VERSION NON CONFIDENTIAL
Addressing the adverse effects of market
power
A FINAL REPORT PREPARED FOR CREG
March 2006
© Frontier Economics Ltd, London.
i
Frontier Economics | March 2006
Addressing the adverse effects of market
power
Executive summary.......................................................................................1
1
Introduction .........................................................................................5
2
Static concerns - range of mitigating measures...................................7
3
4
2.1
The range of possible contractual measures ............................................7
2.2
Rationale for contractual measures ........................................................ 10
2.3
Possible contractual measures................................................................. 10
2.4
Choice of contractual form ..................................................................... 12
2.5
Regulatory measures to support contractual measures ....................... 14
2.6
Contractual measures and security of supply........................................ 14
2.7
Treatment of existing contracts .............................................................. 15
2.8
Conclusions ............................................................................................... 17
The parameters of contractual measures........................................... 19
3.1
Introduction............................................................................................... 19
3.2
Volume and exercise price....................................................................... 19
3.3
Duration ..................................................................................................... 22
Data, methodology and scenarios......................................................25
4.1
Modelling market power using SPARK ................................................ 25
4.2
Overview of data....................................................................................... 26
4.3
Modelling methodology ........................................................................... 30
4.4
Scenarios for analysis................................................................................ 31
4.5
New entry and soft regulatory constraints ............................................ 35
Contents
ii
5
6
Frontier Economics | March 2006
Results ................................................................................................37
5.1
Base case..................................................................................................... 37
5.2
Sensitivity: market entry........................................................................... 45
5.3
Sensitivity: variation in flow over the northern interconnector......... 49
5.4
Summary .................................................................................................... 49
Dynamic concerns – barriers to entry ................................................ 51
6.1
Barriers to entry......................................................................................... 51
6.2
Analysis of incentives to withhold sites................................................. 53
6.3
Possible policy options to address site availability............................... 54
6.4
Conclusion regarding sites....................................................................... 57
Annexe 1: SPARK gaming module .............................................................59
Annexe 2: Data ............................................................................................ 61
Contents
iii
Frontier Economics | March 2006
Addressing the adverse effects of market
power
Figure 1: Illustration of the effect of a contractual measure, for a given level of
demand (an effective contractual measure)...................................................... 19
Figure 2: Illustration of the effect of a contractual measure for a given level of
demand (ineffective contractual measure-insufficient volume)..................... 20
Figure 3: Illustration of the effect of a contractual measure for a given level of
demand (ineffective contractual measure-highly uncompetitive market) .... 21
Figure 4: Winter marginal cost supply curve capacity by owner ............................ 26
Figure 5: Summer marginal cost supply curve capacity by owner ......................... 26
Figure 6: Demand adjustment for pumped storage ................................................. 29
Figure 7: Distribution of demand to be met by Belgium plant and the Southern
I/C, 2007 ............................................................................................................... 29
Figure 8: Winter installed capacity by owner ( +1,500MW CCGT) ...................... 32
Figure 9: Summer installed capacity by owner ( +1,500MW CCGT) ................... 32
Figure 10: Winter installed capacity by owner ( +2,500MW CCGT).................... 32
Figure 11: Summer installed capacity by owner ( +2,500MW CCGT) ................. 32
Figure 12: Illustration of demand adjustment from different interconnector flows
................................................................................................................................. 33
Figure 13: Distribution of demand to be met by Belgium plant and the Southern
I/C with 500 MW more exports to the North ................................................ 34
Figure 14: Distribution of demand to be met by Belgium plant and the Southern
I/C with 500 MW less exports to the North ................................................... 34
Figure 15: SPARK price outcomes for different volumes of contract cover for a
winter demand level of 12,000 MW .................................................................. 38
Figure 16: SPARK price outcomes for different volumes of contract cover for a
winter demand level of 12,800 MW .................................................................. 39
Tables & figures
iv
Frontier Economics | March 2006
Figure 17: Average market price under a given volume of contract cover
(assumed to be baseload) for the base case scenario ...................................... 40
Figure 18: Incremental benefit of an extra 200 MW of contract volume ............. 41
Figure 19: Central case product package (base case): Price outcome by demand
level......................................................................................................................... 43
Figure 20: Central case product package plus 1 GW peak (Variant 2): Price
outcome by demand level ................................................................................... 44
Figure 21: Average profit maximising price under a given volume of contract
cover (assumed baseload) for the entry scenarios compared to the base case
................................................................................................................................. 46
Figure 22: Incremental benefit of extra 200 MW of contract cover for the +1,500
MW new entry scenario....................................................................................... 47
Figure 23: Incremental benefit of extra 200 MW of contract cover for the +2,500
MW new entry scenario....................................................................................... 47
Table 1: Virtual interconnector power plants ........................................................... 27
Table 2: Virtual interconnector power plants ........................................................... 27
Table 3: Base case scenario product mixes and average prices .............................. 44
Table 4: New entry scenario product mixes and average prices............................. 48
Table 5: Northern export scenario product mixes and average prices.................. 49
Table 6: Inventory of installed assets in Belgium ..................................................... 61
Table 7: Installed capacity by plant type .................................................................... 61
Table 8: Fuel prices, taxes and carbon costs ............................................................. 63
Table 9: Availabilities by plant type ............................................................................ 64
Table 10: Run or river and wind power station seasonal duration ........................ 65
Tables & figures
1
Frontier Economics | March 2006
Executive summary
Previous studies of the Belgian wholesale market have indicated that Electrabel is
a dominant player on the Belgian market. The supposition that Electrabel is
indeed dominant is unlikely to be contentious, given the large share of generating
assets in Belgium that are under its ownership and control and the limited import
capacity. In the light of this, The CREG has asked Frontier Economics to advise
on what measures, or combination of measures, could be used to mitigate
Electrabel’s dominance and deliver a more competitive market.
While the title of this report is “Addressing the adverse effects of market power”
this study has not investigated whether there are, in practice, adverse affects
arising from Electrabel’s dominant position. In order to take a view on this one
would need to investigate Electrabel’s past and present conduct and this exercise
was not included in our scope of work. Instead, this study has focused on what
could be done to remove the ability or incentive that Electrabel might have to
increase prices above the competitive level.
Concerns as to the health of competition in a market can be categorised as either:
• static concerns related to the extent of competition given the existing
asset base; or
• dynamic concerns that stem from any barriers to entry that prevent or
hinder the competitive development of the market over time through new
entry.
While static concerns may be important in their own right, any adverse effects
associated with these may be exacerbated if there are also dynamic concerns,
preventing new entry from eroding such static problems as may exist.
In this study we have been asked to look at measures to address both static and
dynamic concerns. This report examines these measures from a purely economic
perspective and does not address any of the legal aspects pertinent to their
potential imposition, nor any detailed aspects of implementation.
STATIC CONCERNS - RANGE OF MITIGATING MEASURES
The Terms of Reference for this study invited us to investigate a wide range of
potential measures1, including:
• Virtual Power Plants (VPPs), which are options to call on power at a
predetermined exercise price;
1
The CREG asked us not to investigate measures where Electrabel was required to sell any of its
assets, requesting that we focus our attention on purely contractual measures instead of measures
involving physical divestment.
Executive summary
2
Frontier Economics | March 2006
• Power Purchase Agreements (PPAs) linking the sale of power to a
specific asset; and
• other long terms forward contracts.
Following the inception of the study, members of the steering group also invited
us to explore further possible measures, including contract swaps between
Electrabel and a generator with capacity outside of Belgium. Since any of these
measures could, in principle, be combined with any other in any quantity, there is
a potentially infinite array of packages of measures, leading to an unmanageably
large requirement for analysis.
However, as we describe in Section 2, in practice all of these contract forms have
a similar effect, i.e. they can decrease the extent to which a dominant firm would
benefit from a rise in price as a result of the exercise of market power in respect
of its remaining plant. As a result, these measures can be viewed as close
substitutes for one another from an economic perspective.
This equivalence means that we can conveniently model one type of instrument
but can then interpret the results as being a requirement for a level of contract
cover that can be met in a wide variety of ways. This aggregate requirement for
contract cover could be met in any way that Electrabel found convenient or
expedient. This equivalence is explored further in Section 3, where we look at
the three main parameters of contract cover: volume; shape; and duration.
STATIC CONCERNS – MODELLING RESULTS
We have used our SPARK model to analyse the possible effect on wholesale
market outcomes of the introduction of measures to curtail Electrabel’s market
power in the Belgian electricity market. Our modelling approach is discussed in
Section 4 and the results presented in Section 5.
Under our base case set of assumptions (described in detail in Section 4), we find
that approximately 8 GW of contract cover would be effective in mitigating
Electrabel’s incentive to exercise its market power. As noted, this aggregate
requirement for contract cover could be met in any way which Electrabel found
convenient or expedient. We also stress that the analysis we have conducted
assumes that Electrabel has no other long term contracts, or other positions that
might have the effect of such a contract (for example as a retail position with
regulated tariffs). Should Electrabel have existing positions, then these should be
netted off from any requirement identified in this report, in order to determine
the remaining contract cover that would be needed to address Electrabel’s
incentive to exercise market power.
In addition we have also analysed cases where we assume that there is entry
(incremental independent entry with a total volume of 1.5 GW and 2.5 GW) into
the Belgian market. In the face of such entry our analysis suggests that the
volume of contract cover required to mitigate Electrabel’s market power is
reduced. With entry of 1.5 GW we would conclude that total contract cover
could be reduced by at least 1 GW. With 2.5 GW, proportionately less contract
cover would be sufficient.
Executive summary
3
Frontier Economics | March 2006
Finally, we have analysed the effect of different patterns of import/export over
the Belgium-Netherlands interconnector. With increased exports (by ~500 MW)
to the Netherlands, contract cover of 8 GW is marginally less effective, but in
our view remains sufficient to mitigate Electrabel’s incentive to exercise market
power for the large majority of the time. In contrast, if exports to the
Netherlands were to fall (by ~500 MW) then 8 GW of contract cover would be
likely to be highly effective at mitigating Electrabel’s incentive to exercise market
power and the extent of the contract cover could potentially be reduced.
However, we recognise that this would be proportionately the largest capacity
release programme by some margin, not only in Europe but world wide. This
quantum should only be reached over enough time for markets and participants
to adjust.
Taking account of existing long term contract positions
Our analysis has not included any other contracts (or positions that might have
the economic effect of a contract) that Electrabel might have in place at present
or might put in place. If Electrabel has in place, or could commit to having put
in place, such contracts and/or positions, then they should be deducted from the
volumes that we report. For example, if Electrabel already has in place contracts
with customers of a reasonable duration at a fixed price for 1,000 MW of base
load power, then our recommendation would fall from 8 GW to 7 GW, with the
volume of baseload product required reduced by 1 GW to reflect this existing
contract position. Similarly, if Electrabel were to have in place a peak load
contract at a fixed price for 500 MW, this could be deducted from the required
volume of peak load product.
DYNAMIC CONCERNS – POLICY OPTIONS TO ADDRESS
SITE AVAILABILITY
One of the possible concerns that the CREG has identified as potentially
restricting competitive new entry is the pattern of ownership of suitable sites for
new generation. Specifically, the concern is that if all (or at least a very high
proportion of such suitable sites) are owned or controlled by Electrabel, then
Electrabel may be able to foreclose potential new entry by not making any of
these sites available to would be new entrants.
There is a range of possible policy measures that could theoretically be adopted
either to provide Electrabel with incentives to release sites voluntarily or to
require it to do so. These include:
• enforced (or negotiated) release of sites;
• a requirement to auction vacant sites;
• revising the licensing regime to limit the scope for new build by
Electrabel;
Executive summary
4
Frontier Economics | March 2006
• resolution of existing issues of market power, thus possibly reducing the
benefit of limiting entry;
• putting in place a tax on vacant sites; and
• using other methods to stimulate entry, such as offering centrally held
contracts for the provision of new capacity (as has been done in Ireland
for example) although this may formally need security of supply to be an
issue.
In order to assess which of these options might be preferred it would be
necessary to conduct detailed analysis based on a range of data. This has not
been possible as we have been unable to obtain the required information within
the scope of this assignment.
We understand through discussion with the CREG that Electrabel may have
agreed with the Minister to release sites that would allow the construction of
1,500 MW of new entry. This understanding has prompted one of the
sensitivities that we have modelled, as described above. However, we have not
been provided with any details of this programme. Similarly, we have not been
provided with any of the data that might have informed the design this
arrangement.
If a release of sites has been agreed with Electrabel and if this route remains open
in the future, then it would seem to represent a useful approach to ensuring site
release over time. We would advise that the effectiveness of the first step in this
policy be assessed before any decision is taken as to whether further site release
would be appropriate.
Executive summary
5
Frontier Economics | March 2006
1 Introduction
Previous studies of the Belgian wholesale market have indicated that Electrabel is
a dominant player on the Belgian market. The supposition that Electrabel is
indeed dominant is unlikely to be a contentious, given the large share of
generating assets in Belgium that are under its ownership and control, and the
limitation that exists on import capacity. In the light of this, The CREG has
asked Frontier Economics to advise on what measures, or combination of
measures, could be used to mitigate Electrabel’s dominance and deliver a more
competitive market. In this report we present our findings on these topics, along
with the methodology and data used in the analysis that led us to these
conclusions.
While the title of our terms of reference is “Addressing the adverse effects of
market power” this study has not investigated whether there are, in practice,
adverse affects arising from Electrabel’s dominant position. In order to take a
view on this one would need to investigate Electrabel’s past and present conduct
and this exercise was not included in our scope of work. Instead, this study has
focused on what could be done to remove the ability and/or incentive that
Electrabel might have to increase prices above the competitive level.
