations can help to track and forecast demand, revealing certain demand patterns. It becomes evi- 15

21 dent regulators are neither fully informed nor fully uninformed. The following paragraphs describe relationship between firm and regulator in the context of limited information. Principal Agent Theory As response to market imperfections and targeting regulators goals, several improvements emerged over the last decades. Neoclassical models simply assume cost-minimizing firms. On contrast, agency models recognize the fact that regulated firms do not necessary minimize costs and use private information to their advantage. Design of regulation must address fundamental problems arising from limited information. Applying Principal-Agent theory with information asymmetry can help to develop more useful regulation. By default, regulators (principal) have incomplete information compared to the firm (agent), which has private information. Details about cost structure is not entirely observable outside the firm, so the regulator is unaware of actual costs next to price signals. Managers do not only have the choice to select a proper input mix but also the degree of their managerial effort in cost-minimizing. Next to that, firms are much better informed about demand they face and its elasticity. Two common problems arise: First, regulators do not know the exact value of marginal costs (adverse selection). Second, managerial effort to reduce costs cannot be observed (moral hazard). Yet, knowledge about this information is necessary to derive proper regulation and incentive schemes. As long as these issues persist, the firm may increase its profits for the disadvantage of consumers (Laffont & Tirole, 1993), act strategically and game the regulator (Jamasb et al., 2004). The effects on welfare distribution among market participants vary by design of regulation scheme. Consequently it is of interest for the regulating institution to reduce the information asymmetry as much as possible. The regulator can however monitor the realized costs ex post through audits and hearings. Additionally, standardized accounting rules have been established to improve data quality. If sufficient amount of firms are present, costs can be benchmarked against each other and reveal relative efficiency, thus reducing information disadvantage of regulators. Accordingly, to deal with principal agent problems firms must be incentivized to perform efficiently Traditional schemes Extreme situation: Cost of service vs. Price cap Assuming regulators objective is welfare maximization, it has to limit rent transfer from consumers to producers under break even constraint. By defining regulators objective function as a basic sum of consumer/taxpayer (S) and producers rent (R), regulation can decide the weight (a) between both parties (Armstrong & Sappington, 2007). This relationship can be drawn as: 16

22 W = S + αr Allowing a having values between 0 and 1, regulator can decide to put emphasis on consumers welfare by setting a < 1. In the case of a > 1, shareholders are in the center of interest. In principle there are two choices when setting regulatory schemes: following a static or dynamic type? A static setting will set tariffs every year in order to cover costs that are considered reasonable. The dynamic way is fixing tariffs for a period of time. After this time is over, tariffs are re-set to a different amount. This allows to decouple costs from revenues. Careful design can utilize this decoupling to generate incentives for cost reduction. It is up to the firm to become efficient on a cost side, increasing risk for the firm. In order to derive optimal schemes, two extreme cases serve as reference scenarios according to Laffont and Tirole s considerations. For several decades static Cost of service (or rate of return) contracts were common. Main principle is: firms announce costs and the regulator reimburses accordingly all production costs. Advantage under cost of service: there is no rent left in the form of excessive profits, as only costs will be reimbursed. This solves the rent extraction issue, leading to no excess profits. Additionally, the system does not expose the company to a high risk, is straightforward and easy to implement. Disadvantage of this system is the danger of inflated costs or over-investment. When the firm can simply recover all costs, there is too little incentive for management to engage in cost-reducing effort. In case of cost saving effort, managers would retain zero percent return. This phenomenon is also called gold plating - using expensive inputs leading to high costs. As a result, it is likely that customers pay a too high price following high costs generating inefficiency. In other words: the moral hazard problem with unknown cost remains, while adverse selection is solved because there is not much rent left for excessive profits. Another major problem using cost of service regulation is regarding the valuation of the specified return on the firms investments. Long-term asset stocks within the regulated asset base (RAB) must be transferred into cash flow streams over its entire lifetime that reflect its value. Challenge is to set prices at a reasonable level that firms can recover its cost of investments. However, this level is difficult to obtain under incomplete information. If the sum of discounted asset cash flows over time is equal or higher than reasonable costs, customers will pay no more than necessary to attract investments into the grid companies in an efficient manner (Joskow, 2007). In 1962, Averch & Johnson analytically derived that a regulated firm under cost of service contract does not minimize costs. Its input proportions are distorted. For the equilibrium level of output, the regulated firm uses too much capital relative to labor or capital us- 17