EPJ Web of Conferences 98 , 0 6 00 1 ( 2015)

DOI: 10.1051/epjconf/ 201 5 98 0 6 0 0 1

C Owned by the authors, published by EDP Sciences - SIF, 2015


Economics of electricity
G. Erdmann(∗ )
Berlin University of Technology - TU 10587 Berlin, Germany

Summary. — The following text is an introduction into the economic theory of

electricity supply and demand. The basic approach of economics has to reflect the
physical peculiarities of electric power that is based on the directed movement of
electrons from the minus pole to the plus pole of a voltage source. The regular grid
supply of electricity is characterized by a largely constant frequency and voltage.
Thus, from a physical point of view electricity is a homogeneous product. But from
an economic point of view, electricity is not homogeneous. Wholesale electricity
prices show significant fluctuations over time and between regions, because this
product is not storable (in relevant quantities) and there may be bottlenecks in the
transmission and distribution grids. The associated non-homogeneity is the starting
point of the economic analysis of electricity markets.

1. – Electricity generation

As with many other goods and services, the demand for electricity is subject to daily,
weekly and seasonal fluctuations (see fig. 1). Electricity demand in winter periods is
usually higher than in the summer periods. In the winter term it has daily peaks during
the evening hours, while during the summer daily peaks occur at noon. At typical

(∗ ) Chair for Energy Systems, TU berlin. E-mail: Georg.erdmann@tu-berlin.de

7KLVLVDQ2SHQ$FFHVVDUWLFOHGLVWULEXWHGXQGHUWKHWHUPVRIWKH&UHDWLYH&RPPRQV$WWULEXWLRQ/LFHQVHZKLFKSHUPLWV
XQUHVWULFWHGXVHGLVWULEXWLRQDQGUHSURGXFWLRQLQDQ\PHGLXPSURYLGHGWKHRULJLQDOZRUNLVSURSHUO\FLWHG

Article available at http://www.epj-conferences.org or http://dx.doi.org/10.1051/epjconf/20159806001

 EPJ Web of Conferences

Load [weekdays; 1000 MW]

January
60

September April
July
40

20
0 2 4 6 8 10 12 14 16 18 20 22
Time of day [h]

Fig. 1. – Daily electricity demand variations.

spring and autumn days the structure of the load profile is different again. The load
curves shown in fig. 1 refer to northern regions. In southern regions the load may look
quite different, due to a different role of electricity for heating and a higher electricity
consumption for air conditioning.
In view of the demand fluctuations, electricity suppliers would ideally use the following
strategies to match supply and demand:

– Storage of off-peak electricity for covering peak-load.

– Variation of supply quality, for example by interruptible load contracts with ap-
propriate customers.

– Price differentiation with respect to time.

These options are used, particularly in the segment of industrial customers. The Euro-
pean Union requires all retailers to offer an electricity tariff, where the price per kilowatt-
hour varies with time. Until today this variation is independent of the generation of wind
and solar electricity generation but with growing shares of these technologies a more
flexible tariff structure will be needed. Also due to missing large-scale electricity storage
facilities(1 ) and so-called smart meters allowing flexible time-of-use tariffs or — more
ambitious — real-time pricing(2 ) a full compensation of load fluctuations is not possible.

(1 ) At the present, a lithium battery with a mass of one kg can just store one kWh of electricity.
(2 ) In some sense electricity is still “too cheap to meter”. This may change if intelligent meters
(smart meters) become more economical.

06001-p.2
 LNES 2014

As a consequence of the non-storability of electricity and the minor role of load man-
agement measures, the generators must adjust their production in order to continuously
synchronize supply and demand. How this task is accomplished today and how it is
different from pre-liberalized markets is now discussed.
.
1 1. Power generation technologies. – Most power plants generate electricity according
to the principle of magnetic induction. The electric side of such power plants consists
of a stator (coil) and a rotor (rotating electromagnet). The generator produces alter-
nating current with a frequency of 50 or 60 hertz and an electric voltage between 6 and
21 kilovolts. For the propulsion of the generator, kinetic energy is required, which can
be supplied by a turbine using the heat from combustion. Alternative designs use hy-
dropower and wind power. Another approach is to use electrochemical principles, e.g.
photovoltaic cells and fuel cells. These devices produce direct current. The electricity
must pass through an inverter in order to feed the power into the public AC grid.