Concerns as to the health of competition in a market can be categorised as either:
• static concerns related to the extent of competition given the existing
asset base; or
• dynamic concerns that stem from any barriers to entry that prevent or
hinder the competitive development of the market over time through new
entry.
While static concerns may be important in their own right, any adverse effects
associated with these may be exacerbated if there are also dynamic concerns,
preventing new entry from eroding any static problems. In this study we have
been asked to look at measures to address both static and dynamic concerns.
This report examines these measures from a purely economic perspective and
does not address any of the legal aspects pertinent to their potential imposition,
nor any detailed aspects of implementation.
Our report comprises the following sections.
| Section 2 identifies a range of contractual measures that have been mooted
as having the potential to address market power and provides definitions for
these. It then analyses from first principles the way in which contractual
measures can affect the ability or incentive of a dominant player to exercise
market power and, in the light of this analysis, reviews qualitatively the
potential effects of the identified measures. This section is completed by a
discussion of the extent to which non contractual measures could play a role
in mitigating market power and the possible interaction between contractual
and non contractual measures.
Introduction
6
Frontier Economics | March 2006
| Section 3 having established the major role that contractual remedies would
need to play, this section addresses the parameters that define the required
contractual coverage (in essence volume, shape and duration).
| Section 4 provides an overview of the data we have used in our analysis,
together with the scenarios we have modelled. We also describe the steps
that our analysis has followed in order to identify the appropriate volume and
shape that any package of measures should have.
| Section 5 presents the results of our analysis with our conclusions on the
parameters of contract cover needed to address static market power
concerns.
| Section 6 discusses potential barriers to entry and in particular potential
problems arising from the difficulty new entrants may have in procuring sites
on which to build generating assets.
We also include two annexes:
| Annex 1 provides some further detail on the model we have used to analyse
static market concerns, SPARK.
| Annex 2 provides further detail on our data assumptions.
Introduction
7
Frontier Economics | March 2006
2 Static concerns - range of mitigating
measures
The Terms of Reference for this study invited us to investigate a wide range of
potential measures2, including:
• Virtual Power Plants (VPPs), which are options to call on power at a
predetermined exercise price;
• Power Purchase Agreements (PPAs) linking the sale of power to a
specific asset; and
• other long terms forward contracts3.
Subsequent to the inception of the study, members of the steering group also
invited us to explore further possible measures, including contract swaps between
Electrabel and another international player.
Since any of these measures could, in principle, be combined with any other in
any mix of quantities, there is a potentially infinite array of packages of measures,
leading to an unmanageably large requirement for analysis. However, in practice,
many of these potential measures have economic effects which are very similar
indeed, and in many cases identical. This section of our report sets out our
rationale for this and the consequences of this for both the study and our
conclusions. We also consider in this section how regulatory measures might
reinforce contractual measures, focussing on one specific proposal raised by the
steering group. However, we begin by providing a definition of the main
contractual measures that we have been asked to consider.
2.1
THE RANGE OF POSSIBLE CONTRACTUAL MEASURES
Before proceeding with an economic analysis of the properties and effect of
contractual measures, we first provide a brief definition of each measure. As we
describe in more detail in Section 3, each contractual measure has a number of
parameters that can be set to modify the impact of the contract(s). We provide a
brief introduction to those parameters here.
2
The CREG asked us not to investigate measures where Electrabel was required to sell any of its
assets, requesting that we focus our attention on purely contractual measures instead of measures
involving physical divestment.
3
We note that long term contracts can, in certain circumstances, give rise to competition concerns.
However, contractual measures would only function as intended if they are long term in nature. As
is explained in a number of places in this report, any ability that Electrabel might have to increase
prices for long term contracts beyond competitive levels should be undermined as long as the total
package of measures to which it is subjected is substantial enough.
Static concerns - range of mitigating measures
8
Frontier Economics | March 2006
2.1.1 VPPs
VPPs are option contracts. The purchaser of a VPP pays a fixed sum, usually
monthly, which confers the right, but not the obligation, to buy a specified
volume of power at a specified exercise price for the duration of the contract.
The seller of a VPP must deliver the required volume of power for the hours
during which the VPP is exercised. A VPP is typically not linked to any physical
asset. The seller of a VPP can therefore provide power sold under a VPP from
any source.
A VPP will be exercised whenever the market price (or expected market price) of
power rises above the exercise price specified in the VPP. If a VPP is exercised
then it will be typically be exercised for its full volume. However, buyers may
exercise less than the full volume. The value of the VPP to the purchaser when it
is exercised is equal to the volume multiplied by the difference between the
market price and the exercise price.
If the exercise price of a VPP is low enough, the VPP is certain to be exercised in
all periods. In this case, the VPP will have the same effect as a baseload contract
of the same duration for the delivery of a fixed volume at a predetermined price.
If a generator sells a VPP contract, this will have the effect of reducing the profit
of that generator from any given increase in the market price at times when the
market price exceeds the VPP exercise price. In the absence of a VPP, the
generator would receive the benefit of the price rise on the corresponding
volume it is able to sell into the market. With a VPP in place, the generator
receives the benefit of the price rise only on a smaller volume. .
2.1.2 PPAs
In this report, when we use the term PPA, we refer to a contract that effectively
provides physical control of a station to the purchaser through a carefully
specified set of terms and conditions. PPAs of this kind are linked to a specified
physical asset. PPAs are most common in developing countries where there are a
small number of assets that are typically all under contract to a central power
procurer, or during the early stages of liberalisation.
The purchaser of a PPA has the right, but not the obligation, to take power from
the station, usually at some predetermined price linked to the marginal price of
the station. In this sense, a PPA resembles a VPP – it has a volume (i.e. the size
of the station) and an exercise price (as determined in the contract).
However, since a PPA is linked to a real asset, these contracts typically include
terms setting out how the volume of power called is allowed to vary over the
course of a typical day. Such terms are linked to the physical ramps rates of the
asset to which the PPA is linked. PPAs will usually impose a cost on the
purchaser for each time the generating unit is stopped and restarted. Similarly,
PPA contracts also need to deal with periods where the station is unavailable due
to either planned or unplanned maintenance.
Static concerns - range of mitigating measures
9
Frontier Economics | March 2006
In addition to paying a marginal cost when the station is called, the purchaser of
a PPA will also pay a capacity payment to the seller of the PPA, linked to the
availability of the station.
2.1.3 Fixed long term contracts
A fixed long term contract is an agreement that determines the sale of a given
volume of firm power from the seller to the buyer in exchange for a fixed and
pre-determined price4. Contracts of this kind are not associated with any
particular physical asset. As such the power could be provided by any asset in a
portfolio.
Typically, fixed long term contracts are either baseload, in which case deliveries
are continuous at the same level for every hour of the day, or peak load, in which
case deliveries occur only during specified hours, when demand is typically
highest. The hours of the day covered by peak load contract can vary according
to the convention of the region (e.g. 7am to 7pm).
A fixed long term contract of this kind has an effect similar to that of a VPP
when it is called, i.e. the seller of the long term contract no longer benefits from
price rises on the volume of power covered by the contract. While a VPP is
given shape through the level of its exercise price, which will determine the
periods when it is likely to be called, a fixed long term contract is given shape
through the hours of the day when it operates. With appropriate calibration, the
effect of either of these two contract forms can be replicated by the other.
2.1.4 Contract swaps
The final form of contractual measure that we have been asked to consider is a
contract swap. A remedy of this form involves the exchange of power in one
country with that of power in another country with no price attached. For
example, in theory Electrabel could agree to a swap with EDF, where EDF
receives the right to some agreed volume of power in Belgium while Electrabel
receives the same volume of power in France.
In principle, a swap creates an obligation on each party to provide power in their
own region to another party for a value which the party cannot influence. In
return, they are given rights to a comparable volume of power in the other
region. Again, the swap could be base load or peak load, as with a fixed long
term contract. We do not envisage swap contracts being linked to specific
physical assets, although this is, in principle, possible.
The effect of a swap is therefore, likely to be similar to a simple long term fixed
contract.
4
In principle the agreed price could be indexed to some exogenous price marker. As long as this
price marker was unrelated to the wholesale price of electricity, this would still serve the same
purpose as a fixed price.
Static concerns - range of mitigating measures
10 Frontier Economics | March 2006
2.2
RATIONALE FOR CONTRACTUAL MEASURES
Contracts can be used to create obligations such that Electrabel has a reduced
incentive to try to raise the ‘market’ price of power.
A dominant firm that has the ability to withdraw plant and raise prices, faces a
trade-off. In withdrawing plant, the dominant firm loses any profit that it would
have made on the plant that is withdrawn (we can think of this as the cost of
exercising market power) but increases the profit from plant that continues to
supply power (we can think of this as the benefit of exercising market power). It
will have the incentive to withdraw plant if the extra profit from remaining plant
exceeds the lost profit on the withdrawn plant (i.e. if the benefits of exercising
market power exceed the costs).
Contracts can be used to alter the nature of this trade-off and hence the
circumstances in which it is profitable to exercise market power.
In principle, a contract could be designed either to:
• increase the profit lost in relation to the withdrawn plant (i.e. increase the
cost of exercising market power); or
• decrease the extent to which the firm benefits from the rise in price of
power in respect of its remaining plant (i.e. reduce the benefit of
exercising market power).
For reasons which will become clear below, the latter has always been found to
be much more practicable than the former. However, for completeness, we
discuss both here.
2.3
POSSIBLE CONTRACTUAL MEASURES
2.3.1 Contracts to increase the profit lost on withdrawn plant
Normally, when a plant is withdrawn, there are savings from not operating the
plant. In theory a contract could:
• decrease the cost reduction from not running the plant; or
• introduce an explicit penalty for not running when it ‘should’.
When an asset is withdrawn from the market, the owner of that asset no longer
incurs the variable cost of operation while the asset is withdrawn. In principle, a
contract could be designed that reduces this benefit of withdrawal, thereby
increasing the cost of exercising market power. Hypothetically, one could offer
the firm rights to subsidised fuel only exercisable if it used the fuel to generate
from the plant. However, such subsidies would be neither desirable nor, in all
probability, legal.
Static concerns - range of mitigating measures
11 Frontier Economics | March 2006
Introducing a contractual penalty for not generating, which would also act to
increase the cost of exercising market power, might be more feasible but there
are many problems. In particular:
• the penalty would have to apply only at the times when that plant should
reasonably run on the system. Hence some mechanism would be needed
to determine when the plant was wanted and when it was not; and
• the penalty would need to be fair in the sense that it did not unreasonably
penalise the operator for planned and forced outages that would arise if
the plant were not controlled by a dominant firm.
Neither of these are easy conditions to meet efficiently. Doing so could well
mean moderating any penalty to a level at which it might not alter incentives
enough to see a marked reduction in the incentive to exercise market power. It is
also not clear who should be the counterparty for a contract based solely on a
penalty for non availability.
Penalties for non availability could be included within a classical station specific
PPA contract where the buyer pays a capacity price and an exercise price based
upon its nominations, receiving a penalty payment if electricity is not delivered.
However, such contracts are most suitable in electricity systems based around the
single buyer model where the buyer is worse off when delivery is not made from
the specific station.
Regions that operate a single buyer model typically have a central power procurer
that has a suite of contracts with a number of different generators in that region.
The power procurer therefore decides how these stations should be dispatched in
order to meet the profile of demand over the day. In such circumstances, the
power procurer is, effectively, playing the role of system operator. The power
procurer needs to be sure that when a specific station is called upon to run then
that station does actually produce the required power, otherwise it will need to
take close to real time actions to ensure system stability and integrity. In order to
provide a financial incentive for stations to run when called, contracts in
environments such as this typically include a penalty clause that creates a cost for
the generator when it fails to generate when called, i.e. a contract clause that
increases the profit lost on withdrawn plant.
The purchasers of power contracts in more developed regions, including
Belgium, are typically retailers (or large consumers) who are completely
indifferent as to where the power they consume is generated as long as it is made
available. In such circumstances there is no obvious rationale for including a
station specific penalty clause for failure to generate from a specific site. The
purchaser of the contract will be happy to receive power from any source as long
as it is delivered. This is discussed further in the context of station specific PPAs
in Section 2.4.1 below.
Static concerns - range of mitigating measures
12 Frontier Economics | March 2006
2.3.2 Contracts to decrease the profit from a price rise
Any contract for the sale of electricity at a price that is determined independently
from the ‘market’ price will reduce the incentive of the seller to raise prices
because the seller benefits from the higher prices only in respect of the
proportion of its sales not covered by such contracts.
There is an infinite variety of contracts that could achieve this result, including all
of the contractual forms identified above in Section 2.1.
2.3.3 Summary
Given the discussion in the preceding subsections, it is seems clear that remedies
focused on reducing the benefit of exercising market power, rather than those
that increase the cost of exercising market power, are likely to be easier to design,
implement and monitor. This is consistent with what we observe in those
regions where such remedies have been imposed, where contractual remedies
such as VPPs have typically been used.
2.4
CHOICE OF CONTRACTUAL FORM
In Section 2.3 we have identified what we would like any proposed measure to
achieve in terms of its affect on the incentives of any large player on which it
might be imposed. When we consider the definitions provided in 2.1, it seems
clear that any of the proposed measures would have a broadly similar effect.
Here we explore further whether there are substantive differences between the
measures we have described from an economic perspective.
