The European Electricity Market – Does Liberalisation Bring

The European Electricity Market Does Liberalisation Bring Cheaper and Greener Electricity?

Claudia KEMFERTa, Wietze LISEb, and Robert STLING c

1. DRAFT October 2003

a Scientific Pool of Environmental Economic Disciplines (SPEED), Oldenburg Univer-

sity, Oldenburg, Germany. b Institute for Environmental Studies, Faculty of Earth and Life Sciences, Vrije Univer-

siteit, Amsterdam, The Netherlands c Stockholm School of Economics, Stockholm, Sweden

Address for correspondence:

Claudia Kemfert

University of Oldenburg

D- 26111 Oldenburg

E-mail: [email protected]

Tel: +49 441 798 4106

Fax: +49 441 798 4101


2

Abstract This paper investigates the impacts of the liberalisation of the European electricity mar-ket. We apply a game theoretic modelling framework in order to assess the economic consequences of different market behaviour. It turns out that, because of different devel-opments of European countries, some countries benefit substantially by strategic market behaviour whereas others could also profit from more competition. The effects on the environment are ambiguous.

Keywords: Electricity market liberalisation, Game theoretic model, Environmental effectiveness, Europe

JELclassification: C7, D2, Q4, R3


3

1. Introduction

The liberalisation of electricity marketsthe introduction of competition, the reduction of external, particularly political interferences and adjustments, and the opening of the market for new providersis a worldwide phenomenon. Although the reasons for an opening of markets vary from country to country, the main goal apart from higher pro-duction efficiency is to offer the customers lower electricity prices.

Though only a few countries in the world have accomplished a complete liberalisation, it can be observed, however, that most countries aim for a complete opening of the elec-tricity market in the near future (see Figure 1). In Europe, all EU member states shall liberalise the electricity market according to the 1997 directive of the European Com-mission (Directive 96/92/EC). The directive provides that European electricity markets should already have been opened up to an average of 25 percent in 1999.

The directive is, however, converted differently into actual policy in different countries, and the progress of liberalisation of electricity markets in Europe vary between coun-tries. In Germany the market was liberalised in 1999, whereby the markets in Norway, Sweden and United Kingdom had already been completely opened up by that time. Aus-tria and Denmark have liberalised their electricity markets almost completely as well (see Table 1). Spain is likewise aiming at an imminent opening of the market. France and Italy have not yet decided when they intend to open the market for external competition. These countries are characterized by the predominant position of a few electricity pro-duces, i.e. a rather uncompetitive monopoly and/or duopoly prevail (France: EdF, Italy: Elettrogen and Enel).

This unequal distribution of market opening and liberalisation of the electricity markets in Europe involve some competition distortions some utilities already face complete competition, whereas others can continue operating in a monopolistic position. Since utilities have to compete with each other after opening of the market, providers, in order to survive, need to alter their behaviour. In Germany, for example, utilities reacted very dynamically after the liberalisation of the electricity market in 1999 by firm mergers and strategic behaviour. A rise of the market shares of certain producers might lead to a rather uncompetitive market structure, which will not reduce, but rather increase electric-ity tariffs.

Whether an electricity supplier is able to convert strategies in the electricity sector de-pends on the market situation, in particular on the dominant market conditions. Thus the market entrance conditions on the different levels of the current market (production, trade (and selling)) play a crucial role.

Furthermore, also electricity trading options can offer additional incentives for the prac-tice of market power, unless uniform price structuring for tradable electricity are created. In Germany for example, a federation agreement regulates the prices for the power trade. However, it has been observed in the past that due to strategic market behaviour, provid-ers for third party access of extraneous electricity were completely delayed or refused. In its second benchmark report the European Commission criticizes that competition distor-tions and market power can arise through high entrance net access fees, which obstruct


4

the entrance of new providers. The different market opening degrees diminishes the ad-vantages for the customer.

Whether the liberalisation of the European electricity markets involves a sustainable, i.e. environmental friendly development, remains to be observed. The current market liber-alisation, and the political decisions in some European countries on nuclear energy and renewable energy law, can lead to increased current production costs, which can involve an increased import of comparatively low-cost electricity. Cheap electricity is not neces-sarily beneficial from an environmental point of view as it is often produced by depreci-ated production plants like nuclear or coal An environmental beneficial situation would increase the share of renewable energy like wind, hydro or biomass technology. Electric-ity from nuclear energy is to be judged positively from a climatic point of view, since it sets free only few greenhouse gases like the main greenhouse gas carbon dioxide. How-ever, nuclear energy poses other dangers and risks to the environment, which usually are not included in the production calculation. As in the current liberalised market, green energy involves higher production costs, and is less competitive. Hence, a harmonisation of environmental laws seems to be necessary in Europe.

Newberry (2002, 2002a, 2002b) studied potentials and opportunities for European utili-ties. Day and Bunn (1999) investigated these aspects by a game theoretic model of mar-ket power and strategic actions of firms in the UK. Bower and Bunn (1999) assess trade opportunities within a pool versus a bilateral trade system in the electricity market of the UK. Admundsen and Bergman (2002) studied these issues for the Norwegian and Swed-ish power market. Here, transmission and transport pricing plays a crucial role.

Experiences in Scandinavia and the UK suggest that a uniform tariff is preferred over distance charges. Moreover, market opportunities and grid owners significantly influence trade. Dawson and Shuttleworth (1997) studied transmission pricing in Norway and Sweden. Green (1997) examined this effect for the UK. Cardell et al (1994) investigated the negative effects of market power and transmission constraints on trading by an im-perfect competition model for North American electricity suppliers.

Different non-cooperative games within various markets have been examined by diverse authors. Murphy et al (1986) demonstrate how mathematical programming approaches can be used to determine oligopolitistic market equilibria. Salant and Shaffer (1999) il-lustrate the theoretical impacts on production and social welfare via two-stage Cournot-Nash equilibrium solutions, where learning by doing and investments in R&D determine marginal costs of identical agents differently. For Europe, Jing-Yuan and Smeers (1999) have modelled an oligopolistic electricity market with a sophisticated game theoretic model, calculating the Nash equilibria. More generally, Helman et al (1999) investigated different kind of trade options and strategic price setting within the electricity market. Stern (1998) investigates the liberalisation of the European gas market.

Bower et al (2001) have simulated the liberalised German electricity market by an agent-based model. They conclude that mergers increase market power, increasing the electric-ity prices. Their model is very sensitive to the out-phasing of expensive oil-fired plants, nuclear energy, or to closing the borders to imports of (cheap) electricity. In all these in-stances, prices jump up considerably. Bigano and Proost (2002) conducted a four coun-try study (France, Germany, Belgium, the Netherlands) which is linked through electric-


5

ity trade. The environmental impacts are quantified, where a 3-stage game is calculated in a partial equilibrium framework. They compare strategic action with perfect competi-tion to conclude that phasing out nuclear energy leads to a substantial decrease in social welfare.

In a liberalised electricity market, electricity suppliers can act strategically, which influ-ence electricity prices, due to changing market shares in favour of large firms. Further-more, mergers can become attractive, as it increases the electricity price and hence prof-its. While enhancing competition in the electricity market, strategic behaviour deter-mines the structure of the market and energy supply network (see also Kemfert, 1999, and Kemfert and Tol, 2000).

