NEGOTIATED THIRD-PARTY ACCESS IN THE GERMAN ELECTRICITY SUPPLY INDUSTRY
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
NEGOTIATED THIRD-PARTY ACCESS IN THE GERMAN ELECTRICITY SUPPLY INDUSTRY Gert Brunekreeft1 University of Freiburg, Germany September 2001 forthcoming: Economia Della Fonti Di Energia E Dell'Ambiente Corresponding address Institut für Verkehrswissenschaft University of Freiburg D-79085 Freiburg i.Br. Germany Tel: (0049)-(0)761-2032373 Fax: (0049)-(0)761-2032372 Email: brunekre@vwl.uni-freiburg.de URL: http://www.vwl.uni-freiburg.de/fakultaet/vw/lehrstuhl.html 1. Introduction Germany opted for negotiated third-party access (nTPA) as the institutional frame for the electricity supply industry (ESI). With this option Germany is the exception, because all other EU member states opted for regulated third-party access (rTPA). The German option has attracted both practical and academic attention and above all criticism, from inside and outside the country.2 In Spring of 2001, the European Commission (2001a), presented a proposal to modify the EU-electricity directive. One important modification was indeed to omit nTPA as an option and instead concentrate fully on rTPA. This would have implied for Germany to modify the sector-specific energy act, such that some agency would be authorized to set or approve ex-ante network-access charges. Curiously, in a background document, the European Commission (2001b) assesses the market developments in Germany as positive, but does not hesitate to criticize nTPA. In contrast to general expectations, the developments have been surprisingly good at first sight and only closer examination reveals that the system does have its flaws; it is only recently that developments have turned negative. This paper will attempt to characterize what negotiated third-party access means and how it works out in the German ESI. Saying that there is no sector-specific ex-ante regulator does not mean that there is no control. Negotiated TPA implies that the antitrust issues in the sector are left to the cartel agency, whose control relies on the competition act. An important policy turning point came in April 2001 with a review of network access in the German ESI, 1 The author would like to thank the editor for useful remarks. 2 The interested reader may be referred to Bergman, Brunekreeft, Doyle et. al. (1999) for a general overview and discussion of liberalization of European power markets.
undertaken by a task force of various cartel offices (Bundeskartellamt, 2001).3 This review explicitly distinguishes between the level of the network-access charges on the one hand and discrimination against third parties on the other hand. The review recognizes that the network- access charges may be excessively high and explores the possibilities the competition act provides to counter this. Notably, the cartel office, by authorization of the competition act, is not allowed to set or approve the network-access charges ex ante; it can only intervene ex post. The paper is organized as follows. Section 2 will characterize the institutional frame. The phrase "negotiated" means in practice that the industry worked out a framework for network access, called the Association Agreement. Currently into force is the second version, which will be described in some detail. The German ESI will be presented in section 3, whereas the market developments will be characterized and assessed in section 4. Section 5 concludes. 2. Negotiated TPA 2.1. The institutional frame The EU-electricity directive (96/92/EG) has been implemented into national law with the Energy Act entering into force 29th April 1998.4 The energy act did not introduce fundamental changes compared to previous legislation. The ESI in Germany has never been monopolized in a strict sense (i.e. close to only one firm), as in e.g. England and Wales, Italy or France. Monopolization in an alternative sense has been achieved by allowing cartelization among various firms (pre-liberalization, about 1,000 asymmetrical firms). Cartels are prohibited by clause 1 of the competition act,5 but the very same competition act made several exemptions to the general prohibition, among which for the ESI. In practice, this worked out into a comprehensive system of so-called demarcation contracts, which were binding and enforceable agreements not to compete in each other's service areas. Cartels were thus stabilized artificially. In the early proposals for a new energy act, the government thought it should suffice to skip the exemption on the prohibition of cartels; hence, demarcation contracts would be prohibited and thus competition would result from the potentially competitive market. True as this may be for non-vertically related markets, in an ESI it neglects the monopolistic networks. Competition in the ESI, however, was thought to be an intra-industry affair, which would concern small, domestic end-users only indirectly. The government held strong beliefs in the countervailing powers in the industry, which might 3 It may be recalled that Germany is a federation, where, apart from the federal cartel office, each state has its own cartel office. Unless stated otherwise, the federal cartel office will be taken to represent the state cartel offices. 4 Energiewirtschaftsgesetz (EnWG); BGBl. 1998, I, 23. 5 Gesetz gegen Wettbewerbsbeschränkungen (GWB). 2
arrange things among themselves, while the government would take care of protecting the small end-users. End-user charges for small (domestic and commercial) end-users have for long been capped by a federal decree,6 which authorizes state-ministries to approve price ceilings (or, more precisely, to check whether price increases are cost based). Paradoxically, this decree is still in force, although it is generally considered to be non-binding. In fact, various political levels have unsuccessfully attempted to abolish the decree. For remaining antitrust issues (like network access), the cartel office was thought to be sufficiently powerful to handle these. This background sets the frame for the energy act 1998. Institutional changes to the previous situation remained modest indeed, with one main exception: network access. Parliament could not accept that a TPA-provision would not be taken up into the law and simultaneously, the cartel office emphasized that the competition act as it was, would not be sufficiently strong to handle network-access issues. As a result, a TPA provision was taken up into the energy act (clause 6), explicitly stating that network-access conditions should be negotiated within the industry, and more or less simultaneously the competition act was modified to include an essential-facilities doctrine (clause 19(4)4, GWB, 6th version). The latter has been carefully formulated to ensure that it is only applied to physical infrastructure; the target of the essential-facilities doctrine was the ESI (and in the meantime, the gas sector), and not non- physical infrastructure, say e.g., a newspaper distribution network.7 The essential-facilities doctrine has two important elements: first, access to the essential facility should be granted in a non-discriminatory manner and second, against a fair and reasonable charge. This one clause has become the main regulatory instrument; the review of network access by the cartel offices, mentioned above, focusses on precisely this clause with these two aspects. During the first three years after liberalization, application of the essential-facilities doctrine concentrated on the aspect of non-discrimination (e.g. the BEWAG-case). As of Spring 2001, however, the focus has shifted towards the level of the network-access charges. The cartel office already threatened to intervene in what it suspected to be excessively high network-access charges; in response, the distribution network operator, E.Dis (which is a 100% subsidiary of E.On), "voluntarily" lowered its charges. One more aspect is important; the review explicitly mentions that the competition act does not allow the cartel office to set or approve network- access charge ex ante; it is only authorized to intervene if a justifiable suspicion of an abuse of market power can be shown, which is ex post by definition. 6 Bundestarifordnung für Elektrizität (BTOElt), 1990. 7 This refers to the Bronner-case settled by the European Court of Justice on 26. November 1998; EuGH Rs. C- 7/97 (Bronner), I-7791. 3
