CONGESTION MANAGEMENT IN THE NORDIC

POWER MARKET – COUNTER PURCHASES AND

ZONAL PRICING

Mette Bjørndal1
Kurt Jørnsten2
Virginie Pignon3

Abstract
In this paper, we investigate methods for managing congestion on the grid in the
Nordic power market. Specifically, we have considered the differences between using
counter purchases as opposed to pricing out the transmission constraints of the grid.
We show that the specific method used for congestion management greatly affects prices
and therefore the surplus of the various agents, including the system operator. This
means that the market agents may have preferences for one method, and take actions
in order to influence which method is to be used. Based on this, we have studied the
incentives and possibilities of “moving” capacity constraints, and the effect this has
on system performance. We have also looked into the differences between various
pricing schemes, i.e. optimal nodal prices versus optimal zonal prices. The effects that
are demonstrated by the examples in this paper are especially relevant when designing
coordination mechanisms and regulation for integrated markets, like the (emerging)
European electricity market.

1. INTRODUCTION

The choice of an effective method to allocate scarce transmission capacity is

a central public policy decision in the path towards the creation of a genuinely
common market for electricity in Europe. Congested paths may indeed
impede access to the market for generators located far from the consumption
centers, thus reducing the benefits due to the competition for electricity
consumers, creating potential market power for incumbent generators isolated

1
Department of Finance and Management Science, Norwegian School of Economics and Business
Administration, Helleveien 30, N-5045 Bergen, Norway, Tel: (+47) 55 95 92 90,
Fax: (+47) 55 95 96 50, E-mail: mette.bjorndal@nhh.no.
2
Department of Finance and Management Science, Norwegian School of Economics and Business
Administration, Helleveien 30, N-5045 Bergen, Norway, Tel: (+47) 55 95 95 52,
Fax: (+47) 55 95 96 50, E-mail: kurt.jornsten@nhh.no.
3
Université Paris I, Centre ATOM, 106-112 Boulevard de l’Hôpital, 75634 Paris Cedex 13, France,
Tel: (+33) 1 44 07 83 21, Fax: (+33) 1 44 07 83 20, E-mail: vir_pignon@yahoo.fr.

Journal of Network Industries, Volume 4 (2003), No. 3 271

 Mette Bjørndal, Kurt Jørnsten, Virginie Pignon

from competition, and limiting the benefits from the European market for
electric power. Seven years after the adoption of the European Directive 96/92
(European Commission (1996)), there is nevertheless no consensus on the
method that should be favored. The sole recommendation in fact is that scarce
transmission capacity should be allocated through a market-based mechanism
(Boucher and Smeers (2002)).

Theoretically, the most efficient method to signal the cost minimizing use of
the grid to generators and consumers is by means of nodal pricing (Schweppe
et al. (1988) and Hogan (1992)). Nodal pricing throughout the European
electric power grid would result in electric power prices defined at each node4
of the grid depending on the marginal generation cost5 and the marginal
congestion cost incurred to transmit electric power to consumers. However,
the transmission pricing methods used in the European countries have not
converged towards nodal pricing and are still characterized by a relatively high
level of diversity. This situation is mainly due to the fact that the European
Directive 96/92, while proposing general principles for the establishment of
a European electricity market, granted a high level of freedom to the countries
for the implementation of these principles.

In this context, the Nordic power market is often held up as an organizational

model that should be followed by the rest of Europe (European Commission
(2003), ETSO (1999, 2002)). This market, with one (non-mandatory day-
ahead) energy market, run by NordPool, covers indeed 5 trading zones, i.e.
Norway, Sweden, Finland and Denmark (split into two areas). The creation
of the Nordic power market implied the implementation of congestion
management methods to alleviate the congested paths that may appear at the
borders between zones, but also inside each zone6. Nevertheless, the congesti-
on management methods have not been standardized for the whole Nordic
market, and differences remain according to the location of the congested
paths. Two main methods are used:
– Zonal pricing7 for the large and long-lasting internal constraints in Norway,
but also for congestion at the borders between different control areas, and
– Counter purchases8 for the internal constraints in Sweden, Denmark and
Finland, and also for the management of smaller internal congestions in
real time in Norway.

4
The nodes are physical connection points of the grid. They represent the points where generators
inject energy and consumers withdraw energy from the grid.
5
At least, without market power. Market power would indeed introduce a difference between the
marginal generation costs and the energy prices offered by the generators.
6
See for instance Westre (1995), Wangensteen (1997), and Bjørndal (2000).
7
Also named market splitting (refer to Knops et al. (2001) and De Vries and Hakvoort (2002)).
8
Also named counter trading (refer to Knops et al. (2001) and De Vries and Hakvoort (2002)).