In this paper, strategic behaviour of energy suppliers and their impacts on the economic and environmental situation in the liberalised European electricity market is studied with the game theoretic modelling tool EMELIE (Electricity MarkEt Liberalisation In Europe).1 EMELIE is calibrated to the main European energy suppliers, which are linked by capital flows. These firms produce electricity through different technologies, consid-ering others and their own capacities, and variable production costs. Within this context, the analysis focuses on the impacts of strategic action by firms, which is compared to the case without strategic action, i.e. when the market is fully competitive. We investigate the electricity prices and the environmental effectiveness of applied technologies.

This paper is outlined as follows. Section 2 highlights some important aspects of Europes electricity market structure. Section 3 gives a general introduction to the com-putational game theoretic model EMELIE, while the Annex shows a more detailed mathematical formalisation. The required data and the calibration of the model to the European liberalised electricity market are presented in Section 4. Section 5 discusses the main model results for seven European countries, namely Denmark, Finland, France, Germany, the Netherlands, Norway and Sweden. The final section concludes.

2. Europes Electricity Market Structure

As the benchmark report of the European Commission of March 2003 testifies, the European electricity market can be characterised by increased market opening, im-provements of unbundling of net owners and more transparent regulation methods.2 Whereas Italy and United Kingdom registered decreased electricity prices for large con-sumers, also Austria, Germany and the Netherlands observed increasing activities of cus-tomers. However, as not all countries have liberalised their electricity markets com-pletely, some unsolved difficulties still emerge, as the degree of unbundling, increased market shares of dominant utilities in some European countries and the lack of infra-

1 A first version of EMELIE is applied to study economic impacts of the German electricity mar-

ket in Lise et al (2003). First application to the European market is given by Kemfert et al (2002).

2 See European Commission (2003).


6

structure of inter country electricity trade. Considerable electricity price decreases could also be observed in Austria, as Figure 3 demonstrates.

Since the market conditions for the current providers are still very different in the indi-vidual countries in Europe, there is so far no clear tendency to observe in the develop-ment of the electricity market. In Germany the liberalisation of the electricity market first led to sinking electricity tariffs, particularly for the main customers, due to increased competition. For the private customer range the prices reduced first. However so far, only a few private customers in Germany have changed provider (see Table 1). There-fore, within the private customer range in Germany, the electricity tariffs rose once again. In England, however, private customers frequently changed provider due to strong electricity tariff variations.

Through various developments of the electricity market in individual European countries and the additional competitive pressure on the individual providers, strong strategic be-haviour of individual providers increases. Mergers of large power suppliers in Germany reduced and rather obstructed competition, since the free decentralised market access also is decreased for small electricity providers. Although France so far has opened its electricity market, the largest French provider EdF dominates the electricity supply in France and increasingly also in Europe. Because the nationally controlled giant pursued a strong policy of expansion overseas, it is however hardly possible for current European providers to expand into the French market. Therefore the European commission has al-ready demanded extra time to regulate the European market as uniformly as possible and to decrease market power by an independent adjustment authority.


7

Table 1: Liberalisation of the Electricity Market in Europe Country Percent

of liber-alisation

Date of com-plete liberali-sation

Main providers Market share of the main providers

Consumers who changed providers

Austria 100% 2003 EVN, Verbund, Wiener Stadtwerke

68% 5-10%

Belgium 35% 2007 Electrabel 97% 5-10% Denmark 90% 2003 SK Power Company 75% N/A Finland 100% 1997 Fortrum, Ivo Group 54% 30% France 30% Discussion

not ended EDF 98% 5-10%

Germany 100% 1999 Bewag, E.On, EnBW, RWE, Veag

63% 10-20%

Greece 30% Not discussed AEH (public company) 100% None Ireland 97% 2007 ESB 97% 30% Italy 35% Not discussed Elettrogen, Enel 79% Less than 5% Luxemburg 50% 2007 Cegetel 90% N/A Netherlands 33% 2003 Essent, Nea 64% 10-20% Portugal 30% Not discussed EDP 85% Less than 5% Spain 45% 2003 Endesa, Hidroelectrica

del Cantabrico, Iberdro-la, Union Fenosa

79% Less than 5%

Sweden 100% 1998 Sydkraft, Vattenfall 77% N/A UK 100% 1998 British Energy, Innogy,

Powergen, Scottish and Southern Energy, Scot-tish Power

44% 80%

Source: Benchmark Report of the European Commission, see European Commissions (2003)

In Germany, the liberalisation of the electricity market first led to substantial electricity price reductions, but increased again because of firm mergers and increased market shares. Lise, Kemfert and Tol (2002) report on this and provide detailed model simula-tions.

The European Commission criticised especially the net access price developments in the German market. Whereas the differences between net purchase and sales prices are at a constant low level (with few exceptions in peak times) in Austria and United Kingdom, the German net access sales prices are permanently at a very high level. In addition, the purchase prices are very low in Germany so that the difference between purchase and sales price is continually at a very high level, independently of hourly differences, see Figure 4.


8

3. The Game Theoretic Modelling Approach

We investigate the market developments in seven European countries using the game theoretic model EMELIE (Electricity MarkEt Liberalisation In Europe). EMELIE can be characterised as a small numerical model, with which market behaviour by firms in oli-gopolistic markets can be investigated. Energy suppliers in EMELIE produce electricity through different technologies. One player can own several power plants, of which total capacity is considered, as well as variable production costs. Investment decisions are not considered here; we limit ourselves to the study of the economic and environmental im-pact of liberalisation with the present capacity structure.

The model is calibrated by providing the electricity retail price and the actual demand in a particular base year. In this paper we take the year 2000 as the base year. In the calibra-tion of the model, production costs are minimised to see whether it is possible to meet the required demand. Below we refer to the results from this calibration as the reference case (REF).

After calibration the EMELIE model use three different sets of assumptions about the behaviour of the producers in the model to compute the market outcome in equilibrium. Besides data on electricity producers, price elasticities of demand and transmission grid capacity are specified exogenously. EMELIE then determine national electricity prices, marginal electricity production costs, and produced and traded electricity per technology and firm.

The first case uses the assumptions of the perfectly competitive market, i.e. that firms are price takers and act as if they couldnt influence the market price. This case is referred to as the perfect competition case, COMP. The perfectly competitive outcome is equivalent to Bertrand oligopoly competition, i.e. when firms set prices taking into account the ac-tions of the other firms.

In contrast to the perfectly competitive situation, in the second case producers are as-sumed to act strategically. Each player decide on production quantity taking into account the strategic choices of the other players. This is the Cournot-Nash oligopoly model, which is characterised by mutual, strategic reactions by market players. This result leads to a Nash equilibrium where the strategies of all market actors are best responses to the actions of all other market players. However, a so-called competitive fringe consisting of the total sum of small decentralised production units has been included, and this fringe is assumed to always behave as price takers. Strategically behaving firms can influence prices by changing production. We refer to this as the STRA case.