TPA is not fully enforced in the former East. After reunification in 1990, the East German generator VEAG was created which largely owns the lignite generation capacity and the East German transmission network. In order to secure the production of indigenous lignite, VEAG (and thereby its shareholders) had been obliged to modernize the lignite-fueled capacity by investing a total sum of 20 billion DM until 2002. With art. 4 clause 3, the energy act explicitly states that in the assessment whether refusal of TPA is deemed discriminatory, the share of lignite in the production should be carefully taken into account.8 The legal basis is that competition might unduely endanger the obligation to invest and thereby put a vast number of jobs in the East German lignite mines at risk.9 Although the interpretation differs among lawyers (cf. Theobald & Zenke, 2000), it is nevertheless clear that this clause allows VEAG opportunities to refuse TPA. In contrast to most other EU member states, Germany opted for complete eligibility as from the start; at liberalization in April 1998 only the Scandinavian countries knew 100% eligibility. As before, the previous frame appears to have been decisive. End-users were "captives" by the demarcation contracts; i.e. by the agreements to stay out of each other's service areas. In lack of another provision which might have closed the service areas, the prohibition of the demarcation contracts thus made the end-users eligible de jure. If the government had wished to close the service areas (i.e. constrain eligibility) up to, say, the threshold values in the EU-directive, it would have had to include a corresponding provision into the energy act explicitly. Since the government did not believe that retail competition for small end-users would occur anyhow, it seemed rather redundant to take such steps. Nevertheless, under political pressure from the states and communities, a provision was taken up offering distributors were left the possibility to opt for a modified single buyer; no distributor actually applied the provision. Minimal modification of the previous situation also implied that the industry structure remained unchanged. Horizontally, this is largely justifiable. Concentration in generation was fairly low to European standards (cf. section 3), and concentration in retail was considered irrelevant. Large industrial users, or power brokers do not need a retail stage and competition for small end-users was not supposed to occur. More disputable is the unchanged vertical industry structure. The degree of vertical integration was high and has been increasing in the recent past. More importantly, two blocks of firms can be identified. A small number of 8 This clause expires at the end of 2003 with an option to be prolonged to 2005. The European Commission authorized the clause with reference to article 24 of the Electricity Directive (cf. EC, 11.12.1999, Offical Journal L-319/18). 9 It did not quite work out this way; modernization reduced the number of miners from 96,000 in 1989 to 20,000+ in 1994 (Hansen, 1996, p. 556). 4
supra-regional firms (the so-called Verbundunternehmen), which are heavily integrated between generation and transmission, and a large number of communal distributors, which are vertically integrated in distribution (i.e. the network) and retail. Thus, whereas vertical integration among the competitive stages (generation and retail) on the one hand and the monopolistic stages (transmission and distribution networks) on the other hand is acceptable, vertical integration among monopolistic and competitive stages is at least problematic. Structural separation of the (transmission) networks would have required expropriation, which was considered to be contrary to the constitution and thus not a feasible option. This may be so, but it is also clear that the government never considered the vertical industry structure to be a serious problem. The EU requirements concerning vertical unbundling (accounting separation) have been implemented only minimally and have not been seriously applied up until this day. Recently, parliament inquired the federal government about the lack of enforcement of the provisions on accounting separation; in response, the government shifted responsibility to the state ministries.10 The possibility of having a mandatory pool (similar to the former electricity pool in England and Wales) has been discussed, but both the industry and the government refuted the idea. Hence, bilateral contracting is the main trading mechanism, whereas simultaneously private and voluntary initiatives for spot and future markets were allowed and promoted. In the meantime, two spot markets have been established: LPX in Leipzig and EEX in Frankfurt. Irrespective of academic controversy, recent developments in England and Wales (UK) and California (USA) appear to support the government's decision for bilateral contracts and voluntary spot trading. 2.2. The Association Agreement II Negotiated TPA implies that there is no sector-specific regulator and consequently no ex-ante regulation. Issues concerning network access, be it charges, be it other conditions, are left to the industry, whereas the cartel office restrains abuse of market power. The industry associations (i.e. the ESI, co-generators and large industrial users) have been required to work out a framework for network-access conditions; this resulted in the so-called Association Agreement (VV).11 After a false start with VV I, a significantly improved version entered into force in December 1999 with VV II, while momentarily VV II+ is in preparation, which is expected to come into effect at the beginning of 2002.12 10 Cf. BT-Drs. 14/5519 and 14/5733. 11 In German: Verbändevereinbarung. 12 At the moment of writing, unfortunately, no details are public yet. The name, VV II+, suggests that it will only update the VV II, without significant changes. 5