2.4.1 Measures linked to physical assets
Notionally, PPAs might appear to have the same effect as the other possible
contractual remedies, together with the potential additional advantage that they
could help to inhibit withdrawal of specific plant. If this meant that withdrawal
to exercise market power could only be effected by withdrawing lower cost plant,
it could increase the loss of profit on withdrawal and thereby have an additional
beneficial influence on incentives.
However, owing to their complexity, PPAs would be more costly to administer
and it is not at all clear to us that purchasers would have either an interest in or
the ability to police where the power comes from. The purchaser will request
power from the station and expect the seller to make the corresponding block
exchange notification to Elia. The purchaser will have no further interest or
knowledge as to where the power actually came from or indeed whether the
seller really produced the power at all. If there is no economic incentive to
enforce an aspect of the contract, enforcement is unlikely to materialise.
In practice, therefore, we believe that station specific PPAs would have little or
no real advantage over the other possible contractual forms and they would be
administratively more expensive. Similarly, it is likely that a PPA would have less
Static concerns - range of mitigating measures
13 Frontier Economics | March 2006
appeal than some other contractual forms to a potential purchaser. For example,
the purchaser of a PPA needs to deal with periods when the plant is unavailable,
whereas other contractual forms are typically for firm power. Given this analysis,
it is not clear that a PPA is necessarily a better measure from the perspective of a
market regulator or from the perspective of a potential purchaser.
2.4.2 Firm contracts versus option contracts
The key differentiating feature of a VPP is that it is an option rather than a fixed
contract.
If the exercise price of a VPP is such that it is always called, there is no economic
and little practical difference between such VPPs and contracts for a baseload
strip. A contract for a baseload strip would be administratively simpler, at least if
there were no other VPPs. However, even that marginal advantage is eroded
substantially if there are going to be other VPPs with higher exercise prices
anyway.
The aim should be to identify the set of instruments to be imposed that is
proportionate in the sense that it meets the criterion of being the minimum
necessary to address market power. Given this aim, a significant proportion of
the requisite instruments are likely to be targeted at durations less than baseload.
Market power is always likely to be more significant when demand is high and
the supply/demand balance tight.
The issue is whether it is better to meet this requirement with profiles that are
fixed ex ante or with VPPs which are called when the market price threatens to
rise above their exercise price. VPPs have the advantage that they will always
apply if they are needed. An ex ante guess at when instruments are needed may
target most of the occasions correctly. However, there are likely to be times
when an ex ante profile means that there is insufficient contract cover, unless the
profile is made so broad/large that there are many occasions on which there is
more contract cover than is necessary to meet the desired objective.
In short, VPPs do a better job of applying when needed and being irrelevant
when they are not. The key causes of market power are either that a generator is
pivotal (demand cannot be met without them) or there is a significant step in the
marginal cost plant stack where a modest withdrawal sends the price up to the
next level. VPPs are better than firm power contracts as instruments for targeting
these circumstances, especially steps in the marginal cost stack.
Therefore, while a fixed contract (whether a PPA, a simple contract or a swap)
could be tailored to have a shape which closely follows the effective shape of a
VPP, there will inevitably be some periods where fixed contracts are either
disproportionately large, or inappropriately low.
It is appropriate to mention at this point that some members of the steering
group have expressed the view that the previous VPPs in Belgium have been
ineffective. One member said ‘ VPPs were just another way for Electrabel to sell
Static concerns - range of mitigating measures
14 Frontier Economics | March 2006
power expensively’. While it is beyond the scope of this assignment to do a
detailed review of the previous VPP releases, we believe that it is at least very
plausible that the releases will have had minimal effect because the volume of
them was ineffective, not because there is any inherent fault in the nature of the
instruments. Our modelling results reinforce this view.
Once market participants understand that conditions with the VPPs in place will
be competitive, there will be no incentive for them to pay more than the
competitive market price to acquire them.
2.5
REGULATORY MEASURES TO SUPPORT
CONTRACTUAL MEASURES
The CREG has also asked us to consider whether regulatory measures could be
implemented, in addition to the contractual measures described above, in order
to enhance the effectiveness of any particular package of measures. Specifically,
the CREG has asked us to consider an arrangement where generating assets that
have been fully depreciated are required to make their output available at a
regulated price determined by the CREG.
If this arrangement were put into effect, it would require Electrabel to sell some
agreed volume of output from some of its assets at a fixed price. Given this, it is
clear that an arrangement of this kind would have an identical effect to a long
term contract.
One complication that an arrangement of this kind would give rise to is deciding
who should be allowed to purchase power at the agreed lower price (and in turn
who should have to pay the higher prevailing market price). The most obvious
mechanism for allocating these imposed regulated contracts would be to auction
them, as one would any other contractually imposed release programme. If this
were the case, then the effect of a regulatory measure of this kind would precisely
replicate any other imposed long term contract and would therefore be captured
by the discussion above (and also by the analysis we present later in this report).
2.6
CONTRACTUAL MEASURES AND SECURITY OF
SUPPLY
We have been asked to consider whether the introduction of a package of
contractual measures to mitigate market power could give rise to a concern with
regard to security of supply.
Concerns over security of supply are typically centred on whether there is
sufficient physical capacity to meet demand, or whether there is an existing or
future danger of a shortage of physical capacity occurring. Since none of the
contractual forms that we consider would be likely to result in the withdrawal of
capacity (more likely the opposite) we do not believe that any imposed
contractual measure(s) could give rise to any concern over security of supply.
Static concerns - range of mitigating measures
15 Frontier Economics | March 2006
2.7
TREATMENT OF EXISTING CONTRACTS
As we describe below, in all our analysis we have assumed that Electrabel has no
contracts in place at present. Any contracts that Electrabel does have in should
therefore be netted off from the total volume that we identify as being sufficient
to mitigate any incentive it might have to exercise its market power. This gives
rise to a number of issues that warrant discussion.
| What characteristics would a contract or similar position need to have in
order to be deducted from the total proposed volume? and
| In particular, how should contracts between Electrabel and ECS be treated?
We have also been asked to consider how existing contracts should be treated if
they were struck at relatively high prices.
We address each point in turn.
2.7.1 Characteristics required
Any contract which, relative to the contract not being there, reduces the profit
that would be earned by Electrabel when the market price rises could be a
candidate for subtraction from the proposed volume. The most likely contract
form (based on typical contracting structures in the power sector) would be a
contract that fixes, in advance, the price that Electrabel would receive for a given
volume of power. A standard forward contract (with a large consumer for
example) would be an obvious form of a contract that could be subtracted from
the proposed volume. A contract swap, had Electrabel been persuaded to accept
one, would also be suitable for subtraction from the headline quantity, since
Electrabel’s revenues from the power covered by the swap would be decoupled
from short run prices in Belgium.
Contracts where the strike price is indexed according to some variable, or basket
of variables, could qualify for subtraction but the terms of indexation would need
to be scrutinised carefully. In particular, contracts that are indexed according to
changes in headline markers of electricity prices in Belgium would most likely not
qualify. This follows as, depending on the detail of the terms of the contract
Electrabel could profit indirectly from increasing prices in the electricity market,
as this would in turn increase revenue from the contract.
Contracts would also need to be assessed for their “shape”, e.g. whether they are
baseload, peak load. We discuss contract shape further in Section 3.
The duration of contracts is also relevant. Contracts need to be long enough
such that the potential effects on the price achieved in their renewal does not
become a substantial influence on the shorter term behaviour of Electrabel.
There is no hard and fast rule in relation to this. We suggest that contracts which
are three years or more should be recognised fully. The volume of contracts with
shorter durations might be taken account on a less than 1 MW for 1MW basis.
Static concerns - range of mitigating measures
16 Frontier Economics | March 2006
While it is possible to write down simple criteria for contracts that are likely to
have the required effect, some scrutiny of any contract that might qualify will
inevitably be needed. Therefore, if the CREG proceeds to implementation of
some package of measure then we would strongly recommend that the CREG
exercises its own judgement on whether any given contract or position should be
netted off from the total volume imposed
2.7.2 Contracts between Electrabel and ECS
On the basis of our present understanding of the relationship between Electrabel
and ECS, together with the prevailing regulatory environment for retail
customers in Belgium, we do not believe that contracts between Electrabel and
ECS should be netted off from the total volume proposed.
This follows since:
• Electrabel and ECS are under common ownership and we must assume
that they maximise profits jointly; and
• all consumers will soon (from 2007) operate in liberalised retail markets
where tariffs are determined by competition and not by regulation.
The first of these points implies that the price in any contract struck between
Electrabel and ECS is essentially equivalent to a transfer price between business
units. If Electrabel were able to raise the wholesale market price, this would
become the input price to all retailers. As a consequence, the retail price will rise
in response to higher wholesale prices – the second point above confirms that
prices will not be fixed and would therefore be free to vary in this way. While
Electrabel will not profit from this directly, if a contract exists with ECS at a
fixed price, it will nonetheless do so indirectly as a higher price will be achieved
on retail sales by ECS. Whether those profits are booked by ECS on retail sales
or by Electrabel on wholesale sales is irrelevant if Electrabel and ECS jointly
maximise profits. Any contract between Electrabel and ECS would therefore
have no affect on incentives Electrabel might have to exercise market power in
the wholesale market.
2.7.3 Treatment of existing contracts if struck above market
prices
We have in addition been asked to consider the position of existing contracts
were these to have been struck at a price exceeding the current market price.
In terms of its impact on future behaviour, the price at which a contract is struck
is irrelevant. In order for a contract to be effective in mitigating market power all
that is required is that for some volume of power, the price is fixed in advance,
i.e. that it has the facets described above. Contracts struck at “high” prices
should therefore qualify for subtraction provided they have the facets described
above.
Static concerns - range of mitigating measures
17 Frontier Economics | March 2006
Of course, if the price written in the contract is “high” then this implies that the
counterparty to Electrabel receives no benefit from lower wholesale market
prices until it is able to renegotiate the contract or the contract expires.
However, this has no impact on the incentives that Electrabel has to exercise, or
not, any market power that it might possess.
If the original contract was struck on a fair basis that did not reflect any abuse of
market power, then in our view it would be appropriate to let it run its course. If
on the other hand it were established that a contract was either anti competitive
in duration or was set at a price reflecting abuse, cancellation of the contract
would be feasible. However, we note that simply being in a position where the
contract appears to be ‘out of the money’ is not evidence of abuse.
2.8
CONCLUSIONS
VPPs have slightly better characteristics of applying when needed and not when
they are not. Other things being equal they are probably the preferred
instrument. However, all contracts serve a similar purpose and any set of
contracts which provide a similar profile over time will have essentially the same
effect in mitigating market power. With an appropriate definition of the
parameters, any of the contractual forms we have identified could produce a
broadly similar result. As a result, it follows that any of the measures described,
with parameters set appropriately, could be combined together in almost any mix
of volumes to achieve the similar results.
This equivalence means that we can conveniently model one type of instrument
but can then interpret the results as being a requirement for a level of contract
cover that can be met in a wide variety of ways. This aggregate requirement for
contract cover could be met in any way which Electrabel found convenient or
expedient. As the discussion in this section has illustrated, there is little or no
substantive difference between the different contractual forms that we have been
asked to consider. As a result, any combination of these contractual forms could
have the desired result. Given this, there is no obvious reason to limit the choice
that Electrabel might have in how to put together a package comprising the
required volume. If one package of measures is more desirable to Electrabel and
still delivers the outcome desired then it would seem unreasonable to
unnecessarily restrict Electrabel from choosing its preferred package.
One concern raised with us is that Electrabel might choose to meet its obligation
using long term bilateral contracts negotiated under its own terms. The issue of
transparency of pricing in imposed contracts is essentially one of implementation.
However, as long as market participants knew that Electrabel was being required
to comply with a programme of a known size, they would (and indeed should)
embody this information in their negotiations with Electrabel and therefore be
able to achieve a contract price consistent with their expectation of market
outcomes after the imposition of contractual measures.
Static concerns - range of mitigating measures
19 Frontier Economics | March 2006
3 The parameters of contractual measures
In Section 2 we set out the main features of all of the typical forms of contractual
measure, including all of those that we have been asked to analyse here. In this
section we describe the key parameters of these contractual measures.
3.1
INTRODUCTION
Contractual measures diminish the incentives of a large player to increase prices.
The precise effect of a contractual measure diminishing a large player’s incentive
to exercise market power will, however, depend on the actual characteristics of
the contracts in terms of volumes, shape (i.e. the periods during which deliveries
will normally take place) and duration.
In the following sub-sections we describe in more detail the impact of different
contract volumes, the shape of the contracts and the duration of contracts on a
large player’s market power.
3.2
VOLUME AND EXERCISE PRICE
3.2.1 Effects of different volumes
How effective a given volume of contract cover is in creating competitive
constraints on pricing depends principally on the degree of market power already
existing in the market (and also on the details of the cost structure of generation).
As an illustration, suppose that a large player were in a position in which
increasing its price always resulted in increased profits. Then assume that a
contractual measure is introduced, for illustrative purposes we initially assume
that the contractual remedy is designed so that it is always in effect (e.g. a fixed
long term contract or a VPP with a very low exercise price). The relationship
between the large player’s profits and its bid price might be like this:
No contractual measure
profit
With contractual measure
profit
Profit-maximising
market price
Profit-maximising
market price
market price
market price
Figure 1: Illustration of the effect of a contractual measure, for a given level of demand
(an effective contractual measure)
The parameters of contractual measures
20 Frontier Economics | March 2006
The contractual measure reduces the additional profits resulting from a higher
market price. The slope of the profit line is less, “tipping” the profit line down.
In this case, the contractual remedy is effective because the change in the slope
of the profit line results in a local maximum profit point at which the market
price is lower than in the case without the contractual measure.