In the Stackelberg game, one firm, in this case the one with the largest production capac-ity, is a leader and moves first, while other firms are followers and take the action of the leader as given. The followers behave just like the STRA case, except for the competi-tive fringe who always behaves as in the COMP case. The leader chooses its output knowing ultimately the outcome will lie on the followers reaction curve, which means that the followers reaction curve is a constraint for the leader. The leader chooses its output to maximise profit given this constraint. We refer to this case as the STACK case.


9

The Annex provides a more detailed description of the EMELIE model.

The model is implemented in GAMS and formulated as a mixed complementarity prob-lem (MCP). This is both a fast and efficient way to solve the model for different scenar-ios and with different data sets.

4. Calibration of EMELIE to the European Electricity Market

As was explained in the previous section, the EMELIE model is a game-theoretic com-putable model of the wholesale electricity market. The model uses different assumptions about market structuresBertrand, Cournot-Nash and Stackelberg oligopolyto com-pute the market outcome in terms of prices, quantities and technology use. The current version of the EMELIE model is static and considers international trade exogenously, namely by modelling each neighbouring country as a separate player with a production capacity equal to the transmission capacity from that country. The model is run sepa-rately for each country.

The supply side of the model consists of a number of electricity producers that produce and sell electricity. For each producer the production capacity of different production technologies are specified. To account for the great number of small decentralised pro-duction units, a competitive fringe is added to the model. As their sizes are small, they always act as price takers in the model.

The demand side of the model consists of one sector demanding the electricity produced, as we generally observe only a single power exchange market within a country. This market trades electricity measured in terawatt hours (TWh). As we are mainly interested in the economic consequences of liberalisation, we consider trade over the period of one year. It is not possible to associate emissions to a model that only considers the possibil-ity to meet peak demand. Hence, the model outcome shows to which extent demand can be met within the current capacity structure.

The model allows for many different production technologies. At this stage 12 different production technologies have been considered. Conventional thermal power technologies are nuclear (N), coal (C), gas (G), lignite (L) and oil (O). There are five different types of combined heat and power production (CHP): gas (CHP-G), coal (CHP-C), oil (CHP-O), biomass (CHP-B) and other fuels (CHP-X). Finally, hydro (H) and wind power (W) are also available technologies in the model.

We calibrate the EMELIE model by considering data for the benchmark year 2000: pro-duction capacities of the largest producers, variable production costs for different tech-nologies, transmission capacities between countries, together with wholesale market price and demand data.

The data used in this report is from the base year 2000 and includes the following coun-tries: Denmark, Finland, France, Germany, the Netherlands, Norway and Sweden. A brief summary of the total production capacities in these countries is given in Table 2.


10

Table 2 Electricity production capacities in 2000

(TWh / year)

Denmark Finland France Germany Netherlands Norway Sweden

N I N I N I N I N I N I N I

Nuclear 0 9.2 21.3 16.7 403.6 23.8 167.9 35.6 3.8 5.4 - 9.1 68.1 3.5

Coal 16.8 2.6 16.0 - 88.8 13.1 178.9 26.8 34.6 - - 3.9 - 17.3

Lignite - 1.5 - - - 2.4 99.3 - - - - - - 0.4

Gas - 2.7 6.3 0.4 13.2 18.5 157.7 10.2 61.2 - - 0.5 - 1.5

Oil - 2.8 8.7 1.8 85.6 10.3 64.1 3.5 8.5 - - 2.5 22.8 1.4

CHP-G 8.6 0.2 8.1 0.06 - 0.2 10.2 3.5 39.8 - - 2.1 0.6 5.8

CHP-C 7.1 1.0 6.6 0.2 - 1.3 52.1 2.9 - - - 2.0 2.5 5.0

CHP-O - 0.3 0.7 0.3 - 0.06 2.6 0.07 - - - 0.4 2.9 0.1

CHP-B - 0.2 4.7 0.2 - - - 0.05 5.4 - - 0.3 2.1 0.6

CHP-X - 1.0 6.5 0.4 29.9 0.8 32.9 1.2 - - 0.9 0.6 4.5 1.1

Hydro 0.03 15.7 12.5 6.8 72.6 38.9 37.4 33.9 0.3 22.1 142.3 8.2 64.4 23.1

Wind 4.4 0.06 0.08 0.05 0.2 0.02 0.7 1.8 1.9 - 0.03 1.1 0.5 2.4

Total 37.0 37.3 91.4 27.0 693.9 109.5 803.6 119.6 155.7 27.5 143.2 30.7 168.5 62.2

Notes: N shows total domestic capacity and I refers to import capacity. There are differences in available technologies between countries, and not all technolo-gies are present in each country. For each country and technology variable production costs have been specified, which are shown in Table 3.

Table 3 Electricity production costs in 2000 (/MWh) Denmark Finland France Germany Netherlands Norway Sweden Nuclear 14.22 2.96 5.00 11.19 5.00 SWE 4.32 Coal 14.88 DEN 15.11 FRA 23.00 DEN DEN Lignite FRA - GER 15.65 - - GER Gas 18.29 30.52 44.45 34.28 41.73 SWECHP - Oil 18.61 46.91 48.46 36.70 53.83 SWE 79.69/54.40 CHP-G 10.54 14.99 GER 20.21 21.90 DEN 29.68 CHP-C 8.61 9.10 GER 9.64 - DEN 21.94 CHP-O 9.30 23.33 GER 20.32 - SWE 35.22 CHP-B 11.76 SWE SWE SWE 19.06 SWE 19.15 CHP-X SWE 16.64 20.98 10.00 - SWE 26.50 Hydro 1.28 0 12.93 1.30 9.08 0 3.55 Wind 1.28 0 0 1.30 9.08 DEN 0 Notes: When a production cost estimate for another country has been used, this is indicated by

the first three letter of the country from which the cost estimate is borrowed. SWECHP indi-cates that the Swedish cost estimate for CHP has been used. The oil cost estimate for Sweden refers to gas turbine and condensing, respectively.


11

There are differences in variable production costs between countries, which is the result of different fuel and production taxes in different countries. These differences also re-flect that it is very difficult to get exact production cost estimates, although it should be noted that relative costs between different technologies are roughly the same in all coun-tries. In some cases, national cost estimates were not available, and then cost estimates for some other country has been used, as indicated in Table 3.

Capacity and production cost data and the price elasticity of demand are crucial model data and parameter. Electricity consumers are not very price sensitive. Therefore, we as-sume a rather moderate price elasticity of demand of 0.400 for five of the countries, while we use slightly smaller price elasticities for Denmark and Netherlands (0.295 and 0.290, respectively), based on average values found in the literature (see for example Andersson, 1997, Brubakk et al, 1995, Grohnheit and Larsen, 2001 and Koopmans et al, 1999).