– Steam turbine power plants are based on a thermal cycle. The heat from burning
fossil fuels is used to transform water into high-temperature and high-pressure
steam. The steam is fed to a turbine, where it expands and generates rotational
energy, which drives the generator. The remaining energy dissipates through a
steam condenser into the environment. With present technologies up to 46% of the
thermal energy contained in the fuel can be converted into electricity. All fossil or
nuclear energy sources can be used to generate the required heat. Solar thermal
and geothermal power plants function according to the same thermodynamic cycle.

– Gas turbine power plants are based on the combustion of a gaseous fuel in the
turbine. In the cold part of the turbine a gas/air mixture is compressed and fed
to the combustion chamber. In the combustion chamber the fuel mix becomes ig-
nited, producing high temperatures and pressures. In the hot part of the turbine
the gas/air mixture expands and generates rotation that drives the generator. De-
pending on the pressures and temperatures, thermal efficiencies of up to 42% can
be reached (known as the Carnot efficiency).

– Combined cycle gas turbines integrate the principle of gas and steam technologies.
The exhaust gas from a gas turbine is fed into a waste heat boiler where it is used
to produce steam, which is subsequently used to drive a steam turbine. Combined
cycle power plants can reach fuel efficiencies of more than 60%.

– Hydropower plants transform the kinetic energy from flowing water into rotational
energy. Run-off-river plants use elevation differences of only a few meters, while
the elevation difference of storage power plants can reach several hundred meters.
Storage power plants use the potential energy of water collected in an (artificial)
lake in the mountains only once. In contrast, pump storage power plants can use
this energy several times. Water is collected in a lower basin and pumped back into
an upper basin by using cheap excess power. While run-off-river plants produce

06001-p.3
 EPJ Web of Conferences

electricity more or less continuously, storage and pumped storage power plants are
used for interval and selective peak-load generation.

– Wind power plants use the kinetic power of wind in order to drive a generator. Like
hydropower stations, they do not need fossil fuels and therefore have no greenhouse
gas emissions. Their variable costs are also comparatively lower. However, the
generation of these plants depends on the wind availability. The capacity factors
of onshore wind turbines are usually below 30%, while offshore wind installations
can exceed capacity factors of 40%.

– Photovoltaic power generation is based on the electrochemical transformation of

sunlight into a direct current of electrons. An inverter is needed if the output
must be alternating current (AC). Like wind power, photovoltaics represents an
environmentally friendly way of power generation, but this form of generation is
again intermittent. The capacity factor of photovoltaics lies between 10 and 25%.
With growing shares of photovoltaic capacities, additional refinement concepts (e.g.
batteries, load management) are required for satisfying electricity demand.

– Fuel cells produce an electron current by using electrochemical principles. One

design option is the use of a hydrogen rich gas that flows over the anode and
gives off electrons at the presence of a noble metal catalyst. Between the anode
and the cathode is a proton-conducting membrane (Proton Exchange Membrane
or PEM). Oxygen receives electrons at the cathode side of the cell and reacts with
the proton to produce steam. As the operation temperature is below 100 ◦ C, the
power generation of the fuel cell can be quite dynamic.
.
1 2. Power plant dispatch on day-ahead markets. – Under perfect supply competition
a power plant operator looking to maximize profits will generate electricity as far as
the (short term) marginal cost of the power plant is smaller than the market price he
receives from selling the electricity. Otherwise, the power plant will not run and produce
electricity. The formal analysis is quite simple. Assuming an individual supplier who
maximizes its periodic profit

Π(Q) = p · Q − C(Q)

(difference between the periodic revenues p · Q and the periodic costs C(Q)) the solution
is to be found by setting the derivative of the profit function with respect to the produced
quantity Q equal to zero:

dΠ d(p · Q) dC dC dC
= − =p− =0→ = p.
dQ dQ dQ dQ dQ

In case of atomistic competition the individual producer cannot influence the sales price
p so that it is given: p = p. Thus, the supplier should adjust the individual production
Q so that the marginal cost — the cost of the last produced unit — is equal to the fixed

06001-p.4
 LNES 2014

sales price. If the marginal cost is smaller, the supply should be expanded, otherwise it
should be reduced. Any sale at a price below the marginal costs is economically stupid.
Depending on the type of plant, marginal costs are the sum of several cost components:

– A large share of the marginal costs consists of the costs of fuels needed to generate
one unit of electricity. There is a clear ranking of power plants using the same type
of fuel. Growing fuel efficiencies lead ceteris paribus to lower fuel cost per unit of
electricity produced. Thus, modern high efficient power stations have an advantage
over older, less efficient facilities.