However, the contractual measure will be ineffective in constraining pricing if no
such maximum is produced. This may occur if the volume of contractual
measure is insufficient in comparison to the degree of market power that exists.
For example, suppose that in the same market conditions as those illustrated
above, a smaller volume of contractual measure were implemented. Then the
contractual measure may not result in a reduction in market price, as illustrated in
Figure 2:
No contractual measure
profit
With contractual measure
profit
Profit-maximising
market price
market price
-
Profit maximising
market price
market price
Figure 2: Illustration of the effect of a contractual measure for a given level of demand
(ineffective contractual measure-insufficient volume)
The contractual measure shown in the figure above would not reduce market
prices. Even though the large player gains less from increasing its bid with the
contractual measure in place than without, the point of maximum profit is still
strictly increasing with the market price.
How much contract cover is sufficient depends on the degree of competition
already existing in the market. If one firm has very high market shares, then it
will lose almost no sales as it increases the market price beyond the point at
which all competing capacity is in use. Profits then simply increase linearly with
the market price. We illustrate this situation in Figure 3:
The parameters of contractual measures
21 Frontier Economics | March 2006
No contractual measure
profit
With contractual measure
profit
Profit-maximising
market price
market price
Profit maximising
market price
market price
Figure 3: Illustration of the effect of a contractual measure for a given level of demand
(ineffective contractual measure-highly uncompetitive market)
3.2.2 Effects of shape
The amount of contractual cover that might be required to mitigate market
power is likely to vary over the course of a day/week/year. In periods of high
demand it is likely that a larger volume of contract cover will be required, since in
such periods the balance between aggregate demand and supply is likely to be
tight, making the market more prone to the exercise of market power. In periods
of low demand, less cover is likely to be required. Unless measures are shaped to
match this requirement, there will be periods of time when the level of contract
cover is either inadequate with no effect on market power, or disproportionately
large.
Contractual measures can be calibrated in order to provide them with shape. For
example, a set of VPP products can be put in place at different exercise prices.
VPPs with low exercise prices will always be called and will always be in effect.
VPPs with higher exercise prices will only be called in periods where the market
price is sufficiently high, which is likely to be in a subset of periods of higher
demand. In this way a VPP programme can be designed to deliver the
appropriate level of cover over time, although practical considerations are likely
to limit the number of different product types that could be included in any
programme and hence limit the extent to which any programme can precisely
match the shape of cover required. This also demonstrates that the effect of a
VPP with a lower exercise price will be greater than the effect of a VPP with a
higher exercise price.
The parameters of contractual measures
22 Frontier Economics | March 2006
Just as a VPP programme can be constructed with different exercise prices to
provide the desired shape, a set of long term contracts could be sold that operate
over different periods of the day, or over different seasons, to provide the
required level of cover at the appropriate time. Again, there will be practical
limitations on the degree to which a programme can be perfectly matched to
requirements. As described in Section 2, such difficulties are likely to be more
substantive with fixed contracts than with VPPs, since the exercise of VPPs will
better track prevailing market conditions.
While baseload products are at least as effective at mitigating market power, they
are also, obviously, a more onerous burden for the party required to sell them
than products with more limited time coverage. If all of the capacity of any
proposed package of contractual measures were baseload, it is likely that there
would be periods where the volume of cover would be excessive – a smaller
volume would have the desired effect. By using VPP products at different
exercise prices, or long term contracts that are in force over only certain periods
of the day, one can create “shape” in the release programme such that, as far as
possible, the volume of cover in force in any given period is proportionate to the
level required to have the desired competitive effect. In this way the effect of a
programme of contractual measures can be managed such that it is generally not
materially more than is required. In any event, some buyers may prefer some
mid-merit or peaking products rather than solely baseload, to fit better the load
profile of their customers. Therefore, including such products in any programme
is likely to increase participation by potential purchasers.
3.3
DURATION
In general, longer duration products are likely to have a greater competitive effect
on market prices. An extreme example illustrates this point. Suppose that some
form of contractual measure were auctioned in durations of one hour. This
should surely produce almost exactly the same effect as a spot market operating
with no mandated release programme. The reason is that a large player is likely
(for significant volumes of contract) to become as interested in the auction
proceeds from reselling the imposed contract as in profitability in the spot
market itself, and may therefore be prepared to bid prices up in that market even
when to do so does not maximise profits from spot market sales. The
contractual measure’s effect in diminishing the incentive to bid price up in the
spot market is ineffective, because the profits from the next hour no longer
matter as much as the effect that a higher price in the next hour will have on the
future price at which the contractual measure will be sold.
At the other extreme, if a contract is in force for ever (no renegotiation of terms
at any point) then the large player has no interest in manipulating the wholesale
market to influence the proceeds of subsequent auctions, because there will be
no subsequent auctions. If a sufficient volume of the contractual measure is sold
to create a competitive market structure, then, a large player acting to maximise
its profits will behave competitively in the wholesale market.
The parameters of contractual measures
23 Frontier Economics | March 2006
As a general proposition, therefore, longer durations lead to more competitive
market outcomes. It is therefore possible to argue that only longer term
products should be sold and that secondary markets will arise to repackage those
products into shorter-term instruments as required.
However, there is no real demand for long term products so a requirement to sell
longer term products could lead to unreasonably low proceeds and hence could
justifiably be viewed as partial expropriation. In most markets only a small
proportion of contracts have a duration of more than 1-2 years. 3 years would
therefore seem to be the maximum duration that should be imposed and a mix
of durations up to 3 years would seem to be appropriate.
The parameters of contractual measures
25 Frontier Economics | March 2006
4 Data, methodology and scenarios
In this section we describe our modelling methodology and present some key
data points.
4.1
MODELLING MARKET POWER USING SPARK
We have used our SPARK model to analyse the possible effect on wholesale
market outcomes of the introduction of measure to curtail Electrabel’s market
power in the Belgian electricity market.
SPARK consists of two modules, one of which identifies profit-maximising
strategic behaviour (the SPARK gaming module), while the other simulates plant
dispatch given that strategic behaviour (the SPARK dispatch module). In this
assignment we have only use the gaming module.
Using the SPARK gaming module, given the portfolios of plant present on the
system, their capacities and costs, we can analyse what incentives players have to
withdraw capacity or mark up the price at which they bid their plant.
Our approach to analysing the scope for strategic interaction is rooted in game
theory. SPARK simply enables us to make the necessary calculations to identify
the range of possible non-cooperative Nash equilibria efficiently and robustly.
The gaming module of our SPARK model requires six key inputs:
• available assets and their capacities;
• ownership of the available assets, together with an assumption on
whether each owner might behave strategically;
• asset efficiencies and availabilities;
• fuel prices, taxes and carbon costs;
• interconnector capacities; and
• demand.
In a following subsection, we provide an overview of the data we used; a more
detailed description of the data used can be found in Annexe 1.
Data, methodology and scenarios
26 Frontier Economics | March 2006
4.2
OVERVIEW OF DATA
4.2.1 Belgian Asset register
Central to our analysis is an estimate of the marginal cost of generation for each
generating unit attached to the Belgian system. A full list of the assets installed in
Belgium is included in this confidential version of the report in Annex 2. In
Figure 4 we illustrate the marginal cost curve for winter and in Figure 5 the same
curve for Summer. Differences between the two curves arise from seasonal
differences in availability (with maintenance typically planned for Summer
periods) and fuel prices. Assumed fuel prices, efficiencies, availability rates etc
are asset out in Annexe 2. Data on Belgian generating assets were provided by
Elia.
[Contains confidential information]
Figure 4: Winter marginal cost supply curve capacity by owner
Source: Elia and Frontier estimates
[Contains confidential information]
Figure 5: Summer marginal cost supply curve capacity by owner
Source: Elia and Frontier estimates
Definition of marginal cost
Spot price outcomes in wholesale electricity markets are determined by the
interaction of supply and demand on the day. Such short run analysis should
therefore be conducted by analysing short run costs, specifically short run
marginal costs. By this we mean the costs directly incurred in providing one
extra unit of power. This will vary from station to station and will be determined
by input fuel costs (e.g. gas prices for gas fired stations, plus carbon allowance
prices) at the time, the conversion efficiency of the station, variable operations
and maintenance costs5.
4.2.2 Interconnectors
In electricity markets, the capacity of interconnectors can be a crucial
determinant of prices in any region. If the interconnector capacity between two
countries, for example, is very large, relative to market size, and is never
constrained, that is its use is always below its maximum available capacity, then
prices in both regions will equalize (except potentially for a small loss factor).
5
Long run marginal costs would only become relevant when considering entry decisions that might
be taken. However, given relatively lengthy lead times for plant construction, they have little
influence on short and medium term pricing. If it is possible for a generator to sell power today at a
price in excess of long run marginal cost because of prevailing competitive market conditions it will
most certainly do so.
Data, methodology and scenarios
27 Frontier Economics | March 2006
However, if the interconnector capacity is small relative to market size and it is
constrained, prices may differ between regions. It is therefore very important to
treat interconnectors properly.
For the purposes of this study we are simulating the Belgium electricity market
explicitly and surrounding regions implicitly. We model the interconnector
between Belgium and France as an adjustment to demand. This reflects our
understanding that the Southern interconnector imports into Belgium with a high
load factor. By assuming that the interconnector is, in effect, fully loaded at all
times, we derive a conservative estimate of the extent of Electrabel’s incentive to
increase prices. The capacity that we assume for the Southern interconnector, by
season, is set out in Table 1. The capacity of the interconnector has been
provided by The CREG.
We also note that this treatment assumes that Electrabel is unable to extract rents
from the Southern interconnector. This reflects The CREG’s recent ruling on
forthcoming access arrangements to the Southern interconnector, in which
existing long term contracts over the interconnector have been torn up and
replaced by an auction. If Electrabel were to raise the price in Belgium, it would
not benefit for the interconnector because the price that it would have to pay to
acquire the rights to its use would rise by the same amount.
Virtual
interconnector
Winter capacity (MW)
Summer capacity (MW)
2,700
1,700
France
Table 1: Virtual interconnector power plants
Source: The CREG
The interconnector between Belgium and The Netherlands is also treated as an
adjustment to demand. The details of this adjustment are described in the
following sub-section. This treatment reflects the fact that the extent and
direction of flows over the Northern interconnector are less certain than flows
over the Southern interconnector.
Interconnector
Winter capacity (MW)
Summer capacity
(MW)
Belgium - The Netherlands
2,350
2,000
The Netherlands - Belgium
2,400
1,900
Table 2: Virtual interconnector power plants
Source: ETSO NTC Values for 2005, Frontier estimate
4.2.3 Demand
In order to make full use of our SPARK results, we need to map results for
representative periods (see the Results of our analysis in Section 5 for a more
detailed explanation of our modelling approach) onto actual demand data. For
Data, methodology and scenarios
28 Frontier Economics | March 2006
this, we used the hourly demand data taken from the Elia website for the year
20046. We then made the following adjustments:
• demand growth – 1.5% per year up to 2007;
• impact of the Belgian/Dutch interconnector – flows from Belgium to the
Netherlands (exports) have the effect of increasing the demand to be
served by Belgian generating assets, while flows from the Netherlands to
Belgium (imports) have the opposite effect; and
• pumped-storage pumping and generating decisions – using a pumped
storage optimizer module (described in more detail below).
Adjustments for flows across the Northern interconnector
The Elia website provides data on flows over the Northern interconnector. We
have adjusted Belgian demand to take account of these flows, scaled up by 1.5%
(capped at the capacity of the interconnector) to reflect demand growth, i.e. our
central analysis assumes that prevailing flows persist.
As we describe below, we have analysed the sensitivity of our analysis to this
assumption by modelling further scenarios that embody different assumptions
regarding the extent of flows over the interconnector.
Pumped storage stations
The pumping and generating decisions of pumped storage units are intertemporal, i.e. is determined within the day by the off-peak/peak price spread;
and, in some cases, the spread between weekdays and weekends. To capture this
we have dispatched pumped storage assets over the course of representative days
based on typical patterns of the total volume of generation. This has the effect
of “shaving” demand in peak periods when these assets are generating and
“filling” demand in off peak periods when these assets are pumping.
6
This was the most recent year for which data was available at the time the analysis was conducted.
Data, methodology and scenarios
29 Frontier Economics | March 2006
Figure 6: Demand adjustment for pumped storage
Final demand characteristics
In Figure 7 we illustrate a frequency chart of actual demand levels after both the
interconnector flow and hydro pumped storage adjustments.
Demand level frequencies
700
600
Frequency
500
400
300
200
100
15,200
14,800
14,400
14,000
13,600
13,200
12,800
12,400
12,000
11,600
11,200
10,800
10,400
10,000
9,600
9,200
8,800
8,400
8,000
0
Demand levels (MW)
Figure 7: Distribution of demand to be met by Belgium plant and the Southern I/C, 2007
Source: Elia raw data and Frontier adjustments and estimates
Data, methodology and scenarios
30 Frontier Economics | March 2006
4.3
MODELLING METHODOLOGY
Using our SPARK model we have analysed the scope for Electrabel to increase
prices above marginal cost at a wide range of representative demand levels. We
have then investigated how the imposition of a given level of contract cover
reduces this incentive.
| We model two representative seasons, Winter and Summer, to take account
of different fuel prices and plant availabilities.
| We model each of these seasons in the presence of a soft constraint on prices
(set at €150/MWh for demand levels in excess of 10,600 MW in the Winter
and 10,000 MW in the Summer, otherwise it is €100/MWh – see Section 4.5
for an explanation of how these assumptions have been derived).
| In all analysis, we have assigned Electrabel a strategy space that ranges
between bidding all units at their marginal cost up to bidding all units at 20
times their marginal cost, increasing in increments of 0.1. This is an
extremely rich strategy space and allows us to be sure that our analysis is not
driven by arbitrary restrictions on pricing behaviour.
| At each level of demand, we have determined the equilibrium price outcome
based on the profit maximising behaviour of Electrabel, taking account of an
assumed volume of contract cover.