5. Country Results and Sensitivity Analysis

The EMELIE model has been run for seven countries and four cases. Table 4 summarises the results by stating the number of (strategically acting) firms, the number of trading partners, and the equilibrium prices and quantities for the four different cases:

Table 4 Model results for seven countries POP REF COMP STACK STRA n c (mil) p q p Q p q p q

Denmark 2 3 5 12.75 32.83 10.54 34.73 14.20 31.81 14.29 31.75Finland 2 3 5 14.88 76.41 14.88 76.41 16.64 73.07 16.64 73.07France 1 6 60 20.81 410.67 12.93 496.78 24.85 382.55 48.46 292.85Germany 4 8 83 15.19 477.00 15.11 478.01 20.33 424.53 20.36 424.28Netherlands 6 2 16 39.65 100.60 29.11 110.03 38.71 101.30 38.71 101.30Norway 6 4 5 12.25 110.92 5.23 155.87 5.50 152.85 5.50 152.85Sweden 6 5 9 14.26 135.46 9.83 157.20 14.98 132.82 15.78 130.08Notes: n refers to the number of modelled firms in each country, c refers to the number of mod-

elled neighbouring countries, p refers to the price in euros per MWh, and q refers to national supply in TWh.

A number of insights can be derived already from Table 4 . For instance the following relation holds:

COMP STACK STRA COMP REF and p p p p p

On the one hand, we find that the reference price for Finland and Germany are (nearly) equal to the price in perfect competition. This means that the actual production capacity and costs are so stringent that they are just enough to meet demand in the fully competi-tive market. It also indicates that in these two countries almost perfect competition or full liberalisation is already reached. Especially in Germany, we observed huge changes due to increased competition within the last two years.


12

On the other hand, we find that the reference price is even higher than Cournot price for the Netherlands, Norway and Sweden. Therefore, further liberalisation of the markets in these countries should lead to lower prices. It also indicates that in Norway, in the Neth-erlands and in Sweden firms choose the market price far above the competitive price (marginal cost) in order to increase profits. However, in comparison to the full competi-tion case no additional profits could be reached by becoming a Stackelberg leader in these three countries.

In Norway firms obtain prices far above the full competitive price, as the COMP, STACK and STRA price is less than half the reference price (see Table 4). The main ex-planation for this is that the price difference among competing technologies is quite low. Furthermore, Norway has a high hydro production capacity. First, hydro power capacity varies from year to year, and the measure that has been used in the model is the actual production in 2000, when hydro production was very high. Second, transmission capac-ity between Sweden and Norway is very high, and Norway can therefore benefit from importing cheap hydro and nuclear power from Sweden. The same reasoning holds for Sweden.

In Finland we can see that firms already have chosen the perfectly competitive price. Full liberalisation of the electricity market in Norway has already brought a fully com-petitive situation. However, firms could increase their gains by acting strategically.

Finally, the model finds very high prices for the Cournot equilibrium in France as com-pared to other countries. The Cournot price is about four times the competitive price, which is of course a result of the dominant, near monopoly, position of EdF. These high prices are caused by a (strategic) drop in production by EdF. Possibly, the consumer price elasticity of French consumers is underestimated here, as when this value is raised to 0.8, we no longer find this extreme result (pSTRA=22.33 /MWh). It could also be noted that despite the predominant position of EdF, the electricity price in France is not particularly high (except for the Cournot price). This is most easily explained by the large nuclear power capacity, which has a low marginal cost of production.

The price of electricity in the Netherlands is very high compared to other countries in 2000, which is a result of high oil and gas prices in year 2000. The results for the Nether-lands are similar to the results for other countries. The Cournot price of 38.71 euros might appear high, but it is about 33 percent higher than the competitive price, which is similar to most other countries.

In Finland, the Netherlands and Norway, the STACK outcome is equal to STRA, while it is interesting to know that in the other countries the Stackleberg leader always in-creases profits at the cost of lowering the profits of all other firms. According to eco-nomic theory, the Stackelberg equilibrium price is expected to be in between the Cournot and perfect competition price. The model run results are therefore in line with economic theory. Economic theory also predicts that the Stackelberg leader is generally better off in the Stackelberg equilibrium as compared to the Cournot equilibrium. This is also what the model results suggest for the various countries studied (the result on the profits is not shown in this paper).

Besides, we can also study the differences in competitiveness within the various coun-tries by calculating the HHI (the formula can be found in the appendix).


13

Table 5 The Hirschmann-Herfindahl index for the seven countries and four model runs.

Denmark Finland France Germany Netherlands Norway Sweden REF 1520 2063 9048 1667 1066 1575 1888 COMP 1405 2096 7414 1651 985 894 2115 STACK 1315 1237 3871 926 885 892 1565 STRA 1314 1237 1432 923 885 892 1495 Note: The Fringe which consists of many small firms is excluded from the HHI calculation.

Table 5 clearly shows that France is close to a monopoly. This may also explain the ex-treme price increase once the dominating firm starts to act strategically. Once a firm ex-ploits its market power, the HHI value naturally drops due to the lower supply. Further-more, it is interesting to see that besides France, Finland and Sweden have particularly high market concentrations, as their HHI values are above the threshold of 1800 in the reference situation.

The comparison of the results for different countries showed that in all studied countries (Cournot) strategic behaviour of firms leads to higher electricity prices and lower pro-duction quantities than in both the reference and perfect competition cases. Because of the monopolistic electricity market situation in France, strategic behaviour leads to very high prices and a substantially reduced electricity production. In Norway, we find very little difference between different scenarios. This is primarily because utilities always produce at full capacity in all scenarios, as electricity can be produced at zero cost (hy-dro and wind). Therefore, the Norwegian electricity market is resistant against any mar-ket changes towards more competition or strategic actions.

These broad statements about the model results give a general picture of the performance of the EMELIE model with data for year 2000. A more detailed picture of the results is given in the next section, where the results are analysed individually for each country.

5.1 Denmark

In contrast to other Nordic countries, Denmark liberalised its electricity market slowly but reaches presently a nearly full market opening. There are two large producers in Denmark, Elsam and Energi E2, which control 40 and 34 percent of total national gen-eration capacity, respectively.

There are two peculiarities with the Danish electricity market that are noteworthy. First, the market has a natural regional division. The electricity grid in Denmark is divided be-tween the eastern and western parts of the country, and in many respects each part is more integrated with the surrounding countries (Sweden, Norway and Germany) than with each other. This situation, however, does not influence the model results since Denmark is treated as one integrated market in the model. Secondly, CHP is the second most important electricity generation technology after conventional coal combustion. The higher efficiency of CHP is, like for other countries, captured by lower technology prices in the model. Table 6 presents the results for Denmark.


14

Table 6 Model results for Denmark p

Market shares, % (Elsam / E2 Energi / Fringe / Sweden / Norway / Germany)

Technology shares, % (CHP-G / CHP-C / CHP-O / H / W)

Marginal technologies

REF 12.75 20 / 6 / 24 / 21 / 26 / 4 15 / 24 / 1 / 46 / 13 CHP-G COMP 10.54 19 / 8 / 26 / 20 / 24 / 4 20 / 23 / 1 / 44 / 13 CHP-G STACK 14.20 12 / 8 / 30 / 20 / 27 / 5 20 / 19 / 0.1 / 48 / 14 CHP-G, CHP-C, CHP-O STRA 14.29 12 / 8 / 30 / 20 / 27 / 5 20 / 19 / 0.1 / 48 / 14 CHP-G, CHP-C, CHP-O Notes: The market share of a player is defined as the production of the player divided by the total

production of all players. Technology shares denote the share of a certain technology as a percentage of total production. Marginal technologies are technologies for which only a part of total capacity is used in production. The Stackelberg leader is in bold font.