– Fossil fuels used for power generation are subject to the European system of CO2
emission allowances. The marginal cost of electricity generation depends on the
quantity of CO2 emitted (in tons of CO2 ) when producing one MWh of electricity
multiplied by the CO2 market price.

– If the output of thermal power plants falls below the optimal design point, efficiency
losses occur and must be taken into consideration when calculating marginal costs.

– Thermal power stations need to be in operation temperature before being able to

generate electricity. This determines turn-on and shut-off costs that belong to the
marginal costs.

Competition between generators occurs on markets. The most important market is

the day-ahead market. Products traded on this market are power deliveries for single
hours of the next day as well as blocks of these contracts(3 ). The European Power Ex-
change EPEX offers the most liquid day-ahead market in Europe. It is a uniform price
auction market. All bids and asks for an individual hour are collected and aggregated
to an hourly supply and a demand curve. After gate closure, the energy exchange calcu-
lates the hourly market clearing price (MCP), which is equivalent to the intersection of
both curves.
The optimal supply strategy of each power plant operator is to offer electricity at
(short term) marginal cost. In the aggregate supply curve all power plants are thus
ordered according to their merit order. Figure 2 shows two wholesale market equilibriums,
each describing a different demand situation. During off-peak hours (night and weekends)
the hourly price is low, and only power plants with small marginal costs are dispatched,
while the more expensive plants remain off-line. During peak hours additional and more
expensive power plants (in terms of marginal costs) are required to cover the demand.
Therefore the MCP is high.
Generators using renewable energy sources have rather small marginal costs (with the
exception of biogas). If these generators would offer their capacities under competitive
conditions and sell at their marginal costs, hydropower, wind and photovoltaic would

(3 ) While in Europe the day-ahead wholesale market is regarded as a spot market, the North
American discussion defines this market as a future market.

06001-p.5
 EPJ Web of Conferences

Peak load
Marginal costs,
day-ahead price MCP
100 [Euro/MWh]

80 Off-peak load

60

40 Oil and gas

Hard coal
20 Lignite

0
20 30 40 50 60 70
Available load without wind power [1000 MW]

Fig. 2. – Merit order of power plants.

stand at the beginning of the merit order dispatch. If renewable electricity generation
increases, the MCP will shrink below the marginal costs of thermal power generators, in
particular gas fired power plants. This is the “merit order effect of renewables”.
.
1 3. Intraday markets. – Because electricity cannot be stored, all market participants
must close their open positions prior to each execution period (e.g. quarter hour in-
tervals). Whether the positions are open or not is determined by generation and load
forecasts. These forecasts depend on many parameters and must regularly be revised
whenever new information is available, in particular when customers are lost or acquired.
A significant source of quarter hour forecasting errors is power generation from wind
and photovoltaics. Even when using sophisticated forecasting models, the aggregated
day-ahead forecast errors can exceed several thousand megawatts, if the cumulated share
of these capacities is large(4 ). Hour-ahead forecast errors are much smaller. Intraday
markets that allow trades up to an hour before execution (or less) can be used for
closing deviations between day-ahead and intraday forecasts. Several intraday market
concepts exist:
– Pool-type markets use the schedules and bids that have been submitted for day-
ahead scheduling to recalculate the optimal dispatch of power plants close to real-
time delivery.
– Bilateral trade over an energy exchange without an institutional link to the day-
ahead market is the model applied in most European markets.
The liquidity of intraday markets is relatively small, particularly if it is organized as
a continuous trade market. Still the observed intraday price volatility appears not to

(4 ) Average day-ahead wind forecast errors (RMSE) are presently in the range of 6 to 7%.