The first version of the association agreement, VV I, mirrors the attitude and expectations the industry had towards the competitive potential of the power market; i.e. a severe underestimation. It might be claimed alternatively, that the expectations were correct, but that VV I was simply an attempt to hold off competition. As stressed above, competition was thought to be an intra-industry affair. Some large industrial end-users and some distributors/retailers would probably renegotiate their comprehensive, long-term contracts and possibly even switch supplier. Within this logic, TPA meant making arrangements for a small number of long-term contracts, containing large amounts of bulk power. Why set up a comprehensive and expensive system, if the expected small number of requests for TPA can be dealt with on a case-by-case basis? There is a flaw to this line of argument; the number of requests was neither small, nor did it concern only long-term contracts. Especially traders (like Enron, TXU), which were not directly represented in the association agreement, had severe difficulties with the system and pushed hard to break it. The main flaw in the VV I, in accordance with the logic explained above, was that it relied upon the contract-path principle. Each trading contract had to be backed up by a request for TPA. The network-access charge was basically a two-part tariff. First, a load-based fixed charge calculated as a mean of the charges at the feed-in and extraction point. The load, for which the charge was calculated, was contract-based, not connection-based. Second, a distance-related (km) charge, for which the distance was calculated as a straight line between feed-in and extraction point. The load and the points of feed-in and extraction, and thus the distance, had to be specified in the contract; hence, the contract-path principle. Because the network-access charges relied heavily on the specifications of the trading contract, the system turned out to be cumbersome, intransparent and leaving options for the transmission system operator (TSO) to manipulate the charges. There were a number of other difficulties. First, the distance-related component clearly advantages nearby power plants, which is an impediment to competition. Normally, the power plants in a control area belong to one and the same firm (vertically integrated generation and transmission). End users in this control area will thus find it cheaper to purchase power from this firm, because it saves on the distance-related charge. Moreover, it lacks an economic justification. Second, strong reliance on contracted load in kW (rather than energy in kWh), creates a bias towards long-term contracts. Consider the following two options. In option one, a trader could have a contract with a maximum load of 100 MW valid for the entire year. In option two, the trader could have a 100 MW-contract for six months with supplier A and have another 100 MW-contract for the subsequent six months with supplier B. In the first option, the trader would have to pay the load-based charge only once, while twice in the second 6
option. In both options, however, the amount of energy actually traded might be the same. The problem with this is not so much the price structure in itself, but that it impedes the development of spot markets, which by definition rely on short-term transactions. For every transaction on the spot market, the trader would have had to pay the load-based charge. Spot markets, residual quantities, reserves and imbalances, have turned out to be essential for fine- tuning competitive power markets. Third, for outside observers (e.g. the cartel office) it is close to impossible to assess the cost-relatedness of the level of the access charges, since it depends strongly on individual contracts. It is possible to calculate examples (cf. e.g. Brattle Group, 1998), but do these examples relate accurately to the real world? Only the TSO has the overall picture. Driven by criticism and market forces the system broke down and was replaced by the VV II in December 1999. The VV II is a significant improvement and appears well-suited for a competitive market. It relates to what appears to be the European standard. Reliance on the contract-path principle has been omitted. Instead, the network-access charges rely on network connection in the form of postage stamps. Only end-users and generators pay, according to load and energy at the point of connection. Still, however, the generator/consumer-split is set at 0/100; in other words, the generators do not (yet) contribute. The costs of higher-voltage networks are cascaded to lower-voltage networks and finally to the end-users; i.e. a lower- voltage network pays a connection charge, also based on load and energy, to the higher- voltage network. The revenue allocation on the extra-high voltage level is negotiated among the network operators, which implies that pancaking at this level has been avoided. Obviously, the voltage level of the point of connection of an end-user is important; an end- user connected at medium voltage does not pay for the low-voltage network. An examination of published tariffs (see below) reveals that network-connection level is the main revenue driver, which seems to correspond to underlying costs. The VV II sets out the structure of the access charges in terms of maximum load and a coincidence factor, which is calculated using load duration, which in turn relies on annual energy consumption. In short, after recalculation, the access charges are presented as a two- part tariff of which one part depends on maximum load (kW) and the other part on annual energy (kWh). In practice, it turns out that the structure only modestly differentiates between different user groups, whereas one would expect that the large fixed costs of networks would allow more pronounced price discrimination between user groups. The network-access charges in e.g. the UK,13 but also the end-user prices in Germany are far more differentiated. 13 In the UK, the fixed charge is a daily-standing charge which does not (or only indirectly) depend on load. In the German system, the "fixed" charge increases with load and thereby modifies the degree of differentiation. 7
Moreover, there is no explicit factor for peak-load pricing, which is remarkable because end- user prices do usually account for peak loads. As figure 3 further below illustrates this implies that larger users are faced with a higher share of network-access charges in the end-user prices, which seems to be the wrong way around. Supposedly, this flaw in the price structure will be repaired in VV II+. There are no separate charges for energy losses. The energy losses are simply allocated to the other network costs and passed through into the access charges irrespective of causation. More efficient would be a cost-based system, but this requires something like nodal spot pricing, real or virtual. Momentarily there is no reliable spot price to base the system on, and installing a virtual system implies high transaction costs which may not compensate the cost- savings (in terms of energy losses or postponed investment). Furthermore, there are no spatial investment signals. In principle, the exact point of connection affects neither the generators nor the end users, except that there are different network owners which may charge different prices. Momentarily these points of criticism are of little concern. As mentioned, the generators do not pay (yet), and thus all price signals, be it short run (energy losses) or long run (investment) would only affect end-users, which does not seem to be very effective. Nevertheless, as soon as generators start contributing, these minor drawbacks are a point of attention. Initially, the VV II included what was called a T-component. The extra-high-voltage networks within Germany had been subdivided in a north-zone and a south-zone. If trading took place from one zone to the other, a charge of 0.25 Pf/kWh was levied. The T-component has been omitted as a requirement in the recent merger cases (VIAG & VEBA into E.On and RWE & VEW into RWE).14 The merging firms must have known that they would have to sacrifice something to get the mergers authorized; throwing out the T-component is one "sacrifice". In fact, it might be conjectured that it was taken up to be able to sacrifice it in the first place. To summarize the assessment of the VV II: • The VV II is transparent; the network-access charges rely on data which concern the end- users and not the supplier. A trader is not involved in the network-access charges. A trader only pays for imbalances. • The structure is no longer biased against short-term transactions, which allows the development of spot markets. 14 The other firms agreed, although they were not (directly) involved in the mergers; consequently, the T- component was abolished for the entire country. 8