Table 6 shows that hydro power production in Sweden and Norway account for almost half of the supply in Denmark. It should be noted, however, that international trade is exogenous in the model, which means that Danish imports do not reduce the supply in Sweden and Norway. Table 6 also verifies the important role of co-generation (CHP) in Denmark CHP accounts for almost half of total supply and it is the marginal technol-ogy in all cases. This means that the outcome hinges critically on assumptions about CHP capacities and production costs.

5.2 Finland

The Finnish electricity market is fully liberalised since 1997. The market is dominated by the state-owned producer Fortum which controls about 35 percent of total national electricity generation capacity. The second largest player, PVO, is owned by electricity-intensive industry and controls around 20 percent of total capacity. Decentralised pro-duction, the competitive fringe, is quite substantial and accounts for about 40 percent of total capacity. The electricity production technology is mixed, although nuclear is the most important technology.

Table 7 Model results for Finland p Market shares, %

(Fortum / PVO / Fringe / Sweden / Russia / Norway)

Technology shares, % (N / C / CHP-G / CHP-C / CHP-X / H / W)

Marginal tech-nologies

REF 14.88 36 / 18 / 16 / 17 / 13 / 1 48 / 18 / 0 / 9 / 0 / 25 / 0.2 C COMP 14.88 36 / 18 / 15 / 17 / 13 / 1 48 / 18 / 0 / 9 / 0 / 25 / 0.2 C STACK 16.64 24 / 14 / 31 / 17 / 14 / 1 51 / 3 / 8 / 8 / 5 / 26 / 0.2 C, CHP-G,

CHP-C, CHP-X STRA 16.64 24 / 14 / 31 / 17 / 14 / 1 51 / 3 / 8 / 8 / 5 / 26 / 0.2 C, CHP-G,

CHP-C, CHP-X Notes: See notes under Table 6.

5.3 France

France has liberalised its electricity market gradually. Although the market is liberalized in some respects, the state-owned producer EdF dominates the market with some 85 per-


15

cent of total capacity; a near monopoly situation. The dominating technology is nuclear power, accounting for a major part of total electricity production in France. Additionally, France is a net-exporter of electricity with net exports of about 70 TWh in 2000.

The result in the STRA case is quite different from the others, as can be seen from Table 8. In the Cournot equilibrium EdF restricts output so that its market share drops from 86 to 36 percent! Interestingly, in this scenario the market share of the competitive fringe is much higher than in the other scenarios producing a higher share of electricity produc-tion with a completely different mix of technologies, see Table 8.

Finally, we can also observe that the Stackelberg price is twice the competitive price, which is a result of the Stackelberg leader, EdF, being so large compared to the other players. EdF is able to increase profits by acting as a leader. Then the consumer is much better off than under STRA as the price is halved in STACK.

Table 8 Model results for France p Market shares, %

(EdF / Fringe / Switzer-land / Italy / Belgium / Germany / UK / Spain)

Technology shares, % (N / C / L / G / O / CHP-G / CHP-C / CHP-O / H / W)


REF 20.81 95 / 0 / 0 / 0 / 3 / 1 / 1 / 1 100/ 0/0/0/ 0/ 0/ 0/ 0/ 0/0.04 N COMP 12.93 86 / 2 / 5 / 2 / 2 / 1 / 1 / 1 84/ 0/0/0/ 0/ 0/0.2/16/ 0/0.04 H STACK 24.85 62 / 18 / 7 / 2 / 3 / 4 / 3 / 2 68/12/1/0/ 0/0.1/0.3/ 8/12/0.04 N, C, H STRA 48.46 36 / 37 / 9 / 3 / 4 / 5 / 4 / 3 44/15/1/4/10/0.1/0.4/10/15/0.05 N, C, G, O, HNotes: See notes under Table 6.

5.4 Germany

Germany is one of the European countries that have fully liberalised their electricity markets. There are four large electricity producers in Germany, namely Vattenfall, E-ON, EnBW, RWE and a competitive fringe having 20% of total national capacity. Vat-tenfall is the largest electricity producer, owning 32% of total national capacity.

Table 9 Model results for Germany p Market shares, %

(E-ON / EnBW / RWE / Vattenfall / Fringe / Austria / Benelux / Switzerland / France/Denmark/Poland/Sweden/Central)

Technology shares, % (N / C / L / CHP-C / CHP-X / H / W)

Mar-ginal tech-nologies

REF 15.19 17 / 19 / 6 / 31 / 12 / 3 / 2 / 3 / 4 / 1 /0/1/2 41 /26 /0 /11 /7 /15 /1 C COMP 15.11 17 / 19 / 6 / 30 / 12 / 3 / 2 / 3 / 4 / 1 /0/1/2 41 /26 /0 /11 /7 /15 /1 C STACK 20.33 18 / 10 / 9 / 18 / 21 / 4 / 2 / 3 / 5 / 3 /2/1/5 46 /15 /2 /13 /8 /16 /1 N, C, L STRA 20.36 18 / 10 / 9 / 18 / 21 / 4 / 2 / 3 / 5 / 3 /2/1/5 46 /15 /2 /13 /8 /16 /1 N, C, L Notes: See notes under Table 6.

As we have already seen from Table 3, in Germany firms are already producing elec-tricity in an almost fully liberalised and competitive market situation. Both market shares of producing firms and technology shares are almost identical in the reference


16

and competition situation. Strategic behaviour leads to different shares of firms and technology mix.

5.5 The Netherlands

In the Netherlands, gas, exploited from Dutch sources, is the most important fuel used in electricity production. In 2003, the Netherlands has opened up its electricity market to 33 percent. Six different electricity producers have been modelled for the Netherlands to-gether with a competitive fringe consisting of around 20 percent of total capacity. The largest firm is Electrabel which controls around 25 percent of total capacity. Essent and Reliant are the second largest firms, each controlling about 20 percent of national capac-ity.

Table 10 Model results for the Netherlands p Market shares, %

(Electrabel / Reliant / E-ON / Essent / Delta / NUON / Fringe / Belgium / Germany)

Technology shares, % (N / C / CHP-G / CHP-B / H / W)


REF 39.65 4 / 9 / 7 / 21 / 4 / 1 / 28 / 5 / 21 9 / 24 / 38 / 5 / 22 / 2 C COMP 29.11 5 / 10 / 8 / 19 / 5 / 3 / 26 / 5 / 20 8 / 31 / 35 / 5 / 20 / 2 STACK 38.71 6 / 11 / 9 / 12 / 6 / 3 / 28 / 5 / 21 9 / 25 / 38 / 5 / 21 / 2 C STRA 38.71 6 / 11 / 9 / 12 / 6 / 3 / 28 / 5 / 21 9 / 25 / 38 / 5 / 21 / 2 C Notes: See notes under Table 6.