• We have modelled contract volumes from 0 MW to 10,000 MW in
increments of 200 MW.
• In the first instance, we have assumed this contract volume is firm and is
baseload7. This ensures that all volume included in a given run is
exercised allowing easy comparison.
• By mapping this analysis onto our demand data we have been able to
estimate the impact of putting in place a given volume of baseload
contract cover on average annual prices. In particular we can assess the
incremental benefit of a further increase in contract volume, informing
the choice of the size of any set of contractual measures.
• Finally, we have used this analysis to determine what mix of VPP
products by strike price, might be optimal. We have modelled a variety of
volumes and mixes for each scenario in order to provide The CREG with
a range of options.
While the final step of the analysis, which focuses on the appropriate shape that
should be adopted for any package of contractual measures, has been based on
use of VPPs, similar analysis could be conducted to identify the mix of fixed long
7
In effect, this is consistent with modelling, for example, VPPs with a strike price of €0/MWh, a
fixed long term contract, a contract swap or a PPA with a station that would be expected to run in
all periods when it is available (ignoring planned/unplanned outages).
Data, methodology and scenarios
31 Frontier Economics | March 2006
term contracts that would be appropriate. However, the broad proportions of
different products identified through analysis of VPPs also provides a reasonable
indication of the how any other contractual remedy might be broken down into
different product types.
Note that the analysis we have conducted assumes that Electrabel has no other
long term contracts, or other positions that might have the effect of a contract
(such as a retail position with regulated tariffs). Should Electrabel have existing
positions, for example any volume released under the existing VPP programme,
then these should be netted off from any requirement identified in this report.
4.4
SCENARIOS FOR ANALYSIS
The Belgium electricity generation market is dominated by Electrabel and this
situation confers upon Electrabel market power and thus the potential to raise
market prices above competitive levels. Broadly, the aim of this study is to
analyse different measures to mitigate that market power. Our approach to the
analysis involves assessing the extent of market power for:
• a Base Case scenario to determine the total volume and mix of
contractual measures required to mitigate Electrabel’s dominance;
• scenarios for new entry(where we assume entry of 1.5GW or 2.5GW of
CCGT) to determine how such entry might modify the total volume (and
mix) required; and
• scenarios to explore the results with various different flows over the
Northern interconnector (increased and decreased exports to The
Netherlands) to determine how sensitive our analysis might be to such
variation in external conditions.
The base case uses the full set of assumptions and modelling approach described
above. Below we describe our sensitivity scenarios in more detail below.
4.4.1 Sensitivity: entry
We modelled two entry scenarios as sensitivities around the base case. The first
assumes entry of 1.5 GW of CCGT and the second assumes entry of 2.5 GW of
CCGT. We illustrate in the following figures the Winter and Summer MC stacks
for both sensitivity scenarios.
All new entry is assumed to belong to a non-strategic portfolio, ie one not owned
/controlled by Electrabel and by assumption one not able to exercise market
power. As a result of this new entry, Electrabel has a smaller share of the
available generating assets on the system. One would therefore expect that there
is a smaller requirement for contract cover under these two scenarios than under
the Base Case (as Section 5 demonstrates, this is indeed the case).
Data, methodology and scenarios
32 Frontier Economics | March 2006
[Contains confidential information]
Figure 8: Winter installed capacity by owner ( +1,500MW CCGT)
Source: Elia and Frontier estimates
[Contains confidential information]
Figure 9: Summer installed capacity by owner ( +1,500MW CCGT)
Source: Elia and Frontier estimates
[Contains confidential information]
Figure 10: Winter installed capacity by owner ( +2,500MW CCGT)
Source: Elia and Frontier estimates
[Contains confidential information]
Figure 11: Summer installed capacity by owner ( +2,500MW CCGT)
Source: Elia and Frontier estimates
4.4.2 Sensitivity: Northern interconnector flows
We also modelled different interconnector flows to the North (illustration of
effect in the following figure):
• one increasing exports to the North by 500 MW by hour (subject to the
capacity limit); and
• one decreasing exports to the North by 500 MW by hour.
Given the increasing steepness of the supply curve as demand increase, an
increase in flow over the interconnector has a greater positive affect on prices
than the negative affect of a similarly sized decrease.
Data, methodology and scenarios
33 Frontier Economics | March 2006
More exports increases
load to be met by Belgian
generators
Demand
(MW)
Less exports decreases
load to be met by Belgian
generators
Hours
24
Figure 12: Illustration of demand adjustment from different interconnector flows
Scenario demand distributions
In Figure 13 and Figure 14 we illustrate frequency charts of actual demand levels
after both the interconnector flow and hydro pumped storage adjustments for
both the 500 MW export increase and decrease. As would be expected, higher
demand levels occur more (less) often with more (less) exports to the North.
Data, methodology and scenarios
34 Frontier Economics | March 2006
Demand level frequencies
700
600
Frequency
500
400
300
200
100
15,200
14,800
14,400
14,000
13,600
13,200
12,800
12,400
12,000
11,600
11,200
10,800
10,400
10,000
9,600
9,200
8,800
8,400
8,000
0
Demand levels (MW)
Figure 13: Distribution of demand to be met by Belgium plant and the Southern I/C with
500 MW more exports to the North
Source: Elia raw data and Frontier adjustments and estimates
Demand level frequencies
700
600
Frequency
500
400
300
200
100
15,200
14,800
14,400
14,000
13,600
13,200
12,800
12,400
12,000
11,600
11,200
10,800
10,400
10,000
9,600
9,200
8,800
8,400
8,000
0
Demand levels (MW)
Figure 14: Distribution of demand to be met by Belgium plant and the Southern I/C with
500 MW less exports to the North
Source: Elia raw data and Frontier adjustments and estimates
Data, methodology and scenarios
35 Frontier Economics | March 2006
4.5
NEW ENTRY AND SOFT REGULATORY CONSTRAINTS
The effect of any programme of contractual measures depends on the behaviour
that the dominant party would be likely to follow in the absence of such a
programme. While there is no formal regulation of generation prices, we believe
that the threat of regulatory or competition law based intervention almost
certainly does constrain Electrabel’s pricing behaviour. It is clear that at present
Electrabel does not price up to anywhere near the level that consideration of its
market power alone would imply is feasible and profitable. We refer to the
constraints imposed by the implicit threat of such interventions as ‘soft
regulatory constraints’. While it is always difficult to quantify such constraints it
is important to recognise that they exist and will influence behaviour.
For the purposes of this study we have assumed that it is unlikely that Electrabel
would choose to price power in a way that left the price permanently well in
excess of the price of new entry. We have therefore informed our choice of soft
constraint by an estimate of the cost of new entry. We estimate that new entry as
a CCGT operating baseload would cost approximately €80/MWh at the
prevailing Belgian gas price and the EUA carbon price. With this in mind, we
have assumed that Electrabel would not consistently price off peak power above
€100 /MWh, nor peak power above €150/MWh at prevailing fuel and carbon
prices
As noted, it is difficult to identify with any certainty whether soft constraints do
in fact operate at the levels we identify. However, a full and comprehensive
review of the level at which soft constraints might begin to bind is likely to be
unnecessary, since our modelling indicates that the results of our analysis are not
highly sensitive to the level of the soft constraint.
Data, methodology and scenarios
37 Frontier Economics | March 2006
5 Results
In this section of the report we present the results from our analysis for the
different scenarios. The results are reported in line with the methodology
outlined in Section 4.3 for our Base Case and for the sensitivities we have
analysed.
5.1
BASE CASE
In this subsection we present results for the Base Case scenario, including:
• Step 1: raw output from SPARK on the equilibrium price outcome for
given levels of demand and given volumes of baseload contract cover (i.e.
equivalent to a given volume of VPP with an assumed exercise price of
zero, a fixed long term contract in place in all hours, or a PPA for a
station that would be expected to run at full capacity when available);
• Step 2: annual average price outcomes, for a given volume of contract
cover, calculated by mapping our raw SPARK results on to Belgian
demand data for an entire year; and
• Step 3: the average annual price for a number of different VPP product
package (base, mid, and peak) propositions, building on the results of
Step 1, to illustrate how products might be provided with shape to ensure
that they are effective but not disproportionate.
Please note that the price outcomes which we calculate should not be interpreted
as predictions of Electrabel’s behaviour but merely represent the potential were
profit maximising behaviour pursued with regard only to the constraints that we
noted
5.1.1 Step 1: Raw results
The output from SPARK is a set of static Nash equilibria prices by demand level.
In general it is possible to get several static Nash equilibria at any given demand
level. However, with only one player assumed to have the ability to act
strategically, it is typical to identify only one equilibrium at each demand level.
For the purposes of this study we investigated the impact on those price
outcomes of introducing different volumes of contract cover. As would be
expected, we observe that as the volume of contact cover increases for a given
demand level so the equilibrium price outcome (weakly) decreases. In some
cases, even 10,000 MW of contract cover does not lead to price outcomes
equalling the short run marginal cost of meeting incremental demand on the
system. While it would be possible to extend the analysis to higher volumes of
contract cover in order to find the volume required to achieve marginal cost
outcomes, we have not done so as we understand that contractual remedies in
excess of this size are regarded as infeasible.
Results
38 Frontier Economics | March 2006
In Figure 15, we illustrate price outcomes by contract volume for a winter
demand level of 12,000 MW. We observe that price outcomes:
• hit the soft constraints for contract volume levels up to 3,600 MW (i.e. a
package of contractual measures with volumes up to that level has no
effect at all at this level of demand);
• plateau around €140/MWh from 3,800 MW up to 7,400 MW of contract
cover;
• then fall dramatically but remain above marginal cost at €90/MWh; and
• never fall as low as marginal costs even as the volume is increased to
10,000 MW.
Winter - Demand level of 12,000 MW
160
Price
MC
140
120
€/MWh
100
80
60
40
20
0
200
400
600
800
1,000
1,200
1,400
1,600
1,800
2,000
2,200
2,400
2,600
2,800
3,000
3,200
3,400
3,600
3,800
4,000
4,200
4,400
4,600
4,800
5,000
5,200
5,400
5,600
5,800
6,000
6,200
6,400
6,600
6,800
7,000
7,200
7,400
7,600
7,800
8,000
8,200
8,400
8,600
8,800
9,000
9,200
9,400
9,600
9,800
10,000
0
Contract Volumes
Figure 15: SPARK price outcomes for different volumes of contract cover for a winter
demand level of 12,000 MW
Source: Frontier
In Figure 16, we illustrate the results for a higher Winter demand level of 12,800
MW, for which, broadly, the main observation is that higher levels of contract
volume still are required before price outcomes fall towards (yet never hit)
marginal costs. This illustrates a further feature of our results, that the volume of
contact cover required to achieve a given outcome typically increases as demand
increases (i.e. at higher levels of demand more contract volume is required).
Results
39 Frontier Economics | March 2006
Winter - Demand level of 12,800 MW
160
Price
MC
140
120
€/MWh
100
80
60
40
20
0
200
400
600
800
1,000
1,200
1,400
1,600
1,800
2,000
2,200
2,400
2,600
2,800
3,000
3,200
3,400
3,600
3,800
4,000
4,200
4,400
4,600
4,800
5,000
5,200
5,400
5,600
5,800
6,000
6,200
6,400
6,600
6,800
7,000
7,200
7,400
7,600
7,800
8,000
8,200
8,400
8,600
8,800
9,000
9,200
9,400
9,600
9,800
10,000
0
Contract Volumes
Figure 16: SPARK price outcomes for different volumes of contract cover for a winter
demand level of 12,800 MW
Source: Frontier
5.1.2 Step 2: Annual average prices by contract volume
We have mapped the price outcomes from SPARK to our adjusted demand data
in order to calculate volume weighted average prices for the year. This is
illustrated in Figure 17, along with the annual average price that would arise if all
units were bid at marginal cost (which is, of course, constant as contract volume
varies). We observe that average price outcomes:
• are on average capped at close to the soft constraint for contract volumes
up to 2,200 MW;
• decrease with contract volume but only very slowly up to 5,600 MW;
• fall more rapidly with increases in contract volume up to 8,000 MW; and
• decrease more slowly after that, although never reaching marginal cost
levels.
Results
40 Frontier Economics | March 2006
160
Base case
MC
140
Price (€/MWh)
120
100
80
60
40
20
0
200
400
600
800
1,000
1,200
1,400
1,600
1,800
2,000
2,200
2,400
2,600
2,800
3,000
3,200
3,400
3,600
3,800
4,000
4,200
4,400
4,600
4,800
5,000
5,200
5,400
5,600
5,800
6,000
6,200
6,400
6,600
6,800
7,000
7,200
7,400
7,600
7,800
8,000
8,200
8,400
8,600
8,800
9,000
9,200
9,400
9,600
9,800
10,000
0
Contract volume (MW)
Figure 17: Average market price under a given volume of contract cover (assumed to be
baseload) for the base case scenario
Source: Frontier
This analysis suggests that 8,000 MW might be an appropriate gross volume of
contract cover to consider, as there is a relatively modest benefit arising from any
increase above that level. Again we stress that this volume could be made up of
any of the forms of contractual cover discussed in Section 2.