272 Intersentia
 Congestion Management in the Nordic Power Market

From an economic and institutional point of view, this regional power market
seems particularly relevant for the rest of Europe. Scarce transmission capacity,
whether within or between countries, is indeed allocated through market-
based mechanisms. At the same time, the Nordic organization for electric
power trade allows some diversity in the transmission pricing rules adopted
within each country, while entailing energy flows consistent with the physical
constraints of the grid at the regional level.

However, this paper shows that the use of different transmission pricing rules
may raise incentive problems for the market participants, i.e. electric power
generators and consumers, and to the transmission system operators9. As
shown in De Vries and Hakvoort (2002), zonal pricing as well as counter
purchases may be, from a theoretical point of view, economically efficient in
the short term, but they differ in the distribution of the congestion costs. The
aim of this paper is to investigate to what extent the use of both mechanisms,
characterized by different distributive consequences, is sustainable.

This question becomes all the more important as we find ourselves at the stage
of the electricity deregulation process where the US federal regulatory
commission is striving for a standardized market design based on nodal pricing
in the whole country. Precisely, we highlight here that the simultaneous
implementation of zonal pricing and counter purchases may give an incentive
to “cheat” on the rules. Thus, we illustrate by means of simple numerical
examples that it is possible to “fake” a transmission constraint so that the real
one could be managed, while benefiting from one particular congestion
management method. Numerical examples do not enable one to deduce
general rules concerning the efficiency of the congestion management
method, but they usefully illustrate that “faking” a constraint on a given
transmission line, may be highly beneficial for some market participants and
transmission system operators. The relative benefits of “moving” a constraint
are highly dependent on the pricing rule used, as well as the network upon
which it is applied.
The rest of the paper is organized as follows. Section 2 describes the two
congestion management methods used in the Nordic countries. Section 3
provides an example of how it is possible to “move” a constraint from one
transmission line to another. In the example we assume a full nodal pricing
scheme. In section 4, we assume zonal pricing is being used and see whether
the results are affected by the pricing scheme. In section 5 we analyze an
extended network to obtain more flexibility for the zonal prices. Thus, it is
possible to check the effects of increasing the number of zones. This increase

9
The transmission system operator is the entity in charge of the transmission grid on an area often
delimited by the political borders of the countries. Its main task is to coordinate the energy flows
that result from the commercial transactions between the energy consumers and generators.

Journal of Network Industries, Volume 4 (2003), No. 3 273

 Mette Bjørndal, Kurt Jørnsten, Virginie Pignon

makes the example more realistic for practical implementation. Section 6

offers concluding remarks.

2. ZONAL PRICING AND COUNTER PURCHASES

Both zonal pricing and counter purchases allow the allocation of scarce
transmission capacity through market-based mechanisms. Nevertheless, these
market-based mechanisms differ according to their trading process, and the
differences in the trading process have distributive consequences for the
market participants and the transmission system operators.

The trading process for zonal pricing works approximately as follows:

1. Based on the supply and demand schedule bids given by the market
participants to the energy market, the market is cleared while ignoring any
grid limitations. It results in the system price for energy and the amount
of electricity traded.
2. The transmission system operators simulate the net exchanges resulting
from the unconstrained price and quantity equilibrium on a model of the
grid. If these exchanges induce flows overloading transmission lines, the
nodes of the grid are partitioned into different zones on either side of the
bottlenecks.
3. A new pool price is determined in each area from the initial bids of the
energy market, taking into account the maximum transfer capacity between
the areas10.

Thus, in accordance with the supply and demand curves for electric power,
congestion is relieved through a market mechanism. This mechanism results
in a revenue for the transmission system operator, equal to the energy price
difference between the areas multiplied by the amount of energy transmitted
across the zone-boundary. In case of congestion, it results in energy prices
differing between connection areas, and the method can be considered as an
approximation of a fully-fledged nodal pricing scheme. The characteristics
of the zonal pricing approach are studied in Bjørndal and Jørnsten (2001).

Counter purchasing is a completely different method as regards the trading

process, as well as achieving completely different results for the market
participants and the network operator. In short, it consists in constraining off
some generators, which are rather “ill-placed” on the grid as regards the

10
As shown in Boucher and Smeers (2002), the notion of transfer capacity as a measure of the
maximal flow of electricity from one area to another, is inherently ambiguous in an electric power
grid. The effective use of transfer capacity depends indeed on the precise origin and destination
of the transactions, even within the areas. To assess a transfer capacity, the transmission system
operators therefore have to make assumptions on the use of the whole grid.