• On paper, there are no obvious points which appear discriminatory. Remaining issues concerning tariff structure (generator/consumer split, two-part pricing, cost-allocation) do have efficiency consequences, but are not discriminatory against third parties. • There are ineffiencies in the price structure. In particular, the lack of an explicit peak-load component seems an issue for improvement. Explicit and spatially differentiated charges to set proper investment signals and handle energy losses also qualify for improvement. • Lastly it should be emphasized that the VV II only concerns the structure of the network- access charges; the level is determined by each individual network operator. 3. The sector Since liberalization, the sector has seen significant changes. Mergers and take-overs, both horizontally and vertically, have characterized the latest developments, increasing concentration and reducing the number of firms. New firms and new institutions further changed the electricity landscape. Although politics does not openly interfere in the market structure of the ESI, there is the impression that the envisioned structure is an oligopoly with four vertically integrated firms (i.e. RWE, E.On, EnBW and VEAG). On the level of generation and thereby transmission this process is well underway. In distribution and retail, however, there are still many firms. Even if mergers and looser forms of cooperation occur at these levels, the resulting firms may still be largely independent of the four supra-regional Verbundunternehmen (VUs). Until recently, there used to be nine supra-regional VUs, which were heavily integrated between generation and transmission. After the merger of the two smaller southern VUs into EnBW in 1996, the main cases, which attracted much attention, were the mergers between RWE and VEW forming RWE, and VEBA (i.e. PreussenElektra) and VIAG (i.e. Bayernwerk) forming E.On, which reduced the number of VUs to six. These two mergers had to be authorized by the cartel agencies. The former fell under the jurisdiction of the German federal cartel office and the latter under the jurisdiction of the DG-Competition of the European Commission. In close cooperation, both mergers were provisionally authorized. One provision, as mentioned above, was that the T-component was omitted from the VV II. Another provision was that the merging parties had to sell the shares they held in VEAG, which in sum was over 75% of the VEAG shares. VEAG was created as a new firm directly after the German reunification with the former East in 1990. It controls most of the generation and transmission capacity in the former East. The Western VUs have been heavily involved in providing the necessary capital and know-how to modernize the system in the East. In 1995, VEAG was privatized and the shares divided among the western VUs. The three largest firms at that time (RWE, VEBA and VIAG) each held 25% share and a holding of the other VUs 9
(thus including VEW) received the remaining 25%. The process is now more or less completed. Rather than floating the shares on the stock exchange, the shares have been sold to a consortium of Sweden's Vattenfall and the US-firm Mirant (formerly Southern Energy). Another substantial part of the shares of VEAG belongs to HEW, but over 70% of the latter's shares in turn are held by Vattenfall. Finally, Vattenfall and Mirant are also majority owners of the Berlin-based BEWAG. It thus appears appropriate to consider the group of VEAG, HEW and BEWAG as only one firm which may, for the time being, be called VEAG+. The fourth VU is EnBW, which is largely independent of the rest. EnBW is relatively small, but has powerful back-up, since EDF from France was able to purchase over 25% of the shares. Table 1 provides the respective market shares in generation, pre- and post mergers. TWh Pre-merger (%) Post-merger (%) VEBA 89.93 18.77 VIAG } E.ON 47.74 9.97 } 28.74 RWE 138.63 28.94 VEW } RWE 39.89 8.33 } 37.27 EnBW 41.21 8.60 8.60 HEW 12.31 2.57 BEWAG VEAG } VEAG+ 10.21 49.50 2.13 10.33 } 15.03 Other 49.60 10.36 10.36 Total 479.0 100.00 100.00 Table 1: Market shares in production. Source: Based upon Europäische Kommission, 2000, p. 23 and own calculation. Table 2 recalculates these values into generation concentration ratios (both using TWh). The latter are for reasons of illustration compared to the values from the UK. HHI CR1 CR2 CR3 CR4 Germany; pre-mergers ~1560 0.289 0.477 0.581 0.680 Germany; post-mergers ~2525 0.373 0.660 0.810 0.896 UK (1990/91) ~3225 0.455 0.739 0.913 0.983 UK (1998/99) ~1620 0.210 0.420 0.597 0.768 Table 2: Concentration ratios in generation output. Source: Calculations based on Brunekreeft & Keller (2000). Note: Pre- and post-merger refers to the RWE-VEW and VEBA-VIAG mergers and the sale of VEAG. Clearly, the German post-merger concentration ratios are high and are now roughly comparable with the concentration ratios in the UK directly after liberalization in 1990. Most noticeable is the direction of change. The UK had significant problems with a concentrated generation market and the authorities have attempted to decrease the concentration; in Germany concentration increases. In the UK the generators were vertically separated from the monopolistic networks and were consequently forced to attempt to make their profit in the wholesale power market. Since they were quite successful, the wholesale prices were rather 10