We now discuss our reasons for this selection, beginning with the criterion by
which different volumes have been assessed.
One obvious criterion would be to achieve competitive price levels. A literal
interpretation of this objective would lead to a conclusion that there should be
fiercely competitive supply even in the peak hours of the year. However, to do
so would require large volumes of contract cover, for very little additional
competitive effect because the number of periods affected is so small.
Furthermore, even quite competitive markets typically exhibit some potential
market power in peak periods.
We therefore consider, as an alternative, the marginal benefit of additional
contract cover. This is defined to be the additional (demand-weighted average)
price reduction arising from the “last” 200MW of contract cover imposed. Thus,
the marginal benefit of 8,000MW of contract cover is the demand-weighted
average market price with 7,800MW of contract cover minus the price similarly
defined with 8,000MW contract cover. We identify a level of contract cover at
which additional customer benefits from the last increment of product are large
and at which the additional benefits from subsequent increments of products are
lower. We illustrate the marginal benefit (the price reduction resulting from each
200 MW tranche of additional contract cover) in Figure 18 below.
Results
41 Frontier Economics | March 2006
-9
-8
Marginal benefit (€/MWh)
-7
-6
-5
-4
-3
-2
-1
9,800
10,000
9,600
9,400
9,200
9,000
8,800
8,600
8,400
8,200
8,000
7,800
7,600
7,400
7,200
7,000
6,800
6,600
6,400
6,200
6,000
5,800
5,600
5,400
5,200
5,000
4,800
4,600
4,400
4,200
4,000
3,800
3,600
3,400
3,200
3,000
2,800
2,600
2,400
2,200
2,000
1,800
1,600
1,400
800
1,200
1,000
600
400
200
0
Contract volume (MW)
Figure 18: Incremental benefit of an extra 200 MW of contract volume
Source: Frontier
Figure 18 reinforces the description of Figure 17 provided above. At low
volumes of contract cover each extra increment has little effect, until over 5,000
MW are in place. Beyond this level, additional contract volume results in more
substantial benefits until beyond 8,000 MW when the benefit is smaller. Higher
volumes result in negligible additional benefits. This reinforces the view that
8,000 MW of imposed contract cover (before accounting for any other form of
measures or existing contract positions) might be appropriate.
5.1.3 Step 3: VPP product packages
Up to this point we have identified the total volume of contract cover required to
reduce the incentives of Electrabel to raise prices. In this sub-section we present
an analysis of how this total volume of contract cover might be broken down
into a mix of products. For illustration, we focus on VPP products and
determine the product mix that will meet that objective while providing an
appropriate range of products for potential VPP purchasers. In order to allow
potential competitors of Electrabel to compete more effectively, and also to
ensure that the total impact of any programme is proportionate, the VPP product
(or equivalents) cannot all be (as was assumed for the zero exercise price VPP
products analysed so far). For effective competition to occur, potential
competitors will need to offer or have the ability to offer a suite of baseload,
mid-merit and peak products By inspection of the results of our modelling to
identify the volume of products needed at different demand levels, it is possible
to make an informed judgement on the mix of products that would be
appropriate. As with the choice of total volume this is not a precisely
mechanistic exercise but one which seeks to identify a simple portfolio of VPPs
Results
42 Frontier Economics | March 2006
that would, when called, give rise to the right quantum of contract cover for the
demand level. In doing this we restrict ourselves to a choice of 3 VPP products
with exercise prices appropriate to the technology steps in the supply curve.
Following this approach we identify a sensible split of our 8GW total as:
• a total baseload product volume of 3,000 MW at an exercise price of
€11/MWh (just above the nuclear power plant price);
• a total mid-merit product volume of 3,000 MW at an exercise price of
€41/MWh (just around the coal power plant price); and
• a peak product volume of 2,000 MW at a an exercise price of €81/MWh
(just around the CCGT power plant price).
As noted, the exercise prices reflect clear steps in the Belgium merit order and
should therefore be exercised as demand reaches the different steps. In practice
we would recommend that the exercise prices of the different products are
matched ex ante to prevailing input fuel forward prices so they continue to track
the relevant steps in the supply curve. We also note that while this final step of
the analysis focuses on VPP products, similar analysis could be conducted to
identify how an appropriate shape should be given to any of the other
contractual measures discussed above. For example, we could define a similar
range of products (base, mid-merit and peak) for an analysis of fixed long term
contracts and would identify a similar suggested mix of those products.
The volumes that we have selected for each of these products have been
informed by analysis of the raw results underlying this analysis and are aimed at
attempting to ensure that the volume of VPP likely to be exercised at any given
level of demand is not unnecessarily large. However, as neither the overall
volume nor the mix can be mechanistically determined, we also explore through
our SPARK model, a small number of variants which add or subtract volumes
from the volumes of the three products contained in our central package.
The raw price results by demand level coming out of SPARK for our Central
case VPP product mix are illustrated in Figure 19, along with the underlying
marginal costs and demand distribution. Three series are presented in the chart.
| The short run marginal cost of meeting demand at the relevant level of
demand (measured on the left hand y-axis). This acts as a point of
comparison for the Price series described immediately below.
| The equilibrium price that our analysis suggests would be profit maximising
for Electrabel at the relevant level of demand, given the imposition of the
specified package of VPP products (measured on the left hand y-axis).
| The frequency distribution of demand (measured on the right hand y-axis).
This illustrates how frequently we might expect the given demand level to
occur over the course of the year.
Results
43 Frontier Economics | March 2006
We observe a number of steps in the marginal cost curve. These reflect the
points in the supply curve where all assets of one technology type are deployed
and a small increase in demand causes the next most expensive technology type
to be deployed.
We observe that prices are close to marginal costs except along the final “step”
of the supply curve (above 13,200 MW). The ability to price up in peak periods
is unlikely to be a major cause for concern, as this is almost inevitable given the
fundamental characteristics of electricity markets8. As the demand curve
illustrates, such high levels of demand occur relatively infrequently.
160
700
140
600
120
400
80
300
60
Price
MC
Demand distribution
40
Demand frequency
Price (€/MWh)
500
100
200
100
20
0
7,000
0
9,000
11,000
13,000
15,000
17,000
Demand level (MW)
Figure 19: Central case product package (base case): Price outcome by demand level
Source: Frontier
To allow comparison, we have also modelled the affect of increasing the peak
load contract volume by 1 GW (Variant 2 in Table 3). We illustrate these results
in Figure 20. We observe that price outcome around the final steps are much
closer to marginal costs than in the previous results, and that the prices that hit
the soft constraint only occur at demand levels that are very rare – only in the
small proportion of periods in each year demand is very high.
8
Prices above short run marginal cost are almost certain to occur in some periods of time. The most
obvious example is in periods where only a single plant has surplus capacity. In such circumstances
this unit can bid its remaining capacity into the market at a price consistent with the willingness to
pay of consumers, which typically lies well above the cost of production. In cases where demand at
a price equal to the highest marginal cost of plant would exceed available capacity, efficient prices
need to exceed marginal cost of supply in order to ration available capacity. The presence of prices
above short run marginal cost is, of course, required if peaking plant is to recover its fixed costs.
Results
44 Frontier Economics | March 2006
160
700
140
600
120
400
80
300
60
Price
MC
Demand distribution
40
200
100
20
0
7,000
Demand frequency
Price (€/MWh)
500
100
0
9,000
11,000
13,000
15,000
17,000
Demand level (MW)
Figure 20: Central case product package plus 1 GW peak (Variant 2): Price outcome by
demand level
Source: Frontier
In the following table, we present the yearly average prices arising from the base
case and Variant 2 together with two additional variants:
• the impact of 1 GW of mid-merit VPP less on the Central case (Variant
1); and
• the impact of an additional 1 GW of baseload VPP on the Central case
(Variant 3)
Central
package
Package
Variant 1
Package
Variant 2
Package
Baseload (€11/MWh)
3,000 MW
3,000 MW
3,000 MW
4,000 MW
Mid-merit (€41/MWh)
3,000 MW
2,000 MW
3,000 MW
3,000 MW
Peak (€81/MWh)
2,000 MW
2,000 MW
3,000 MW
2,000 MW
Average price
outcome (if Electrabel
90 €/MWh
109 €/MWh
85 €/MWh
80 €/MWh
71 €/MWh
71 €/MWh
71 €/MWh
71 €/MWh
Variant 3
profit maximises)
Marginal cost
Table 3: Base case scenario product mixes and average prices
Source: Frontier
The above results lead us to the following conclusions:
• Package Variant 1 with a lower volume, appears to be substantially less
effective than our Central Package;
• Packages with an increased volume, Variants 2 and 3, both provide a
more stringent constraint on Electrabel; and
Results
45 Frontier Economics | March 2006
• The effect of adding a 1GW of product to the baseload requirement
produces a slightly greater effect than adding a 1GW requirement to
peakload. The average price outcome (if Electrabel were profit
maximising) would move down by just over 5 %).
Against this background, and given the difficulty there may be in getting an
aggressive package agreed, we believe that it is reasonable to adopt our Central
Package. We also note that the price outcome from our Central Package is close
to the new entry price plus a premium (see Section 4.5), implying that while this
level of VPP release is effective in constraining prices, it would still allow scope
for new entry if Electrabel were to price up to the level which would maximise its
profits.
5.2
SENSITIVITY: MARKET ENTRY
In this sub-section we present results from the new entry sensitivity scenarios,
assuming 1.5 GW and 2.5 GW CCGT of new entry respectively. We do not
present the raw results here but only the aggregated price outcomes (if Electrabel
were to profit maximise) by contract volume level, looking at contract cover
generically) and the average prices under relevant VPP packages.
5.2.1 Annual average prices by contract volume
The purpose of this new entry sensitivity scenario is to test for the impact of
non-strategic entry on price outcomes and test, in particular, to what extent entry
introduces more competition to the market.
In the same way that we presented the previous set of results, we illustrate in
Figure 21 the volume weighted average price outcomes with various volumes of
contract cover for the base case +1,500 MW and +2,500 MW CCGT entry. We
clearly observe that new entry reduces potential price outcomes significantly
compared to the base case – therefore new entry has a strong positive impact on
maximum extent to which market power could profitably be exercised. In
particular, we note that:
• as more entry occurs (i.e. going from base case to +1,500 MW, and then
to +2,500 MW) average prices tend to be lower and fall faster, in line with
what one would anticipate; and
• even with new entry, potential price outcomes are not reduced fully to
marginal costs for contract volumes of 10,000 MW or less.
Results
46 Frontier Economics | March 2006
Base case
Base case +1500
Base case +2500
MC
MC +1500
MC +2500
160
140
Price (€/MWh)
120
100
80
60
40
20
0
200
400
600
800
1,000
1,200
1,400
1,600
1,800
2,000
2,200
2,400
2,600
2,800
3,000
3,200
3,400
3,600
3,800
4,000
4,200
4,400
4,600
4,800
5,000
5,200
5,400
5,600
5,800
6,000
6,200
6,400
6,600
6,800
7,000
7,200
7,400
7,600
7,800
8,000
8,200
8,400
8,600
8,800
9,000
9,200
9,400
9,600
9,800
10,000
0
Contract volume (MW)
Figure 21: Average profit maximising price under a given volume of contract cover
(assumed baseload) for the entry scenarios compared to the base case
Source: Frontier
We also present in Figure 22 (for the +1,500 MW case) the incremental benefit
of an additional 200 MW of contract volumes on average annual profit
maximising price. The shape of the figure also has peaks and troughs over a
large band of contract volume tranches, ranging from about 4,600 MW to about
7,800 MW. The incremental benefit of an additional tranche beyond a total
volume of 6,800 MW is, however, relatively small.
We present in Figure 23 (for the 2,500 MW case) the incremental benefit of an
additional 200 MW of contract volume in reducing the level that it would be
profitable for Electrabel to price up to. We observe that the incremental benefit
of an additional tranche beyond a total volume of 6,200 MW is relatively small.
Results
200
400
600
800
1,000
1,200
1,400
1,600
1,800
2,000
2,200
2,400
2,600
2,800
3,000
3,200
3,400
3,600
3,800
4,000
4,200
4,400
4,600
4,800
5,000
5,200
5,400
5,600
5,800
6,000
6,200
6,400
6,600
6,800
7,000
7,200
7,400
7,600
7,800
8,000
8,200
8,400
8,600
8,800
9,000
9,200
9,400
9,600
9,800
10,00
Marginal benefit (€/MWh)
200
400
600
800
1,000
1,200
1,400
1,600
1,800
2,000
2,200
2,400
2,600
2,800
3,000
3,200
3,400
3,600
3,800
4,000
4,200
4,400
4,600
4,800
5,000
5,200
5,400
5,600
5,800
6,000
6,200
6,400
6,600
6,800
7,000
7,200
7,400
7,600
7,800
8,000
8,200
8,400
8,600
8,800
9,000
9,200
9,400
9,600
9,800
10,00
Marginal benefit (€/MWh)
47 Frontier Economics | March 2006
-9
-8
-7
-6
-5
-4
-3
-2
-1
0
Contract volume (MW)
Figure 22: Incremental benefit of extra 200 MW of contract cover for the +1,500 MW new
entry scenario
Source: Frontier
-8
-7
-6
-5
-4
-3
-2
-1
0
Contract volume (MW)
Figure 23: Incremental benefit of extra 200 MW of contract cover for the +2,500 MW new
entry scenario
Source: Frontier
Results
48 Frontier Economics | March 2006
5.2.2 VPP package variants
In Table 4, we present volume weighted average prices for both the Central
Package and for Package Variant 1. (We only analyse Package Variant 1 as new
entry can only reduce the requirement).