5.6 Norway

Norway liberalised its electricity market earlier than all other countries studied here, and it is also the home country of the Nordic electricity exchange Nordpool. Norway is also special in that almost all of the generation capacity consists of hydro power. This means that generation capacity varies from year to year depending on hydrological conditions in 2000 Norway produced a huge amount of hydro power and it is perhaps not a very representative year. The dominance of hydro power therefore makes it difficult to choose a correct electricity generation capacity figure since it varies from year to year. The vari-able production cost of hydro power is set to zero in the model.

The dominating electricity producer in Norway is the state-owned company Statkraft which controls about 30 percent of total capacity. The other five players that have been modelled are rather small, controlling around 58 percent of total capacity. The competi-tive fringe is therefore quite large, accounting for almost 40 percent of total capacity.


17

Table 11 Model results for Norway p Market shares, %

(Statkraft / Oslo / Norsk Hydro / Agder / BKK / Lyse / Fringe / Sweden / Denmark / Finland / Russia)

Technology shares, % (N / H / W)


REF 12.25 35 / 8 / 10 / 8 / 7 / 6 / 21 / 5 / 0 / 0.1 / 0.0 0 / 100 / 0 H COMP 5.23 25 / 6 / 7 / 6 / 5 / 4 / 36 / 10 / 1 / 0.2 / 0.3 6 / 94 / 1 STACK 5.50 26 / 6 / 7 / 6 / 5 / 4 / 36 / 9 / 1 / 0.2 / 0.3 4 / 96 / 1 N STRA 5.50 26 / 6 / 7 / 6 / 5 / 4 / 36 / 9 / 1 / 0.2 / 0.3 4 / 96 / 1 N Notes: See notes under Table 6.

5.7 Sweden

Sweden has like the other Nordic countries reached far in the process of liberalisation. The dominating player is state-owned Vattenfall which controls more than 40 percent of total capacity. The second largest player is Sydkraft, which controls 20 percent of total capacity. The third largest player in 2000 was Birka Energi with 13 percent of total ca-pacity. Later the Finnish producer Fortum has aquired Birka Energi and shares in other companies so that Fortum is now the third largest producer in Sweden. In total, six play-ers have been modelled together with a fringe consisting of 17 percent of total capacity.

Nuclear and hydro power are the most important production technologies in Sweden. In 2000, hydro power production was very high and actual production exceeded the average production, which has been used as a capacity measure here.

In Sweden the Stackelberg equilibrium reduces the price under strategic behaviour with 5%. The situation resembles that of France with a large, predominant producer, Vatten-fall, which results in a quite substantial first-mover advantage.

Table 12 Model results for Sweden p Market shares, %

(Vattenfall / Sydkraft / Birka / Fortum / Skellefte / Graninge / Fringe / Norway / Finland / Denmark / Germany / Poland)

Technology shares, % (N / C / H / W)

Marginal tech-nologies

REF 14.26 51 / 8 / 8 / 4 / 2 / 2 / 4 / 15 / 3 / 2 / 1 / 0 35 / 0 / 63 / 2 N COMP 9.83 44 / 16 / 11 / 4 / 2 / 1 / 4 / 13 / 3 / 2 / 1 / 0 44 / 0 / 54 / 2 STACK 14.98 34 / 19 / 13 / 4 / 2 / 2 / 5 / 16 / 3 / 2 / 1 / 0.3 34 / 0.3 / 64 / 2 N, C STRA 15.78 29 / 20 / 14 / 4 / 2 / 2 / 5 / 16 / 3 / 2 / 1 / 2 29 / 3 / 65 / 2 N, C Notes: See notes under Table 6.

5.8 Environmental effects

The model results show that applied production technologies vary widely. Market behav-iour does not only affect prices and production quantities, but also applied production technologies. Since different production technologies have different environmental im-pacts, model results indicate environmental effects of liberalisation and different market conditions.


18

It is not meaningful to compare the production technology mix in the REF case with the other cases, since the technologies used in the REF case does not reflect the actual pro-duction technologies that was used in the base year. Therefore, we are comparing the COMP case with the STACK and STRA cases.

Prices are higher in the STRA case than the COMP case, which makes it possible to profitably produce with higher-cost technologies in the STRA case. This could lead to an increase of applied production technologies. Model results shows that this is indeed true in all countries we find more applied technologies in the STRA case than in the full competitive market situation.

However, there is no clear relation between which technologies that is used in the STRA and COMP cases. The result depends on the market shares of the firms in the different cases and how production technologies are distributed among firms. For example, in France the share of nuclear production decreases and CHP and hydro production shares increase when moving from a perfectly competitive market to a Cournot oligopoly mar-ket. In Germany, the share of nuclear production decreases, while the coal production share increases. The STACK case either collapses with STRA, in case where the leader does not have substantial market power, or leads to a mixed case between COMP and STRA. Then STACK has the consumer advantage of a price lower than STRA and the environmental advantage of a production lower than COMP.

Therefore, the results and country comparisons are very ambiguous. Higher electricity prices do not necessarily lead to higher gains as electricity demand is decreasing. Nu-clear technology is environmental friendly from the climate perspective as it does not produce greenhouse gases like coal or oil plants. However, nuclear produces wastes that harm the environment. Firms apply a specific technology primarily in the context of their economic effectiveness and not because of the environmental hazard. Because of that, we can detect that countries apply renewable technology primarily because of its tech-nology and natural availability (like hydro in Finland, Denmark and Norway). The mar-ket situation does not necessarily bring an increase of environmental friendly technol-ogy. However, because a fully competitive market situation forces firms to choose mar-ket prices at their marginal production costs, low cost technologies are favoured that are not necessarily environmental friendly.

6. Conclusions and Outlook

Model simulations of the European electricity market show that firm behaviour varies substantially after a market liberalisation. In some countries, simulations show that prices will go down irrespective of the resulting behaviour. However, in other countries, competition is already so severe that prices can only rise in order to avoid bankruptcy of power firms.

The main objective of liberalisation of the electricity market is to bring the prices down. We find the lowest prices in the case with perfect competition (Finland, Germany). Al-ternatively, firms may also revert to strategic action, but dependent on the technology


19

production costs (Finland, Norway) and market power (France), it may or may not make a big difference. It also matters to which extent consumers are sensitive to wholesale market prices of electricity. A higher sensitivity may offset windfall profits under strate-gic action (France).

The liberalisation of the European electricity market leads to an application of low cost technologies which are not necessarily environmental beneficial. However, in some countries with a high share of renewable energy production technology (Norway, Den-mark, Finland) strategic market behaviour increases the share of environmental friendly applied production technology.

Acknowledgements

Funding from the Ministry of Science and Culture in Germany to build the model as well as funding from the EU (contract number NNE5-2001-00519) for the EMELIE project is gratefully appreciated. Any remaining errors are ours.

References

Amundsen, E. and Bergman, L. (2002) Will cross-ownership re-establish market power in the Nordic power market? The Energy Journal 23(2):7395.

Andersson, B. (1997) Essays on the Swedish Electricity Market. Dissertation, Stockholm School of Economics, Stockholm.