high (cf. e.g. Newbery, 1995). The institutional frame in Germany is different. The generators are to a large extent vertically integrated with the monopolistic networks and are the network- access charges are unregulated. It can be argued that the integrated firms' strategy is to set low margins in the competitive markets (generation and retail), and make excess profits on the networks (cf. Brunekreeft, 2001a) and thus, given the current institutional frame of unregulated network-access charges, the relatively high generation concentration ratio does not imply relatively high wholesale prices. Indeed, the wholesale prices (on the power exchanges) are low; possibly as low as (short-run) marginal costs. This claim depends critically on the fact that the network-access charges are unregulated. If the network-access charges are constrained in some way or other, it is likely that the high concentration ratio may turn out to be problematic. As has been the case in so many other countries, it appears that a balance between different policy aims has been struck. Retaining "sufficient" competition on the one hand and allowing the creation of "global players" on the other hand. It is not entirely clear to what extent these VUs are integrated into distribution and retail. Directly or indirectly, they do own distribution networks with their host retail departments, partly in their own respective control area, partly via participation in holdings somewhere else in the country. Moreover, with the exception of the original VEAG, they have all set up their retail department aiming for domestic end-users nationwide. Especially, EnBW with its retail subsidiary Yello is quite active in this segment. Apart from the VUs, some 700 mainly communal firms determine the distribution and retail stage, largely vertically integrated. Still, these communal firms are dominant host-retailers in their distribution network area, because domestic switching rates are very low. The number of communal firms rapidly declines. The scale of many of these firms is too small to sustain in the competitive market and in the monopolistic distribution-network business it has become fashionable to exploit synergy effects. A more rational business attitude drives the firms towards scale enhancement, merger, privatization and outsourcing. The generation mix in Germany is given in table 3. It shows that the total of somewhat over 450 TWh (with little below 100 GW installed capacity) mainly relies on nuclear power and coal (hard coal and lignite). Dependence on oil, gas and hydro is low; other fuels are mainly CHP and increasingly wind energy. The main, although still minor change since 1993, is that the use of gas increased at the expense of lignite. 11
1993 1999 Hydro 4.49 4.61 Nuclear 34.34 34.81 Lignite 29.52 26.02 Coal 26.31 25.26 Oil 0.84 0.32 Gas 3.94 7.43 Other 0.57 1.54 Table 3: Generation mix (based on production). Source: VDEW (1998) & VDEW (2000). Note: Figures in %.. In Summer 2000, the government decided to phase out nuclear power over a period of 32 years; in effect, this means that existing nuclear capacity runs until depreciated, and that no new nuclear capacity may be built (except that which was already under construction). In the face of excess capacity, this was not planned anyhow. The phasing out of nuclear capacity is largely the political will of the Green Party, which is the minority party in the current government. Given the European Commission's renewed interest in nuclear power, it might be questioned whether the "phasing out" of nuclear power is in fact not just a moratorium. Consistent with the European Commission's policy the German government sets on a larger share of renewables. The renewables are promoted with the so-called renewable-energies act, which basically arranges a take-off-obligation for the network operator to which the renewables capacity is connected, and a fixed feed-in price against which the renewable power should be compensated.15 The costs are passed through, via the retailers, to the end- users as a levy on the power price. There are compensation arrangements to spread the burden evenly over the country. The generation mix shows the large share of coal, which in sum makes up for approximately half the electricity generation. Coal is the country's only primary energy source. It has thus been policy to promote the use of inland coal (partly in order to decrease reliance on outside primary energy and partly, in order to save the jobs of miners), which resulted in coal subsidies. In a way, these are still in force, but are phased out. A problem of special interest is the use of lignite. As mentioned above, lignite was and still is the main fuel in the former East. For various reasons, it had been thought to be a good idea to continue the policy of using lignite for electricity production after the reunification. The problem, however, was that the existing capacity was completely outdated, both to economic and environmental standards. It was agreed, although quite controversially, that the Western VUs would take over the responsibility of modernizing the Eastern ESI (in the so-called Stromvertrag), which required huge investments in among other things lignite capacity. 15 It varies for different energy sources, but for e.g. wind power it is approximately 17 Pf/kWh. 12
8000 generation 7000 capacity 6000 (million DM) 5000 4000 total 3000 total old total new 2000 1000 0 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 Figure 1a: Annual investment generation capacity in the German ESI. Source: Based upon Karl (1996). Note: "Old" means the old states (former west) and "new" means the new states (former east). The modernization of the ESI in the former East is interesting; it is in fact the main difference between the developments in the German ESI and other European states. Figures 1a and 1b show that investment, especially in generation, has been exceptionally high. In 1995, investment in generation capacity in the former East was higher in absolute terms than in the West, whereas electricity consumption in the East is less than 15% of national consumption. Over a period of less than five years about 20 billion DM has been invested in the ESI in the former East alone, mainly financed by the Western ESI. The impression emerges that this may contribute to an explanation to the question as to why the German ESI is not regulated, in contrast to all other European member states. 9000 transmission + 8000 distribution 7000 (million DM) 6000 5000 total 4000 total old 3000 total new 2000 1000 0 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 Figure 1b: Annual investment in transmission and distribution in the German ESI. Source: Based upon Karl (1996). Note: "Old" means the old states (former west) and "new" means the new states (former east). 13
4. Developments 4.1. End-user prices The developments since liberalization have been quite spectacular. As expected, switching of large industrial users and communal distributors has set in immediately after liberalization. This quite often did not result in actual switching, but the threat to switch sufficed to renegotiate the contract resulting in significantly lower prices.16 Hence, industrial prices (including industrial network-access charges) fell significantly. Fully unexpected was the development of retail competition aimed at the large group of small, domestic end-users. It may be recalled that Germany opted for 100% eligibility directly with liberalization. Around September 1999, the improved second version of the association agreement (VV II) was presented, which eased network access considerably. More or less simultaneously the larger VUs launched their retail-subsidiaries attacking the domestic-users market (dominated by the incumbent communal host-retailers) with nationwide offers, supported by large-scale marketing campaigns. The communal retailers responded swiftly by lowering their prices. The price development for an average domestic end-users (3500 kWh annually; eurostat's category Dc) is plotted in figure 2. Prices exclude VAT, but include the electricity tax. The latter was introduced in April 1999 with 2.0 Pf/kWh, raised January 2000 to 2.5 Pf/kWh and again raised January 2001 to 3.0 Pf/kWh. The price developments have been divided into the incumbents' and the entrants' prices; to recall, the incumbents are communal host retailers, whereas the entrants are the new retail subsidiaries of the VUs and really new retail entrants. Both prices are calculated averages. The