We observe that the average profit maximising prices:
• are substantially lower than under the base case scenario regardless of the
mix and cumulative volume of VPP modelled and that they are much
closer to the underlying marginal cost (which is itself lower than in the
base case reflecting entry);
• are lower still as more new entry occurs;
• while the outcome of Variant 1 in the Base Case might not be regarded as
acceptable, Variant 1 appears to deliver entirely adequate results in the
presence of 1.5 GW of entry; and
• smaller volumes of VPP still might be regarded as effective in the
presence of 2.5 GW of entry.
Central
Package
(with 1.5 GW
entry)
Central
Package
(with 2.5 GW
entry)
Package
Variant 1
(with 1.5 GW
entry)
Package
Variant 1
(with 2.5 GW
entry)
Baseload (€11/MWh)
3,000 MW
3,000 MW
3,000 MW
3,000 MW
Mid-merit (€41/MWh)
3,000 MW
3,000 MW
2,000 MW
2,000 MW
Peak (€81/MWh)
2,000 MW
2,000 MW
2,000 MW
2,000 MW
Average price
outcome
68 €/MWh
61 €/MWh
76 €/MWh
63 €/MWh
Marginal cost
62 €/MWh
59 €/MWh
62 €/MWh
59 €/MWh
Table 4: New entry scenario product mixes and average prices
Source: Frontier
This analysis would suggest that a smaller volume of VPP (or some other
contractual remedy) is required in the presence of entry. With 1.5 GW of entry,
it would appear that the extent of the VPP programme can be reduced by at least
1 GW. With entry of 2.5 GW, the reduction in the size of the required package
of measures could be at least 2 GW.
Results
49 Frontier Economics | March 2006
5.3
SENSITIVITY: VARIATION IN FLOW OVER THE
NORTHERN INTERCONNECTOR
In this sub-section we present results from the Northern interconnector flow
sensitivity scenarios of 500 MW more exports and 500 MW less exports.
In Table 5, we present the volume weighted average prices for these scenarios
under the Central Package VPP volume mix. We observe that:
• as expected, both average profit maximising prices and the underlying
marginal costs increase compared to the Base Case as exports to the
North increase; and
• again, as expected, both average profit maximising prices and the
underlying marginal costs decrease compared to the Base Case as exports
to the North decrease.
Central Package
(Base Case)
Central Package
(500 MW more
exports)
Central Package
(500 MW less
exports)
Average price
outcome
90 €/MWh
104 €/MWh
81 €/MWh
Marginal cost
71 €/MWh
75 €/MWh
67 €/MWh
Table 5: Northern export scenario product mixes and average prices
Source: Frontier
This analysis would suggest that, with a consistently greater level of exports to
the Netherlands, a VPP programme in excess of 8 GW would be needed. In
contrast, if exports were to reduce, then an 8 GW VPP programme would be
highly effective in constraining prices and the quantum of release could possibly
be reduced.
5.4
SUMMARY
In this section we have demonstrated that under our base case scenario, a
package of measures with a cumulative volume of 8 GW is likely to be quite
effective in restraining prices to levels consistent with a more competitive market
structure.. We have also shown that the volume of contract cover required to
restrain prices at reasonable levels is likely to fall if there is new entry by a nonstrategic player. If there is entry of 1.5 GW then our analysis suggests that 7 GW
of contract cover might be appropriate, even less if there is further entry.
Similarly, 8 GW of cover is likely to be excessive if exports to the Netherlands
were to decrease in the future.
Results
50 Frontier Economics | March 2006
Our analysis has not included any other contracts (or positions that might have
the economic effect of a contract) that Electrabel might have in place at present
or might put in place. If Electrabel were able to demonstrate that it has in place,
or could commit to having put in place, such contracts and/or positions, then
these should be deducted from the volumes that we identify here. For example,
if Electrabel has in place already contracts of a reasonable duration at a fixed
price for the sale of 1,000 MW of base load power, then our recommendation
would fall from 8 GW to 7 GW, with the volume of baseload product required
reduced by 1 GW to reflect this existing contract position. Similarly, if Electrabel
were to have in place a peak load contract at a fixed price for 500 MW, this could
be deducted from the required volume of peak load product.
The total volume we recommend, net of any existing contracts described above,
could be made up from a combination of the different contractual remedies
discussed in Section 2. If Electrabel were to accept a contract swap of 1 GW,
again this could be deducted from the gross requirement we identify in this
section, reducing the required size of the remaining package of contractual
measures.
We note that in proportionate terms, the scale of the package of measures
indicated in this report would be unprecedented. While EDF has implemented a
capacity release programme that approaches the same absolute capacity (6 GW as
opposed to 8 GW), that programme represented a far smaller proportion of the
relevant generation market.
It is not within the scope of this assignment to address implementation issues in
detail. However, we would be negligent if we did not make clear that a
programme of this scale needs the volume released to be ramped up gradually to
the 8GW over a substantial period of time, if unreasonable market disturbance is
to be avoided. The EC required EDF to release capacity at ‘a pace compatible
with demand and the proper functioning of the market’. Even though the EDF
scheme is proportionately very much smaller, full volume was only reached after
two years from the first release.
Results
51 Frontier Economics | March 2006
6 Dynamic concerns – barriers to entry
The foregoing has addressed primarily concerns regarding static competition with
the existing generation park in Belgium. In this section we turn to dynamic
concerns, essentially related to barriers to entry that inhibit the competitive
development of the generation market over time.
Specifically, in this section we discuss:
• a range of potential barriers to entry in the Belgian electricity market and
the possible measures to alleviate at least some of these. A key concern
raised is the availability of sites for new generation;
• the incentives that an incumbent could have to hold on to unused sites
suitable for generation; and
• possible measures facilitate or mandate the release of suitable sites.
6.1
BARRIERS TO ENTRY
There are a number of potential barriers to entry in the Belgian wholesale
electricity market. Some of these are intrinsic to the activity and arise in all
similar markets around the world. Others arise as a result of factors that are
more specific to Belgium.
6.1.1 Lumpy investments and large sunk costs
Since it is not practical to add capacity in small increments, investments in
wholesale generating assets are necessarily large. As a result there are large up
front investments to take account of when considering whether to enter a
wholesale market. While this makes entry into the wholesale electricity market
more risky - and therefore less likely, such costs are unavoidable. Furthermore,
all possible entrants to the wholesale market are likely to be large, sophisticated
agents, probably participating in generation in other markets. As a result,
potential entrants are likely to be well aware of the problems associated with
lumpy investment and prepared to deal with them.
These problems of lumpy investments are likely to more substantive in markets
that are small and poorly interconnected The presence of cross border capacity
can partially mitigate the problem.
In a small market, the addition of even a modest increment of capacity can have
a large impact on prices, whereas larger markets should be able to accommodate
a similar volume of entry without seeing a similar impact on prices resulting.
Similarly, where markets are connected with their neighbours, potential entrants
have the additional option of selling their output to neighbouring regions,
assuming that such sales are economic. Interconnectors can therefore help to
smooth the introduction of new capacity and minimise the entry deterring effect
of large, lumpy investment costs.
Dynamic concerns – barriers to entry
52 Frontier Economics | March 2006
The recent expansion of the France – Belgium interconnector should help, as
potentially would investment in phase shifting transformers to improve capacity.
Furthermore, improved international TSO co-operation working with PTDF
models might also allow greater use to be made of existing interconnector assets
and thereby reduce this barrier to entry.
6.1.2 Ability to find demand for output
Potential entrants into the wholesale electricity market will need to consider
whether there is scope for them to sell their output, at a reasonable price, in the
market following entry. In markets where consumers of electricity are already
supplied under long term contracts (i.e. tied to a rival generator) there might be a
concern that they would be unable to find customers for their output. This
concern might be exacerbated in markets where vertical integration between
generators and retailers is common.
We have seen no evidence to suggest that finding a customer base is likely to be a
substantive issue in Belgium. Discussion with members of the steering group has
indicated that there are many large customers seeking to purchase competitive
power supplies. In line with the relevant Directives, the entire Belgian retail
market will be open to competition in 2007, which should provide further
options for any potential entrant in the wholesale market to sell its output.
6.1.3 Uncertainty arising from illiquid markets
Liquid and transparent forward and spot markets can play a valuable role in
facilitating entry. Where markets provide reliable and robust price signals,
potential entrants will be able to value better their investment opportunities.
Where participants are unsure whether market prices are reliable and unbiased,
they will typically be less willing to enter. Problems with illiquid markets might,
again, be exacerbated by markets in which there is considerable vertical
integration, as such some proportion of the trade between generators and
retailers will be conducted as transfers within a company.
The development of BelPex and the recent increase in the capacity of the link
with France should provide increased liquidity and more certainty over forward
prices for potential entrants. Similarly, the value of products sold at auction as
part of any enforced capacity release programme can also provide very helpful
signals to potential entrants on the value of power. In this way, contractual
remedies can be used to increase market transparency and reduce one potential
barrier to entry.
6.1.4 Availability of sites
There is a perception that the sites on which it is both possible and attractive to
build new generating assets are likely to be scarce. Belgium, in line with most
Western European countries, is a densely populated country. As a result, we
understand, there are relatively few locations where it would be acceptable to site
a new generating facility. An entrant would need to go through the relevant local
Dynamic concerns – barriers to entry
53 Frontier Economics | March 2006
planning processes and gain the relevant local planning and environmental
consents. Such consents might not be granted for many potential sites. It is
likely that a substantial proportion of the possible sites for new generation will be
sites on which there has been electricity generation in the past, where those assets
have been decommissioned.
The CREG has indicated that the availability of suitable sites for the construction
of new generation is a potential concern in Belgium. In particular, the CREG
has indicated that Electrabel may own a large proportion of such sites, including
the most attractive sites for new build. As we describe below, we have been
unable to gather data either to support or refute this assertion. However, it
would not be surprising for Electrabel to hold a high proportion of possible sites,
since sites on which there has been a generating facility in the past are typically
attractive sites for new build. It is at least plausible that Electrabel as the
incumbent generator is more likely to own sites on which there was once a
generating facility.
If all attractive sites are held by Electrabel, this could create an additional barrier
to entry. Electrabel might be unwilling to sell sites to potential rivals, or might
only do so at prices that would render entry unprofitable.
6.2
ANALYSIS OF INCENTIVES TO WITHHOLD SITES
The discussion above has highlighted the availability of sites as a potential
concern. It is also one in which there might be a role for a regulatory
intervention.
A dominant player in the wholesale electricity market that also owns the majority
of the most attractive sites on which new generation capacity could be added is
likely to have incentives to retain these sites, even if they do not plan to construct
assets on them in the foreseeable future. If the dominant player were to release a
site to a new entrant who proceeded to build capacity, this would be likely to
reduce the dominant player’s market share and also to erode prices. Both of
these effects would reduce the profits earned by the dominant player. As such,
keeping sites vacant could have considerable value to a dominant player, allowing
it to maintain a high market share and achieve a high margin on that volume.
The dominant player would no doubt prefer to retain ownership of all possible
sites and build on them itself in the fullness of time. Since a dominant player
profits from higher market prices, it would wish to avoid creating a situation in
which there is “too large” a surplus of plant. By keeping the reserve margin
relatively tight, a large player would be more readily able to demonstrate a scarcity
of supply that would justify higher prices. It is therefore possible that the rate at
which it would build additional capacity would be below that which is socially
optimal, but be nearer the rate that will maximise the value of its business.
Dynamic concerns – barriers to entry
54 Frontier Economics | March 2006
On the basis of this high level analysis it seems clear that, in principle at least,
there is reason to suppose that Electrabel might have both a number of relevant
sites and also faces financial incentives to retain those sites even if it does not
intend to build on them in the foreseeable future. In addition, this analysis also
allows us to consider how a policy response could be used to resolve this
potential problem. This is discussed in the following subsection.
6.3
POSSIBLE POLICY OPTIONS TO ADDRESS SITE
AVAILABILITY
In the absence of concrete information on the relevant facts regarding the
availability of sites in Belgium, it is not possible to provide definitive advice on
what is likely to be the best policy response to deal with a scarcity of sites should
such sites as there are be owned by a dominant incumbent. However, in this
section we explore the measures that could in theory be introduced to make sites
available to potential entrants and discuss what impact these might have.
6.3.1 Enforced (or negotiated) release of sites
We understand through discussion with the CREG that Electrabel may have
agreed with the Minister to release sites that would allow the construction of
1,500 MW of new entry. This understanding has prompted one of the
sensitivities that we have modelled. However, we have not been provided with
any details of this programme.
If a release of sites has been agreed with Electrabel and if this route remains open
in the future, then it would seem to represent a useful approach to ensuring site
release over time.
However, we would advise that the effectiveness of this policy be assessed after
its first implementation is completed in order to understand whether an
extension of the same mechanism is appropriate.
6.3.2 Requirement to auction vacant sites
Although it would undoubtedly have implications for required legislation, policy
makers could in theory decide to create an obligation to auction vacant sites after
they have remained unused for some period of time. This would ensure that the
available stock of sites was either used by their owner or made available for use
by others. However, this would give rise to a number of potential difficulties.
| Definition of a site: adopting this policy would require some independent
body to decide how to define a site suitable for generation. While some sites
might be obviously suitable and some obviously unsuitable the dividing line
could be very difficult to draw and could be disputed. We understand that
identification of sites is a problem with the Government’s current sites
initiative.