06001-p.6
 LNES 2014

Electric load [[1000 MW]]

Ordered load duration

Peak load Base load

0 1000 2000 3000 4000 5000 6000 7000 8000

Annual operation hours
Fig. 3. – Planning of power plants under monopoly.

be excessive compared to the day-ahead price volatility. If more renewable electricity is

offered than predicted in day-ahead forecasts, some power plant operators replace their
generation by purchasing renewable electricity. If the renewable electricity generation
falls short of day-ahead forecasts, conventional plants have to produce more. In both
cases the marginal costs of the respective plant determines the intraday market price
providing that this market is efficient. Imbalances that remain after gate closure of the
intraday market are to be covered by the transmission grid operator in the corresponding
control area. It delivers (positive or negative) balancing power as necessary which he has
been purchased on the market for regulation power. The expected balancing power prices
(real time price) should represent a limit for intraday prices.

2. – Power plant investments

Before the market liberalization in Europe, power generation was organized in closed
areas without competition. Under these conditions investment planning is straight for-
ward. First, the monopolistic utility has to forecast the ordered load duration for each
year of the planning period (see fig. 3). The load duration orders all 24·365 = 8760 hours
according to the expected load. The second step of the monopolistic planning process is
to fill the area below the duration curve by rectangles that represent each available power
plant. The height of the rectangle corresponds to the rated load. The length corresponds
to the expected load factor. This reflects that some power plants are designed as peak
load units with less than 3000 planned operation hours per year and others as base load
units with more than 6000 planned annual operation hours. This planning step also
takes into account power plants that will be retired during the planning period. If the
area below the ordered load curves cannot completely be covered by the available power

06001-p.7
 EPJ Web of Conferences

plants, including a predetermined reserve margin, additional power plants are needed. To
finance the required investments, the regulator may allow power tariffs to increase. Re-
garding the high willingness to pay for electricity, the risk of declining electricity demand
is moderate as long as the surcharge remains moderate.
To support regulators, the theory of optimal electricity tariffs was developed by
economists including Botteux [1], Steiner [2] and Kahn [3]. The starting point is
the marginal willingness to pay (WTP). A distinction is made between off-peak and
peak demand. Deducting the marginal costs of electricity generation results in a WTP
for capacity. Because generation capacity can be regarded as a public good, the WTP
is horizontally (instead of vertically) aggregated in order to get the total marginal WTP
for capacity. The optimal generation capacity is equal to the point in which the marginal
cost for capacity (in particular capital uses cost, i.e. amortization and interest of the in-
vested capital) is equal to the marginal WTP for capacity. With the optimal generation
capacity the optimal power plant investment and capacity costs for both base load and
peak load supply can also be determined. For the optimal capacity cost allocation to the
two demand segments, regulators can use the Ramsey pricing model, which is equivalent
to the optimal structure of grid fees.
One of the aims of electricity market liberalization is effective competition in the gen-
eration segment of the electricity value chain. Once competition is established the market
determines optimal the power generation price. Thus the regulator should restrain from
imposing Ramsey pricing on the generation industry and let the markets work without
intervention. However electricity competition is organized, the grid always remains a
natural monopoly and must be regulated. The theory of optimal regulation and Ramsey
pricing remain relevant for the grid segment of the electricity value chain but not for the
generation segment.
In competitive markets for power generation the availability of sufficient power plant
capacities is a key condition for a safe and reliable electricity supply. Accordingly, incum-
bent generators should avoid investing into additional capacities, even if these capacities
may be indispensable in the future. Once newcomers — independent power generators —
have the right to access the market they will invest into generation capacities if economic
opportunities exist. In this case incumbents risk loss of market share if they abstain
from investing in order to keep generation capacities short. Therefore, no systematical
underinvestment should be expected if generation markets are competitive and market
access is open.
Any power plant investment is economical if the expected cash flows or contribution
margins exceed its capital expenditures. A deeper analysis is based on fig. 4, which shows
the ordered price duration line of the (expected) hourly day-ahead prices in a particular
year of the planning period. The expected annual operation hours of this year are deter-
mined by the intersection of the variable unit cost with price duration. The area above
this point is equivalent to the annual cash flow. Clearly the investment is expected to be
economical if time and price spikes occur because spikes contribute to the margin that is
necessary to finance the capital expenditures of the plant. Should the regulator limit price
spikes on the hourly power markets, sufficient generation investments cannot be expected.

06001-p.8
 LNES 2014

Day-ahead
y pprice ph [[EUR/MWh]]

Are these contributions sufficient

for risk adequate financing of
the power plant investment?

cvar Marginal costs of a new power plant

Annual market clearing

price duration

0 1000 2000 3000 4000 5000 6000 7000 8000

Expected annual Annual operation hours
running hours hexp

Fig. 4. – Merit order of power plants.