Bigano, A. and Proost, S. (2002) The opening of the European electricity market and environ-mental policy: does the degree of competition matter? Center for Economic Studies, Katholieke Universiteit Leuven. Available from: http://weber.ucsd.edu/~carsonvs/papers/602.pdf

Bower, J., Bunn, D. W. and C. Wattendrup (2001) A model-based analysis of strategic consolidation in the German electricity industry, Energy Policy 29:9871005.

Bower, J. and Bunn, D. (1999) A Model-Based Comparison of Pool and Bilateral Market Mechanisms for Electricity Trading, London.

Brubakk, L., Aaserud, M., Pellekaan, W. and von Ostvoorn, F. (1995) SEEM - An Energy Demand Model for Western Europe. Report 95/24, Statistics Norway, Oslo.

Cardell, J., Hitt, C. C. and Hogan, W. W. (1996) Market Power and Strategic Interaction in Electricity Networks, Harvard Electricity Policy Group, Cambridge, Mass.

Dawson, Paul and Shuttleworth, Graham (1997) Efficient Transmission Pricing in Norway and Sweden, Report for NVE, National Economic Research Associates, London.

Day, C. and Bunn D. (1999) Generation Asset Divestment in the England and Wales Electricity market: A Computational Approach to Analysing Market Power, London.

Drake, F.-D., T. Ohler, et al (2000) Vom Preiskrieg zum Servicewettbewerb, Energiewirtschaftliche Tagesfragen 5:286291.


20

European Commission (2003): Second Benchmarkreport of the implementation of the European Electricity and Gasmarket, Brussels.

Ferris, M. and Pang, J. (1995) Engineering and Economic Applications of Complementarity Problems. Wisconsin.

Ferris, M. and Sinaoiromsaran, K. (1998) Formulating and Solving Nonlinear Programs as Mixed Complementary Problems. Wisconsin.

Fudenberg, Drew and Tirole, Jean (1992) Game Theory, 2nd print, MIT Press, Cambridge, Mass. Green, Richard (1997) Electricity Transmission Pricing An Internal Comparison, Department for

Applied Economics and Fitzwilliam College, Cambridge. Grohnheit, P.-E. and Larsen, H. V. (2001) Balmorel Data and Calibration.

Version 2.05. Available from: http://www.balmorel.com. Helman, U., Hobbs, J.B., Cardell, T., Luong, T. and Wander, M.T. (1999) Modelling Strategic

Pricing in Bilateral and Poolco Electricity Markets: A Nash-Cournot Approach with Applica-tions to US Eastern Interconnection, INFORMS national meeting, Philadelphia, Npv. 7-10

Jing- Yuan, W. and Smeers, Y. (1999) Spatial oligopolistic electricity models with Cournot Gen-erators and regulated transmission prices. Operations Research 47(1):102112.

Kemfert, C. (1999) Liberalisation of the German Electricity Market - Strategies and Opportuni-ties. Nota di Lavoro, 95-1999.

Kemfert, C. and Tol, R. S. J. (2000) The Liberalisation of the German Electricity Market -Modelling an Oligopolistic Structure by a Computational Game Theoretic Model. Working Paper 89/2000, Oldenburg.

Kemfert, C., Barbu, D., Kalshnikov, V. (2002): Economic Impact Assessment of the European Electricity market, Working Paper, Oldenburg

Koopmans, C.C., te Velde, D.W., Groot, W. and Hendriks, J.H.A (1999) NEMO: Netherlands Energy demand MOdel a top-down model based on bottom-up information. CPB Research Memorandum (155), The Hague.

Lise, W., Kemfert, C. and Tol, R. S. J. (2003) Simulation of the Process From a Monopolistic To a Liberalised Electricity Market in Germany. Submitted to Energy Economics.

McKinsey (2001) Paying a green premium, The McKinsey Quarterly, 2001 Number 3. Murphy, F. H., H. D. Sherali, et al (1986) A mathematical programming approach for determin-

ing ologopolistic market equilibrium. Mathematical Programming 24:92106. Newberry (2000) Privatisation, Restructuring, and Regulation of Network utilities. MIT Press. Newberry (2002a) Regulatory Challenges to European Electricity Liberalisation. Cambridge. Newberry (2002b) Problems of Liberalisation of the energy utilities. European Economic Re-

view, Vol. 4. Oschmann, V. (2000) Das Gesetz fr den Vorrang Erneuerbarer Energien. Energiewirtschaftliche

Tagesfragen 6:460464. Prognos (2000) Energiereport III: Die lngerfristige Entwicklung der Energiemrkte im Zeichen

von Wettbewerb und Umwelt. Stuttgart, Schffer-Poeschel Verlag Stuttgart. Pfaffenberger, W. and Haupt, U. (2001) Wettbewerb auf dem deutschen Strommarkt- Zwei Jahre

nach der Liberalisierung, 2. IEWT 2001, Strategien und Instrumente fr den Wettbewerb in der Energiewirtschaft.

Rutherford, T. (1993) MILES, A Mixed Inequalitiy and non Linear Equation Solver. Salant, S. W. and Shaffer, G. (1999) Unequal treatment of Identical Agents in Cournot Equilib-

rium. The American Economic Review 89(3):585604.


21

Stern, J (1998) Competition and Liberalisation in the European Gas Market: A Diversity of Mod-els, London.

Strombasiswissen (2002) Kooperationen unde Fusionen verndern die Stromlandschaft. Available from: http://strombasiswissen.bei.t-online.de/SB115-04.htm

Tirole, Jean (1988) The Theory of Industrial Organisation. MIT Press, Cambridge, Massachu-setts.

Verbndevereinbarung (2001) Verbndevereinbarung ber Kriterien zur Bestimmung von Netznutzungsentgelten fr elektrische Energie und ber Prinzipien der Netznutzung.

Wietschel, M., Dreher, M. and Rentz, O. (2001) Grne Angebote in Deutschland: Stand und Perspektiven; 2. IEWT 2001, Strategien und Instrumente fr den Wettbewerb in der Energiewirtschaft.


22

7. Mathematical Description of the EMELIE Model

The computational game theoretic model EMELIE is characterised by following indices, parameters and variables:

Indices: f i Parameters: cvi d0 p0 qmaxi,f Variables: p cmf sf qi,f

- Firms f F - Technologies i I - Variable production costs for technology i - Reference demand for electricity - Reference price for electricity - Price elasticity of electricity demand - Maximum production capacity with technology i in firm f - Electricity net transport losses - Demand price for electricity - Marginal and average costs of electricity production of firm f - Supply of electricity by firm f - Production of electricity by firm f with technology i

EMELIE is a partial general equilibrium model of a liberalised electricity market with multiple actors and is inspired by the theory of industrial organisation (Fudenberg and Tirole, 1992; Tirole, 1988). On the supply side, electricity producing firms maximise their profits. On the electricity demand side consumers maximise utility. In equilibrium, prices p clear at national markets.

Note first that the market needs to be closed on the (consumer) demand side. This is achieved by well-known inverse demand function:

0 0 0

ff F

ps d pp

=

(1)

Let us now consider the case with strategic interaction among firms at the supply side. In this case, electricity producing firms f maximise their net profits. They do this by choosing their strategies as represented by their supplies sf simultaneously by assuming that other firms do the same. This is equivalent to maximising the profit firm f can obtain by supplying to the grid. This profit is the difference between incomes from supplied electricity minus the cost of production:

( ) ( )( )mMaximise f f fS p S c s = (2) where ff FS s= (3)

where the dependence of the demand function p(.) on S constitutes strategic action, while from equations (1,3), the explicit functional form of p(.) can be derived.