Dynamic concerns – barriers to entry
55 Frontier Economics | March 2006
| Definition of timing of release if all such defined sites are not instantly to
be released (something which would seem rash, there would have to be a
policy as to the pace of release.
| Monitoring of compliance: once a definition of a site had been agreed, a
list of the status of all such sites would need to be maintained and kept up to
date to ensure that sites were being made available as required. Again, the
criteria used to assess whether a site is vacant or otherwise could be
contentious and might be disputed.
| Alternative use of the site: a dominant player could circumvent the
intention of such a policy by ensuring that the site is used for some
alternative use aside from electricity generation. The site would no longer be
vacant but it would no longer be able to support a generating asset. The
relevant regulatory body would in effect need a right of veto over change of
use.
| Participation in the auction: to avoid any issue of discrimination it is likely
that it would be necessary to allow any interested party to participate in the
auction, including non-energy sector players. Again this could result in the
sale and use of the site without supporting entry, either because the site is
used for some alternative purpose. If the auction were fair and transparent,
and a party wanting the land for a use other than electricity generation paid
the higher price, it is very difficult to argue that the site was ‘particularly
suitable’ for generation.
| Reserve price: companies required to auction sites might require a reserve
price to ensure that there is no danger of expropriation of value. The value
of the site (as determined by outside assessors) in some other use might
provide an appropriate reserve to ensure that sites are not used inefficiently
for generation and that companies receive an appropriate value for any land
they are required to auction.
These difficulties suggest that while enforced auctioning of vacant sites might be
a helpful policy in stimulating site release, it could be difficult to implement and
enforce in practice.
6.3.3 Licensing regime
Given any required legislation, the CREG could use the licensing regime for new
generating assets to reduce the value of sites to Electrabel. For example, the
licensing regime could be modified to ensure that parties with a market share in
excess of some proportion of the total installed capacity were not allowed to
build further capacity.
This would reduce the value to Electrabel of withholding sites. Over time,
coupled with effective merger control to prevent concentration through
acquisition, this would be sure to result in Electrabel’s market share falling below
the defined threshold. Italy has operated a law prohibiting any company from
Dynamic concerns – barriers to entry
56 Frontier Economics | March 2006
owing more than 50% of installed capacity, although we note that
implementation of this type of restriction is not without its difficulties. For
example:
• How should any mothballed plant be treated in such calculations?
• How should temporarily or permanently derated plant be treated and who
will measure such capacities?
• How should variations in capacity with fuel type be dealt with?
• Is it reasonable for a party to be put in an illegal position by the
unanticipated decommissioning of plant by others over which they have
no control and may have no knowledge?
• Would such a rule incentivise Electrabel to decommission open cycle gas
turbines in order get round the constraint when it would be economically
preferable to maintain them albeit as reserve?
We do not rule out the use of this type of licensing constraint but caution that its
implementation will need very careful specification.
6.3.4 Resolve existing issues of market power
A further measure that might reduce, but not eliminate Electrabel’s potential
incentive to withhold sites would be to eliminate, through contractual measures,
Electrabel’s market power. If it were the case that Electrabel had exerted market
power in order to raise prices, then it would be in a position where the loss that
new entry would induce would be greater than if lower competitive market prices
prevailed. However, if Electrabel is not currently exerting market power, its
incentives with regard to withholding sites would remain unchanged with this
measure.
While resolving dominance might help to reduce the incentives to withhold sites,
it would not remove them entirely. As a generator, Electrabel would still be able
to profit from allowing the reserve margin to fall and creating genuine physical
scarcity.
6.3.5 Tax on vacant sites
We also understand that there is an existing proposal to introduce a tax on vacant
sites and that this is may be part of the Government’s agreement with Electrabel
regarding site release. This would clearly have the effect of reducing the value to
Electrabel of withholding sites, although the level of the tax would need to be set
to ensure that it was material enough to affect their decisions.
Dynamic concerns – barriers to entry
57 Frontier Economics | March 2006
As with a proposal to require vacant sites to be auctioned, this proposal gives rise
to many of the same difficulties regarding definitions and monitoring of
adherence.
6.3.6 Other methods to stimulate investment
The final way in which the sites issue could be overcome is through direct
intervention. In theory, for example, the Belgian regulator could put out a tender
for the provision of a power station at some agreed site, procured by
Government. This would both ensure that a site was made available, but would
also ensure that entry occurred.
A model involving centralised procurement of this kind has been applied in a
number of other countries, including, for example, Ireland. In Ireland this policy
has been successful achieving entry. However, the Irish regulator (CER) only has
powers to do this to ensure security of supply and has no right to pursue this
measure solely for the purpose of reducing the incumbent’s market power. The
policy is also a direct intervention in the market and is very likely to reduce the
incentive of anyone else to invest independently in generation.
6.4
CONCLUSION REGARDING SITES
We have explored the options for dealing with this issue but, given that within
this assignment it has not been possible to obtain hard data, we cannot quantify
the importance of the issue or make firm recommendations on the way forward.
Given the advanced nature of the Government’s current policy to achieve site
release, the prudent policy is to wait to assess the way in which this measure has
worked before assessing whether a repeat of this or some other initiative should
follow.
Dynamic concerns – barriers to entry
59 Frontier Economics | March 2006
Annexe 1: SPARK gaming module
Our approach to market analysis is based on prices and pay-offs to portfolio
generators. We can define a portfolio to be anywhere from a single generating
unit within a power station up to all the power stations on a given market.
The particular strategy decision (e.g. mark-up levels) of players will depend on
their expectation about their competitors’ behaviour. We identify Nash
equilibria, where each player takes into account the strategies that other players
find optimal. Modelling establishes equilibria as situations in which neither party
wants to deviate from a strategy.
We use numerical simulations to compute strategic equilibria in electricity
wholesale markets.
WHAT CONSTRAINS STRATEGIC BEHAVIOUR?
Strategic behaviour may be constrained by:
| Regulatory control –regulatory offices in many countries monitor electricity
wholesale market outcomes closely and have often intervened. In the UK
over the past 10 years, this intervention has taken the form of constraints on
specific types of behaviour, formal price control and the divestment of plant.
| Market entry of new capacity – high prices may induce the construction of
new plants by smaller players who aim to benefit from strategic bidding
behaviour of larger players. It is both feasible and desirable to model the
effect of market entry on market outcomes and, conversely, to evaluate the
impact of the market outcome on the entry decision itself.
| Contract coverage - if players have hedged against volatile spot prices
through financial contracts, net revenue streams for the volumes contracted
are pre-determined (by these contracts). Contract coverage may therefore
reduce incentives to manipulate prices in the spot market. However, spot
prices also affect price expectations and therefore the strike prices in financial
contracts. As incumbents also have an interest in high contract prices, their
incentives to drive up prices may prevail, even if there is full contract
coverage. SPARK allows us to advise clients on the effect of contract
coverage on bidding behaviour if required.
HOW ARE STRATEGY OUTCOMES COMPUTED?
The SPARK Game Module analyses generator strategies on the basis of the main
principles of game theory. In essence the Game Module is set up to identify
profit maximising bidding strategies. This is done by assigning a number of
strategies to players, in terms of:
Annexe 1: SPARK gaming module
60 Frontier Economics | March 2006
• mark-ups of bid prices for individual plants above variable generation
cost; or
• plant withdrawal from the market.
Each player can be assigned a large number of strategies that will be tested
against each other. All possible combinations of strategies that have been
assigned to players define the strategy space. SPARK searches the strategy space
for sustainable strategy combinations by computing prices and payoffs for each
strategy combination.
Annexe 1: SPARK gaming module
61 Frontier Economics | March 2006
Annexe 2: Data
In this Annexe we detail the data we used for the purposes of this study.
Asset register by plant type
The full asset register used in this study is presented in Table 6.
Generating Unit Name
Plant Type
Capacity (MW)
[Contains confidential information]
Table 6: Inventory of installed assets in Belgium
Source: Elia – data confidential. For internal use by the CREG only.
Table 7 contains a summary of installed plant by technology type.
Plant Type
Capacity (MW)
[Contains confidential information]
Table 7: Installed capacity by plant type
Source: Elia and Frontier
Asset efficiencies by vintage of plants
We have assumed a range of efficiencies for different plant types, which vary
according to the date when the station was constructed. Efficiencies for the
main technology types (OCGT, CCGT, hard coal, lignite coal, heavy oil, diesel
and turbojets) are illustrated in the figure below. The efficiency of each plant on
the system is therefore estimated on the basis of the year in which it was
constructed.
Annexe 2: Data
62 Frontier Economics | March 2006
Plant eifficiencies over time
60%
50%
40%
30%
20%
CCGT
Coal_hard
Coal_lignite
Diesel
Heavy_oil
OCGT
Turbojet
10%
19
75
19
77
19
79
19
81
19
83
19
85
19
87
19
89
19
91
19
93
19
95
19
97
19
99
20
01
20
03
20
05
20
07
0%
Plant efficiencies over time
Source: Frontier Economics estimates
Fuel prices, taxes and carbon costs
To estimate the marginal cost of generating assets we combine the assumed
efficiencies with estimated fuel prices, taxes and carbon costs. We take the Argus
September 2005 contract price for all the oil products which we then index to the
NYMEX Futures Light Sweet Crude up to 2007. For gas, we take the average of
the Zeebrugge quarterly forward prices for 2006 index it as per oil products by
season. For coal, we take the Argus September 2005 contract price and keep it
constant nominal. We also assume a constant carbon permit price of €22/tonne
CO2.
Annexe 2: Data
63 Frontier Economics | March 2006
Fuel type
2007 (€/MWh(th))
Carbon permit
cost (€/MWh(th))
Winter : 33.73
4.44
Gas
Belgium taxes
(€/MWh(th))
Summer: 21.89
Coal_hard
5.95
7.49
Coal_lignite
4.51
8.01
Heavy_oil
27.30
6.12
0.57
Diesel
61.41
5.86
1.86
184.24*
5.69
Kerosene
Table 8: Fuel prices, taxes and carbon costs
Source: Argus and Frontier estimates
* we assume the Kerosene price as three times the Diesel price
Availabilities
We assume that generating assets have expected availability below 100%,
reflecting planned and unplanned outages. Availability levels are different for
different technology types and for the summer and winter seasons, reflecting the
reliability of different generating technologies and the fact that planned
maintenance is typically scheduled for summer periods in regions facing peak
demand in the winter. The availability figures we have used are based on publicly
available Generating Availability Data System (GADs) run by the North
American Electric Reliability Council (see http://www.nerc.com/~gads/)
combined with information gathered from a range of assignments previously
undertaken by Frontier Economics. Our estimates of expected outage rates are
shown in the following table.
Annexe 2: Data
64 Frontier Economics | March 2006
Plant Type
Winter outages (%)
Summer outages (%)
CCGT
10%
14%
Coal_hard
7%
12%
Coal_lignite
4%
9%
Diesel
7%
12%
Heavy_oil
7%
12%
Light_oil
7%
12%
Nuclear
8%
25%
OCGT
10%
14%
Table 9: Availabilities by plant type
Source: GADS and Frontier Economics estimates
Special treatment of CHP units
The electrical output from a lot of generation capacity is determined by the
demand for heat. Such units produce heat as their main function and electricity
as a by-product. From a gaming perspective, therefore, the power that they
produce as the by-product from heat generation cannot be withdrawn for the
strategic purpose of raising prices. That amount of power should thus be
modelled as non-strategic (i.e., cannot be withdrawn) and bid at a marginal cost
of zero (i.e., the cost of selling the energy is zero because the electricity is
produced as a by-product).
Gathering information from public sources about the exact amount of capacity
that is heat driven on a plant by plant basis would be a costly exercise and would
not be easy to do properly within the project time frame allowed. However, not
modelling the amount of electricity that is produced as a by-product of heat
generation would grossly over-estimate the potential for market.
We have already described the differences in assumed availability but we make
further assumptions about heat driven plants. We assumed a higher proportion
of heat driven CHP in winter (70% of available CHP capacity) compared to
summer (50%). CHP capacity which is committed to run as a consequence of its
heat load is given a marginal cost of zero, ensuring that it runs at all times and is
not withdrawn from the market. The remaining CHP capacity is bid in with a
marginal cost based on the gas price and an assumed efficiency of 50%.
Annexe 2: Data
65 Frontier Economics | March 2006
Special treatment of Wind and Hydro units
We make special assumptions with regard to the output of constrained power
plants like run of river and wind plants. The generation from these plants is,
beyond planned and unplanned outages, much lower than their full capacity due
to the nature of the force (rivers in the first case, wind in the second) that drives
the electricity production. For this reason, we assume a much lower available
capacity than implied by planned and unplanned outages alone to reflect an
average over the day of the seasonal load factor. The data, presented in Table 10,
was collected from public sources and refined over the course of several
modelling projects.
Plant Type
Winter duration (%)
Summer duration (%)
Run of river
40%
55%
Wind
75%
87%
Table 10: Run or river and wind power station seasonal duration
Source: Frontier Economics estimates
Annexe 2: Data
Frontier Economics Limited in Europe is a member of the Frontier Economics network, which consists of separate companies based
in Europe (London & Cologne) and Australia (Melbourne & Sydney). The companies are independently owned, and legal
commitments entered into by any one company do not impose any obligations on other companies in the network. All views expressed
in this document are the views of Frontier Economics Limited.
THE FRONTIER ECONOMICS NETWORK
LONDON | COLOGNE | MELBOURNE | SYDNEY
Frontier Economics Ltd
71 High Holborn London WC1V 6DA
Tel. +44 (0)20 7031 7000 Fax. +44 (0)20 7031 7001 www.frontier-economics.com