The day-ahead power prices vary from year to year. Therefore, the above calculation
must be repeated for each year using the hourly price forecasts. To reflect the riskiness
of the day-ahead prices, the distribution of the expected annual cash flows should also be
included e.g. by Monte Carlo simulations. This leads to a probability distribution over
the annual cash flows. The variable costs of the power plant, including mostly fuel costs
and the price of emission allowances, can be secured by long-term contracts so that these
price risks become irrelevant. Skipping this assumption, Monte Carlo simulations based
on the expected distribution of fuel and CO2 prices are even possible.
On the other hand, power price risk could be hedged by power sales on forward and
future markets. If the forward markets for electricity, natural gas, coal and emission
rights are sufficiently liquid, the economic assessment of power plant investments can be
based on the price signals from these markets. Therefore the following expressions have
become popular:

– Spark spread for gas plants: the difference between the peak load electricity price
future and the gas price future corrected by the fuel efficiency of the gas plant.

– Clean spark spread for gas plants: derived from the spark spread by subtracting the
specific CO2 costs of gas power generation based on the future prices of emission
rights.

– Dark spread for coal plants: the difference between the base load electricity price
future and the coal price future corrected by the fuel efficiency of the coal power
plant.

06001-p.9
 EPJ Web of Conferences

– Clean dark spread for coal plants: derived from the dark spread by subtracting the
specific CO2 costs of coal power generation based on the future prices of emission
rights.
These spreads can be calculated for each year given a liquid forward/future market for
which transparent prices exist. Otherwise, the spreads must be estimated by forecasting
methods. Once the planner can estimate the clean spark (dark) spreads, he can assess the
economics of a new gas (coal) power station by comparing the spreads with the annual
capital and fixed costs of the plant.
An extension of this approach is based on the real options theory. According to this
theory, a gas power plant can be regarded as a call option on the clean dark spread.
Once the new capacity is available, their operators use this option whenever the clean
dark spread is positive, namely by producing as much electricity as possible. The inter-
pretation of power plants as real options implies that their value exceeds the value of
the underlying (net present value of the cash flows) by an option premium. This option
premium increases with the volatility of the cash flows and accordingly the expected
volatility of the hourly spot prices. This again highlights the importance of spot market
price spikes for power plant investment under competition.
It is reasonable to assume that a period of low plant investments would lead to tight
electricity supply and thus higher spot and futures prices both in the base load and the
peak load segment. The price increase creates an investment signal (once a threshold
is passed) and ends the period of insufficient generation investments. However, one
problem remains. If the lag between the decision to invest and the completion time of
the investment is sufficiently short, the market mechanism should guarantee a stable,
long term power supply(5 ). Otherwise, long term price fluctuations instead of a stable
power price trend may result.

3. – Capacity markets

This raises the question as to whether or not the liberalized European electricity mar-
ket is able to continuously secure the electricity supply by providing sufficient generation
capacities. As electricity cannot be stored at significant volumes, a safe and reliable
power supply requires some excess and reserve generation capacities. The excess capac-
ities cover unexpected loads, while reserve capacities are required to back-up scheduled
and unscheduled power plant outages. Some further generation capacity is required for
supplying regulation power and other grid services, in particular redispatch of genera-
tion in case of grid bottlenecks. As a consequence, reserve margins of about 10% of the

(5 ) At the beginning of market liberalization, long-term power contracts had been an obstacle
for benefits from liberalization to become real. Other than in the US, the European situation at
that time was characterized by excess capacities so that planning for new investments was not
an issue. Once this changes, long-term contracts are inevitable for power markets to efficiently
generate power plant investments. The alternative would be state guarantees, subsidies or other
types of market interventions.

06001-p.10
 LNES 2014

maximum load are usually assumed to be necessary to secure electricity supply at any
time (system adequacy), but this reserve margin may not be offered in the regular power
market. Mechanisms must therefore be installed to finance the reserve margin.
In a market with a single generator the regulator requires the monopolistic utility to
secure the necessary excess and reserve capacities and allows their financing through elec-
tricity tariffs. Competitive markets will not automatically provide appropriate capacities
because supply security has the character of a public good. With the implementation
of electricity market competition the regulators have developed several mechanisms in
order to solve this problem:

– The regulated Transmission System Operator (TSO) purchases capacities that are
needed for supplying balancing power. The plant operators receive capacity pay-
ments determined by pay-as-bid auctions.