incumbents' prices is a regional average, because these do not have nationwide offers. The entrants' price is an unweighted average of the ten cheapest offers, representing a best- practice benchmark. The picture is telling. It indicates that the first round of the electricity tax has been passed through completely, which is the peak in August 1999. Around this time, the retail competition has set in, reducing the price level(s) significantly. The second round of the electricity tax may have slowed down the price fall, but could not be passed through. The lowest price level was reached around Spring 2000; by then the best-practice benchmark was over 20% lower than the original price level less than one year before. Figure 2 also indicates that the incumbents reacted immediately and simply followed further price decreases at a distance. The distance reflects consumer "stickiness". Various market surveys reveal that consumers have been aware of the possibility and are principally willing to switch, but 16 Not switching had the considerable advantage of avoiding the network-access difficulties of the first version of the association agreement. 14
required a premium to be convinced. The (averaged) distance between incumbents' and entrants' prices peaked at about 15%. As can be seen, the price level is rising again since late 2000. This has at least two reasons. First, January 2001 again witnessed an increase of the electricity tax by another 0.5 Pf/kWh, which is passed through. In the same line, a levy for stranded CHP is being passed through. Second, especially independent new retailers encouter financial problems. The profit margin which remains after subtracting the network-access charges apparently is too low to be profitable (see further below), while the actual switching rate has been modest as well. According to the ESI's association VDEW, the switching rate by the end of 2000 was about 2% for domestic and 3.5% for commercial. Quite possibly the latter is the result of the swift reaction of the incumbents. The new retailers respond to the financial difficulties by either increasing their price level or leaving the market. By mechanism, both reactions increase the average entrants' price level. 28 Aug.99 27 Incumbents Oct.99 Aug. 01 26 Apr.00 Jan.00 25 Aug.00 Aug. 01 Pf/kWh 24 April 1999: electricity-tax: 2 Pf/kWh Since January 2000: 2.5 Pf/kWh Jan.01 23 Average of Since January 2001: 3 Pf/kWh Nov.99 nationwide retailers 22 Jan. 00 Aug.00 Apr.00 21 Jul Jan Jul Jan Jul Jan Jul Jan Jul Jan Jul Jan Jul Jan Jul Jan Jul 93 94 94 95 95 96 96 97 97 98 98 99 99 00 00 01 01 Figure 2: Price development for an average domestic end-user. Source: Based upon Brunekreeft & Keller (2000). The developments deep into 2000 have been considered a success story. There has been a lot of attention in the media and the successes in telecommunications seemed to continue with electric power. As a by-effect, the arguments in favour of an ex-ante sector-specific regulator 15
appeared somewhat academic.17 The recent developments (e.g. low switching, rising prices and financial problems of entrants) might by the same token change regulatory attitude. As mentioned in the introduction, in April 2001 the cartel office assessed the overall situation as problematic and indicated to pursue a pro-competitive policy focussing on the network-access charges (Bundeskartellamt, 2001). The high level of the latter, relative to the end-user prices, is indeed the problem. 4.2. Network-access charges Table 4 gives an overview of published network-access charges, averaged for a sample of network operators.18 The overview has been restricted to end-users connected to the low- voltage networks. The users are qualified according to the Eurostat-method (cf. e.g. Eurostat, 2000a and 2000b), where "D" stands for domestic and "I" stands for industrial, which includes commercial. The second letter indicates the customer's size. Pf/kWh Ann. Cons. (kWh) Load End-user price Access charge Ratio Level Total at night (kW) Da LV 600 3 39.48 13.06 0.391 Db LV 1200 4 31.51 12.32 0.484 Dc LV 3500 1300 7 25.71 11.83 0.576 Dd LV 7500 2500 8 23.65 11.69 0.636 De LV 20000 15000 9 14.57 11.62 1.130 Ann. Cons. Max. load Ann. Util. (hrs) (103 kWh) (kW) Ia LV 30 30 1000 27.89 11.60 0.427 Ib LV 50 50 1000 27.26 11.59 0.437 Table 4: Network-access charges. Source: Brunekreeft & Keller (2000). Note: End-user price incl. taxes; Ratios are calculated for end-user prices excl. taxes. The end-user prices include two taxes. First, a concession fee, for which ceiling values are laid down in a federal decree. The level of the concession fee depends in particular on the city's size. Here an average has been estimated of 2.7 Pf/kWh for domestic and 0.22 Pf/kWh for industrial/commercial users. Second, the electricity tax, which now is 3.0 Pf/kWh for domestic and 0.6 Pf/kWh for commercial/industrial users. The last column in table 4 gives the ratio of the network-access charge in the end-user price, for which the latter has been cleaned of taxes. The reader may notice immediately that the overall level of access charges is high - both in absolute terms and in terms of ratio. It should be realized, moreover, that the domestic categories Dc and Dd are by far the most important empirically. The overview shows that the degree of differentiation, in contrast to the end-user prices is low. Because the end-user prices 17 The interested reader may be referred to Brunekreeft (2001b) for a more extensive and more formal analysis of this argument. 18 Although they must be published, many distribution-network operators still fail to do so. 16
differ with customer size, whereas the network-access charges differ only modestly, the ratio varies significantly. Note in particular the category De, for which the ratio is above one; in other words, the network-access charges are higher than the end-user prices (cleaned of taxes). These users rely, by Eurostat's assumption, heavily on off-peak power for heating purposes. As mentioned above, the end-user prices in Germany do have a peak-load element, whereas the network-access charges do not. Ib Netherlands Germany Ia UK Dd Dc Db Da 0,000 0,100 0,200 0,300 0,400 0,500 0,600 0,700 Figure 3: Ratio of network access charges to end-user prices. Source: Based upon Brunekreeft (2001a). Figure 3 plots the ratio of network-access charges in the (cleaned) end-user prices and makes a comparison with the UK and the Netherlands. Both the latter apply price-cap regulation for the network charges; in the UK for about ten years now, while in the Netherlands regulation has started since the beginning of 2001. For the Netherlands, a tough price cap with an average X of 5.9% annually, has not yet been taken into account in these figures. Two aspects are strikingly clear. First, the relation between customer size and the ratio are reversed between Germany on the one hand and the UK and the Netherlands on the other hand. The reason is that the structure of the German access charge is not sufficiently differentiated. Second, for the empirically relevant groups Dc, Dd; Ia and Ib the network-access charges in Germany are high (both in absolute and relative terms). The latter is the problem for new entrants and is now also the focus of the cartel office. For an end-user it is not important how exactly the final price is composed; for a competitor it does matter, because the difference between the end-user price and the network-access charges (subtracting taxes) determines the remaining profit margin. The latter is relatively small in Germany. 17