274 Intersentia
 Congestion Management in the Nordic Power Market

congestion’s location, and constraining on “better-placed” generators so that

demand could be met. Some generators therefore have to reduce the amount
of energy generated, whereas others are called upon to produce more
compared to the result obtained after clearing the market without taking into
account the grid limitations. Thus, counter purchasing can be considered as
a market-based application of the mechanism formerly used by vertically
integrated utilities that used to charge one single electricity price for the whole
control area, and perform re-dispatching in order to balance supply with
demand, within all relevant constraints.

The trading process to implement counter purchases works approximately

as follows:
1. The first stage remains the same as with zonal pricing.
2. If the resulting power exchanges induce flows overloading transmission
lines, the network operators check where injections into the grid have to
be curtailed or increased, so that the congestion could be relieved.
3. These increases and decreases are implemented through markets (e.g. the
“balancing” market in Sweden and the “regulation” market in Norway),
separated from the day-ahead energy market. Agents offer adjustment bids
on these markets, the sole buyer being the system operator.
4. The system operator selects the less expensive bids for increases and
decreases. Thus, some generators may be constrained off and compensated
with the equilibrium price of the adjustment market for generation
reductions, whereas others are constrained on and receive the equilibrium
price for generation increases.

This mechanism induces costs for the system operator since he has to buy and
resell energy according to the adjustment bids. These costs are distributed
among network users through the fixed charges of the network tariff, to
generate the (regulated) revenue of the transmission system operator. For the
generators and consumers, the equilibrium price on the energy market is not
affected by the congestion.

Recently, consumers have been allowed to participate in the regulation market,

and there is also a joint regulation list for the Nordic market. Moreover, the
regulation market is applied both in order to secure momentary balance
between supply and demand, and for congestion management (“special
regulation”). For the latter it is possible to pick bids not only based on price,
but also based on the effect the generation or load has on the specific
congestion in question. These changes serve to improve the potential
efficiency of the counter purchase method.

Even if zonal pricing and counter purchasing both are market-based methods
to allocate scarce transmission capacity in the grid, the description of their

Journal of Network Industries, Volume 4 (2003), No. 3 275

 Mette Bjørndal, Kurt Jørnsten, Virginie Pignon

trading processes highlights many differences in their distributive consequen-

ces. With zonal pricing, congestion management may result in additional
revenues as compared to a situation without any constraints. With counter
purchases, the transmission system operator incurs the cost brought about by
the adjustments in energy generated and consumed. Whether the system
operators retain these revenues/costs depends on regulation, and it may be
so in revenue cap regulation where permitted income is not determined by
incurred cost, or in light-handed regulation with (wide) limits on the allowed
returns.

The distributive consequences for the other market participants depend on

their location on the grid. Counter purchasing entails a higher average fixed
network tariff for the users of the grid, while the energy price is (seemingly)
uniform for the whole market. On the other hand, with zonal pricing, the
constraints of the grid affect the market price for energy and result in several
prices according to the relative locations of the zones and of the grid
constraints. The congestion costs are reflected in the energy prices, and some
consumers will therefore benefit from, and others suffer from, energy prices
that differ from the system price. For generators, the effects will be the
opposite of those of the consumers (consumers in an area obviously prefer
a low price, while generators prefer a high price). The relative winners and
losers will therefore depend on the definition of the price areas and on their
location on the grid (and in the end, also of how the total fixed charges are
determined and distributed).

3. A SIMPLE EXAMPLE WITH NODAL PRICING

In the following examples, we assume a linear and lossless approximation11

of the power flow equations (Wu et al. (1996)), and we focus on real power12,
i.e. the power that provides energy. The flows of the network are determined
by Kirchhoff’s laws, and as explained in Boucher and Smeers (2002), in
meshed networks, these laws direct electric flows in the grid and disconnect
the effective physical path from the contractual one, i.e. the direct least-
distance path between the generator and the consumer involved in a
transaction. One of the main consequences of these physical laws is that trade
between two parties affects flows all over the grid, and that the market

11
Most of the transmission grids are fed with alternating current. The current intensities and the
voltages on the grid are then sinusoidal. Through a linearization of the power flow equations,
the ”DC” approximation facilitates the evaluation of the power flows on the grid for the
transmission system operators.
12
In an alternating current network, we distinguish between the real power, or ”useful power” since
it will have mechanical, thermal or chemical effects, and the reactive power, useful to control
the voltage, but unable to supply energy.