– In a fully liberalized market with retail competition each wholesale market par-
ticipant (balancing group managers), in particular each retailer, has to purchase
balancing power from the TSO to adjust for imbalances between supply and de-
mand in its balancing group. If the imbalances are excessive, the regulator may
impose a fine. In extreme situations the TSO can even prohibit the balancing group
manager from using the grid.

– Over-the-counter capacity markets exist that provide back-up power, for example
in cases of plant outages. The demand comes, among others, from the operators
with a small portfolio of generation plants.

– In some countries regulators purchase emergency capacities that are released if the
power system threatens to fail. The regulator may organize a public auction to
acquire appropriate capacities and decides about the conditions under which these
capacities should be released. Because emergency capacities cannot participate
in the regular day-ahead market, this mechanism increases the regular wholesale
power prices, but when the emergency capacities are released, the peak prices
come under pressure. Obviously, the criterion for releasing emergency capacities
determines the maximum hourly price. State intervention thus can reduce the
incentives for private power plant investments.

The situation of capacity shortages can be explained by the simplified merit order
function shown in fig. 5. Point A shows a regular market situation in which the marginal
costs of the last available power plant (i.e. excluding the reserve margin) sets the market
clearing price p1 . All power plants to the right of the marginal plant also receive the price
p1 and earn infra-marginal rents above their marginal costs. The last unit in the merit
order has marginal costs of p2 . If all available power plants together are not sufficient
to cover the total demand, the market clearing price will spike to p3 > p2 leading to a
capacity surcharge. The price difference p3 − p2 , is also called “capacity rent”.
If such situations would happen regularly, the capacity rent would stimulate genera-
tion investments. It is important to note that in point B of fig. 5 the power price is above

06001-p.11
 EPJ Web of Conferences

a et Clearing
Market C ea g Price
ce [Euro/MWh]
[ u o/ W ]

B
p3
Demand
Capacity surcharge (scarcity rent)
p2

p1
A
Inframarginal rent

short-term
marginal cost

Load [MW] Available capacity

Fig. 5. – Electricity market under capacity shortage.

the marginal cost of any plant in the merit order, but still this situation is not the result
of misusing market power. In liberalized markets the wholesale power price can, and
must from time to time, exceed the marginal costs. But due to the public involvement
(suspicion that generators manipulate market prices and obtain excessive returns) the
regulator may come under political pressure to intervene in favor of lower power prices.
Thus, for political reasons the energy-only market may not generate sufficient revenues
for financing power plant investments — the so-called missing money problem. To cor-
rect for this problem, capacity payments to (certain) power generators are proposed,
that are financed outside the energy-only market by a capacity levy imposed on (all)
final electricity consumption.
Alternatively, a decentralized capacity market design is possible. Retailers and eligible
industrial customers must hold capacity certificates before they can purchase electricity
at the wholesale market. Generators, electricity storage operators and customers with
interruptible loads are on the supply side of the decentralized capacity market. The
capacity price is the result of supply and demand and will be high in periods with tight
capacities (weak renewable electricity supply), but it can also be zero if sufficient capacity
is available (strong renewable electricity supply). Such a decentralized market requires
again an institution that supervises and controls the behavior of market participants,
but there is no central planner who determines the necessary capacity as would be the
case under centralized capacity mechanisms.
In power systems with high shares of intermittent renewable energy sources another
concept could be implemented. This concept is based on the customer’s choice to select
the desired quality of electricity supply. A regular power supply contract could allow

06001-p.12
 LNES 2014

the retailer to cut the supply during one or more hours per day if capacities are short
(usually during evening hours). Electricity customers can opt for a non-interruptible
power supply but they have to pay for the individual cost associated with this service.
The expectation is that such a concept would initialize a variety of innovations so that
the final result will be cost savings compared to capacity markets with politically planned
generation (or electricity storage) investments.