4.3. Third-party discrimination The first three years of liberalization have been characterized by many complaints about discriminatory behaviour (for court, cartel authorities and in the media). As argued above, the real problem for competitors is the low profit margin, which is induced by the high network- access charges, which in turn are a result of the integrated network operators' incentives. Discriminatory behaviour is -given the institutional frame- not the real problem. Even so, where it occurs, it should be prohibited. The network-access review of the task force of the cartel offices (Bundeskartellamt, 2001) explores discriminatory behaviour with a list of eight issues, which play a role in practice; some are not considered to pose a real problem, while others are problems which can be cleared for court. This section will provide a brief discussion of discriminatory behaviour. Refusal to provide TPA TPA should be granted on non-discriminatory terms, be it by the TPA provision in the energy act or by the essential-facilities doctrine in the competition act. However, there are legal exceptions and subtle ways to block TPA. With respect to the latter, firms are reported to make negotiations unreasonably difficult or cumbersome, demanding unnecessary formalities. Making a serious offer can be unduely delayed, possibly at the expense of the competitor. A related issue is that TPA is refused because the legal status of the (long-term) contract of the switching end-user is not cleared. Cases have been settled for court, where the incumbent may have an argument, but which nevertheless hinder the development of competition. Apart from real cases, the incumbent may simply hide behind a legal argument and wait until the competitor goes to court. Exceptions on providing TPA are legally given by capacity constraints and by the lignite clause. In one case, Berlin-based BEWAG justified refusal of TPA to its competitors (among which RWE) with the argument of constrained capacity. The case, which was to be cleared by the federal cartel office,19 was rather exceptional and complex, because it concerned the interconnector between the former East and West Berlin taken into operation in 1994. The cartel office made unambiguously clear that all capacity has to be allocated on equal terms to all those wanting access and that priority rules must be applied with utmost restriction. In cases without capacity constraints there is no obvious problem; this is important because the cartel office explicitly pointed out that if a consumer switches, required line capacity may simply go from one hand to another without requiring additional capacity. The incumbent cannot claim the line capacity "left behind" by the switching consumer, if the consumer needs 19 Cf. Bundeskartellamt, B8-40100-T-99/99; 30th August 1999. 18
it to be supplied by the third party. For an economist this seems rather obvious; for a lawyer, however, this means that property rights are reduced in status. For the case where the interconnector might be congested (peak times), the cartel office ruled that existing capacity should allocated proportionally according to shares in demand. The line capacity was estimated at 400 MW in peak times, whereas the peak load in West Berlin was 1966 MW, and thus each supplier could be allocated 20.3% of the (peak) load of its customers. The lignite clause concerns the former East. As explained above, for socio-political reasons, the production of lignite was thought to require protection against undue competition, resulting in the lignite clause, which states that if an authority assesses refusal of TPA it should carefully take account of the effects on lignite production. In other words, this clause gives VEAG an argument to refuse TPA and it is up to the courts to decide whether refusal is reasonable. What exactly is reasonable in this case is quite controversial, although the courts tend to be pro-competitive. Switching costs Switching costs may be defined here as anything which requires a premium to convince a consumer to switch away from the incumbent. The incumbents have applied a variety of instruments to set or increase switching costs, some of which are unambiguously abusive, while others possibly involve a trade off. An example of the former is refusal of contract's notice on formal grounds; these may have a legal basis, but may also be unreasonable. Another example is plain intimidation of switching customers. At several occasions, incumbents have been reported to suggest that continuity of power supply could no longer be guaranteed after switching, because of e.g. technical reasons or solvency problems of the new supplier. Obviously, the average domestic end-user will not be informed in detail about the technical and legal arrangements concerning continuity of supply. Yet another example is if incumbents match the competitive offer as soon as the consumer announces the switch. This is what potential competition is supposed to do, but in a developing market, it is the incumbent's strategy of informed selection which makes the competitors' lives very hard. Two related issues are more challenging. First, explicit switching costs based especially on additional meter reading involved in the switch. Usually, meter reading is only once a year (normally bundling entire streets). If an end-user switches supplier during the year, it has to be settled how much the end-user consumed up to this point. This implies an additional meter reading at the moment of switching, which is relatively expensive.20 These are real costs and 20 It may be noted, that actual reading may be circumvented, by e.g. applying estimates derived from the load profiles. 19
the incumbents have made a practice of it to pass through these costs to the switching customer (or to its new supplier). For average domestic end-users these costs can be about half the expected savings. The cartel office (Bundeskartellamt, 2001, pp. 49 ff.) goes in pains to argue that such costs can be allocated legally consistent (i.e. cost based) to all end-users (through network-access charges), rather than to the switching customer. The main argument is that such switching costs impede the development of competition, whereas also the non- switching customers would gain from competition by a lower price level overall and thus it can be legally justified that the non-switching customers contribute to these costs. The second issue concerns use-of-system contracts. An end-user demands two separate services: power supply and use-of-system (i.e. network access). This requires two contracts, which can, however, be bundled into one contract. Whereas the new suppliers insist on representing the customer in all issues and insist on offering an all-inclusive contract, the incumbents insist on signing a separate use-of-system contract with the end-user and refuse to accept all-inclusive contracts. The formal underlying argument is the legal question of liability in case of insolvency. The problem is that it applies asymmetrically for switching consumers, which raises perceived switching costs and may thus keep customers from switching. Courts and cartel offices argue in favour of the competitors and are strongly opposed to the incumbents' refusal of accepting all-inclusive contracts. Load