23

The first order conditions for optimality of strategically acting firms follow directly from (1), by taking the partial derivatives with respect to sf:

m: 1 0ff ff F c p s =

, (4)

The sign is used to denote the dual variable to this equation. The equation on the left-hand-side holds when firm f has a positive supply, a result which is well-known from the Karush Kuhn Tucker conditions and is a typical characteristic of a mixed complementar-ity problem (MCP).

Equation (4) shows the equalisation between marginal costs and marginal income. The individual market shares in equation (4) are conveniently determined by:

: 0ff fg

g F

sf F

s

=

(5)

The marginal income in the case of strategic action is reduced by the factor market share divided by price elasticity of demand. The market share represents a monopoly mark-up.

Power supply by firm f to the grid can be generated with various technologies i, like nu-clear, coal, lignite, gas, oil, hydro, and so on. We assume here a % electricity transport loss, which is generally in the range of 05%.

( ) m,: 1 0i f f fi I

f F q s c

= (6)

In order to fully define the model, we need to restrict firms marginal costs. For that pur-pose we have used variable cost data per technology, which do not differ per firm. Then, a logical lower bound of marginal costs are these variable costs.

( ) v m max, ,, , : 0i f i f i fi f I F c c q q (7) where the available technology is restricted by an upper bound too.

Model (1,47) is used to calculate the Nash equilibrium in the case with strategic action, where market information is typically incomplete and we are dealing with the case of imperfect markets. This case of quantity competition shall be referred to as STRA.

A model with perfect markets is established by replacing equation (4) by:

m: 0f ff F c p s = (4)

Furthermore, in the case of equation (4), equation (5) should be eliminated as well from the model, as market share f is no longer a variable. Of course, the market shares now follow exogenously from the model. Model (1,4,6,7) is used to calculate the competitive equilibrium in the case without strategic action. This case is derived by reducing the de-mand function to an identity: p(.)=p. Here firms take market prices as given and we are dealing with the case of perfect markets. This case of price competition shall be referred to as COMP.


24

These model relations are written by the programming language GAMS, which decom-poses the non-linear program as a mixed complementary problem (MCP). This is solved by the non-linear MCP-solving algorithm MILES, which is a mixed inequality and non-linear equation solver. Partially, MILES approximates linear sub-problems by Lemkes algorithm and solves the non-linear program by the generalised Newton algorithm itera-tively with a backtracking line search. An optimal solution is found by maximising re-gional profit conditions reciprocally under all considered constraints.

It is also useful to verify whether the initial prices and demand are viable. This is done in the so-called REF case, where the production cost is minimised. This is expressed in the following equation.

( ) v, ,Minimise Cost i f i f ii I f Fq q c = (8) Model (1,68) is run as a non-linear programming problem to establish the REF case.

Main outcomes of the model are regional prices, interregional trade flows and the opti-mal market shares of each electricity producer from which the regional concentration of the industry can be calculated in terms of the Hirschmann-Herfindahl index (HHI). HHI is a measure for (regional) competitiveness (see also Tirole, 1988, pages 221223). For the industry as a whole the Hirschmann-Herfindahl index is calculated as follows:

2

,

,

HHI 10000f r

r R

f F f rr R f F

s

s

=

(9)

A fourth possible market situation is where one firm is a leader and moves first, while the others are followers and move second in reaction to the chosen supply by the leader. The followers behave just like the STRA case (while the fringe behaves as COMP). The leader chooses its output knowing ultimately the outcome will lie on the followers reac-tion function (4). The followers reaction function is a constraint for the leader. The leader chooses its output to maximise profit given this constraint.

The first order condition of the leader can then be written as:

m 1 fff

dSc p ds

=

(10)

where dS/dsf =1 for the follower who behaves as in the STRA case, while dS/dsf > 1 for the leader, which reduces the monopoly markup to the advantage of the leader. Hence, the outcome of the Stackelberg behaviour will lie between COMP and STRA.

Derivative dS/dsf does no longer cancel out in the case of the Stackelberg equilibrium, as the terms in the aggregate demand S = sf is built up from the supply of all firms, which can be expressed implicitly via (4) in the supply of the leading firm.


25

Let us now simplify the model of f to two firms, where i is the leader and j is the fol-lower.3 Then the reaction function sj(si) can be derived by rewriting (4):

( )( )

m

m

jj i

j

p cs s

p p c

=

( )( ) ( )

mm

m1 ; where ji j i i j j

jj

p c pS s s s s p p cp p c

= + = + = =

m 21

i ji j i j

dS p dp dSs cds dS ds

= +

m

2 21ji j

dpmji j j i j dS

ps cdS p dpdSds s c

= =

(11)

we know from the definition of inverse demand (1) that:

1dp pdS S

=

substituting this in (11) this leads to:

2 mj

i j j i

pdSds pc

=

+ (12)

Finally substituting (12) in (10) and after rewriting the expression leads to:

( ) ( )

( )2 m

m1 1

1 ; where ; the Lerner index1

f f ff ff f

f

p cc p

p

+

= =

(13)

Equation (13) is used as the FOC for the Stackelberg leader, while the followers are competing in quantities (4), whereas the competitive fringe is a price taker (4).

3 To derive, mathematically, the FOC for the leader with n followers, is not an easy task, as

expressing the reaction functions of each follower leads to n equations, with n variables. This can be aggregated to total demand, but the derivative w.r.t p is very complex. The present approach nevertheless already shows the advantage of moving first.


26

8. Figures

Figure 1: Degree of Liberalisation of World Countries (Red: Fully liberalised; Blued: Liberalisation underway)

-25

-20

-15

-10

-5

0

5

10

15

20

25

30

AT BE DE DK ES EU FI FR GR IR IT LX NL PT SW UK

in %

% Change since 1/1999 % Change since 07/2001

Figure 2 Percent Change of Consumer Electricity Price


27

-30

-20

-10

0

10

20

30

40

50

AT BE DE DK ES EU FI FR GR IR IT LX NL PT SW UK

in %

% Change since 1/1999 % Change since 07/2001

Figure 3:Percentage Change of Electricity Prices of Large Industries

Figure 4: Difference of Average net Purchase and Sales Prices in 2002

0

20

40

60

80

100

120

140

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24

hour

in E

uro

/M

Wh

Germany (E.On) Austria (APCS) UK

AbstractIntroductionEuropes Electricity Market StructureThe Game Theoretic Modelling ApproachCalibration of EMELIE to the European Electricity MarketCountry Results and Sensitivity AnalysisDenmarkFinlandFranceGermanyThe NetherlandsNorwaySwedenEnvironmental effects

Conclusions and OutlookAcknowledgementsReferencesMathematical Description of the EMELIE ModelFigures

The European Electricity Market – Does Liberalisation Bring

Documents

Transcript of The European Electricity Market – Does Liberalisation Bring