4. – Power market design options

Before liberalization, electricity markets had usually been organized in closed con-
cession areas in which the state regulator allowed only one single company to distribute
electricity. These local or regional distributors were allowed to sign long-term and full-
service electricity agreements with upstream power generators. According to these con-
tracts, utilities had the exclusive right to deliver electricity to the distribution company
in the concession area, but without directly supplying final electricity users. As a result,
all final users were obliged to purchase electricity from a single distributor, and no compe-
tition between generators was possible. In return distributors were not allowed to refuse
any customer asking for electricity supply. Additionally, electricity prices above regu-
lated tariffs were not allowed, but large industrial customers could benefit from special
power purchase agreements at prices below the regulated tariff.
When implementing competition, the first observation is that a vertically integrated
utility industry does not necessarily represent a natural monopoly. This has first been
shown by Christensen and Greene [4] in the USA. According to their study, economies
of scale exist, but with regard to the already achieved size of the power market unit
costs of vertically integrated companies cannot shrink further if output levels continue
to grow. Similar results have later been found by Thompson and Wolf [5]. Only
the transmission and distribution part of the electricity value chain represent a natural
monopoly. Liberalization should therefore focus on the other parts of the value chain,
in particular generation, trade and marketing, and regulate the electricity grids so that
competition among generators and (eventually) distributors becomes possible.
Guided by these principles, several market design options are possible (see [6, 7]):

– Single-buyer market: All generated electricity has to be offered to a centralized

(usually government managed) pool, the single buyer. The pool operator calcu-
lates the cost optimal dispatch so that the aggregated load is always covered by
the least expensive combination of available power plants. As is the case under
regulation, retailers and eligible customers have no alternative to buying electricity
from the pool. No company can purchase power through bilateral contracts. This
helps incumbent retailers to keep their monopolistic market position, but if the
retail market is legally opened for newcomers, this market is contested and tariff
control can be abandoned. Supply side regulation is necessary as generators may
manipulate their offers. The pool operator can also allocate side payments in order
to stimulate investments into operating reserves.

06001-p.13
 EPJ Web of Conferences

– Mandatory power pool: All generators must offer the generated electricity to a
centralized pool. This model functions like an energy exchange as retailers and
eligible customers can also submit individual bids to the pool in order to satisfy
their power needs. The bids are based on the load forecast and can reflect demand
side strategies and auto generation strategies. As forecasting errors are unavoidable,
a separate balancing mechanism must be in place. The pool calculates the market
clearing price, which is relevant for all buyers and sellers. Again, manipulation of
generators can occur so that price regulation is necessary.

– Free wholesale competition: Power can be traded through bilateral long term con-
tracts (forward and future contracts with physical delivery) and through an energy
exchange. This model offers more options to generators and retailers and is there-
fore more efficient. However, the voluntary participation at the energy exchange
leads to a reduced liquidity and increased efforts for controlling the non-distorting
behavior of generators are necessary.

– Fully liberalized market with retail competition: This is the most sophisticated
model of liberalized electricity markets. In addition to the power exchange, all types
of bilateral power trades are allowed. Generators and retailers can develop and
apply a manifold of contract models (portfolio management). The establishment
of this system takes usually several years but can then work quite efficiently. In
particular new retail companies have a chance with innovative products that beat
the retail market price. But as retail margins are quite small, many authors argue
(e.g. [6], p. 26ff, [7], p. 713ff) against the benefits of retail competition. However, the
European Electricity Market Directives 96/92/EC, 2003/54/EC and in particular
2009/72/EC require the implementation of this model in Europe (see [8]).

REFERENCES

[1] Boiteux M., J. Business, 33(2) (1956) 157.

[2] Steiner P. O., Q. J. Econ., 71(4) (1957) 585.
[3] Kahn A., The Economics of Regulation (MIT Press, Cambridge, MA) 1970.
[4] Christensen L. R. and Greene W. H., J. Polit. Econ., 84 (1976) 655.
[5] Thompson H. G. jr.and Wolf L. L, Land Econ., 69 (1993) 234.
[6] Stoft S., Power System Economics. Designing Markets for Electricity (Wiley, New York)
2002.
[7] Bhattacharyya S. C., Energy Economics. Concepts, Issues, Markets and Governance
(Springer, London) 2011.
[8] EU-Commission: Directive 2009/72/EC of the European Parliament and the Council of
13 July 2009 concerning common rules for the internal market in electricity and repealing
Directive 2003/54/EC, Official Journal of the European Union L 211/55-93 (2009).

06001-p.14