profiles To overcome the problems of expensive time-of-use meters, which if required, would stifle retail competition, statistical load profiles (as in Scandinavia and the UK) have been developed and are applied. Two different methods are used: synthetical and analytical load profiles. The network operator decides which method is used in its area. The cartel office review (Bundeskartellamt, 2001, p. 62) reports that 59% used the synthetical method, 14% the analytical method, while 27% of the questioned network operators did not respond. The synthetical method ex ante estimates a load profile for a (group of switching) consumer(s). Possible imbalances between scheduled and real values (although this cannot be known because these consumers are not time-of-use metered) pass on to the network operator. This method is relatively inaccurate, but given its ex-ante nature, is relatively friendly for competitors. The problem is that the anticipated costs for imbalances are passed through ex ante into the network-access charges; it will be difficult to control whether the pass-through relates to the actual costs or is excessively high.21 The analytical method is more accurate (and more expensive to install). It meters the time-of-use load in the network and subtracts the 21 Since the network-access charges are unregulated anyhow, this argument does not seem to be overwhelmingly relevant. 20
load of all time-of-use metered consumers; the residual per definition is the load of the non- metered consumers. By means of synthetical estimates this residual can be further allocated to various user groups. This method thus is ex post, which leaves the competitors with considerable uncertainty. More importantly, the data on network load and aggregated load of all metered users are only known to the network operator. Within reasonable ranges, this provides a large potential to discriminate against third parties.22 Another issue with load profiles is less challenging, but real. Previously it has been customary to install time-of-use meters for consumers with an annual consumption of over 70,000 kWh. The ESI has agreed that in case of switching load profiles are to be used for consumers with annual consumption up to 30,000 kWh. This leaves a market segment of consumers between 30,000 and 70,000 kWh for which a time-of-use meter has to be installed in case of switching. Apparently, in many cases this is commercially not attractive, which sets competitors at a disadvantage. Imbalances A last issue which deserves attention concerns compensation payments for imbalances, which are differences between scheduled and metered values.23 Nera (2000) brings a criticism of a specific point in the institutional arrangements in the German ESI. Ideally, the price to be paid for imbalances is something like a market price, reflecting the system's marginal costs of electrical power (which can be opportunity costs) at any moment. It would not make a difference whether the imbalance is over-extraction or over-feed-in; it would be charged against or compensated with the same price. In contrast, the association agreement II arranges a fixed predetermined price for imbalances. If there would be only one fixed price, valid for both extraction and feed-in, and if this prices is something like an average of fluctuating market prices, this would provide an opportunity to use the imbalance payments for strategic manipulation. If the market price is higher than the imbalance price, a third party would over- extract and sell in the spot market and reverse. Such strategic manipulation would endanger system's operation. Therefore, the association agreement II sets two prices: a feed-in price (2 Pf/kWh) and an extraction price (6 Pf/kWh). Since normally the former is lower and the latter is higher than system's marginal costs (~ 4 Pf/kWh), this avoids strategic manipulation. The problem is that third parties lose money either way and are put at a competitive disadvantage relative to the integrated network operator. It may stressed that this is not genuine discriminatory behaviour; it only works out that way in lack of an institutional alternative. 22 Currently, this is a hypothetical argument. There are no indications of actual abuse of this potential. Given the current unregulated institutional frame this does not surprise, but it does indicate a substantial opportunity to discriminate should regulation of the network-access charges actualize in the future. 23 It is beyond the scope of this paper to be extensive. The interested reader may be referred to nera (2000). 21
Consequently, the cartel office argues in favour of installing markets for imbalances to tackle the problem. Meanwhile, RWE has started an imbalance market in the internet.24 Looking through the list above, one cannot escape the feeling that all in all the extent of genuine discriminatory behaviour is modest. Apart from unsystematic consumer intimidation and paper-wars (which should be pursued with tough means), various issues fall back on real arguments and/or involve trade-offs, which should be or have been settled for court. Moreover, there are strikingly few indications that the real potential to discriminate through imbalance settlement and load profiling is systematically abused. 5. Concluding remarks This paper provides an impression of the developments in the electricity supply industry in Germany since liberalization in April 1998. The developments in Germany deserve close attention, because Germany is the only EU member state which opted for negotiated third party access, whereas the European Commission recently proposed that all member states install a sector-specific regulator with authority to set or approve network-access charges ex ante. Negotiated TPA means in the German practice that there is no sector-specific regulator. The issues concerning network access (including the level of the charges) is left to the cartel office, authorized mainly by the essential-facilities doctrine in the competition act. The essential-facilities doctrine states that access to the networks should be non-discriminatory and against a fair and reasonable charge. The latter aspect is becoming increasingly important, since the cartel office published a review on network access, which emphasizes excessive access charges. A small, but important remark in the same review notes that the cartel office's authority is restricted to ex-post action; in other words, something like a price cap for the network-access charges in the German ESI requires a modification of existing legislation. Another practical implication of negotiated TPA is that the industry associations worked out a general framework for network-access conditions, the so-called association agreement (abbr. with "VV"). After a complete failure with the first version, since December 1999 the second version is in force (VV II). Apart from a set of more technical conditions and information requirements, the VV II sets the frame for the structure of the network-access charges. In contrast, the level of the network-access charges is left to each individual network operator; there are six (although it will be four soon) transmission network operators (largely integrated with generation), and some 700 distribution network operators (integrated with retail). Basically, the VV II conforms to European standards; it can no doubt be improved on several 24 Cf. www.rwenet.com. 22
You can also read