276 Intersentia
 Congestion Management in the Nordic Power Market

participants can only modify the electric flows by modifying net injections and
withdrawals at different points in the grid. This is very different from networks
where flow is routable, i.e. it is possible to select a specific route through the
grid for a given trade.

The network that we study contains 5 nodes connected by 6 electrically

identical lines, like the grid in figure 1. In every node, there is both production
and consumption, and for simplicity, we assume quadratic cost and benefit
functions, implying linear supply and demand curves. Demand, qid, in node
i is given by pi = ai - bi qid, where pi is the price in node i and ai and bi are positive
constants. We therefore assume elasticity in demand and a minimum level of
consumption at each node. Supply, qis, is given by pi = ci qis, where ci is a positive
constant. Considering a situation without any market power, the prices offered
by the generators should equal their marginal generation cost, variable
according to the quantity produced. The parameters for the example are given
in table 1. Demand is assumed to be identical in all 5 nodes, whereas supply
is relatively cheap in node 1 and relatively expensive in nodes 2 and 4.

Table 1: Cost and Demand Parameters

Node Consumption Production

I ai bi ci
1 20 0.05 0.1
2 20 0.05 0.6
3 20 0.05 0.4
4 20 0.05 0.8
5 20 0.05 0.3

First, we find the least cost dispatch of the generation units, ignoring any grid
limitations, and this results from simply aggregating all supply and demand
curves, finding the market-clearing price and tracking back the equilibrium
quantities from the individual supply and demand curves. This procedure
maximizes the social surplus, which is the monetary assessment of the
consumers’ and producers’ well-being on the whole system, and equal to the
area under the demand curve (consumers’ willingness to pay) less the
production costs. For the example, the unconstrained least cost dispatch
entails the electric flows depicted in figure 1, and since the dispatch is
unconstrained, and we have assumed no losses, there is a unique price, the
system price, which is equal to 16.84. The social surplus amounts to 3157.90,
and the grid revenue is zero since there is no congested path on the grid.

Journal of Network Industries, Volume 4 (2003), No. 3 277

 Mette Bjørndal, Kurt Jørnsten, Virginie Pignon

Figure 1: Unconstrained Dispatch

Let us assume that the flows resulting from the unconstrained energy market
clearing entail two binding constraints: the first one on line (1,2), whose
capacity limit is 51 energy units, and the second on line (4,5) with a limit of
11 energy units. The constrained dispatch, maximizing social surplus while
taking into account the transfer capacity constraints (i.e. reducing flows on
lines (1,2) and (4,5) below their respective capacities), gives the optimal
dispatch depicted in figure 2, where nodal prices and flows are displayed.

Now, due to the congestion, prices differ from one node to another, and
depending on location, the energy prices paid to the generators and charged
from the consumers are higher or lower compared to the (unconstrained)
system price. The variation in prices is about 3.17%, i.e. rather small price
differences across the network. As for the social surplus, not surprisingly, it
is reduced to 3155.49. Indeed, because of Kirchhoff’s laws, the only way to
relieve the constraints is to change the least cost dispatch; hence the demand
will be met by higher cost generation units that, absent the constraint, would
not run. Due to the congestion, these more expensive plants are constrained
on in order to achieve equilibrium between energy consumed and generated.
Nevertheless, the social surplus is just slightly reduced, implying congestion
costs of 2.41 monetary units.

Figure 2: Constrained Dispatch: C12=51, C45=11

278 Intersentia
 Congestion Management in the Nordic Power Market

Let us consider now that rather than directly manage the constraints on lines
(1,2) and (4,5), we put a “fake” constraint on line (2,4). For instance, we put
a capacity limit of 10 units of energy on this line. Once again, the prices
computed are the optimal nodal prices, taking into account a “fake” capacity
limit on line (2,4). The resulting flows are displayed in figure 3.

Figure 3: Constrained Dispatch, C24=10

In this case, the variation in prices is much higher, entailing larger relative
costs and benefits for the generators and consumers located at each node.
Here, the variation in prices across the grid amounts to almost 10.2%. The
social surplus now is much lower and equals 3137.36, entailing congestion costs
of 20.54, where congestion costs are measured by the difference in social
surplus compared to the unconstrained dispatch.

Nevertheless, an interesting point is that the congestions on line (1,2) and

(4,5) are both relieved when we “move” the capacity limits to line (2,4).
Because of Kirchhoff’s laws, the effective use of transfer capacity on each line
depends a priori on the origin and destination of all simultaneous transactions
on the grid. Imposing a “fake” constraint can therefore imply a change in the
quantity generated and consumed that modifies the effective physical flows
of electricity on each line of the grid, in this case such that the real constraints
are fulfilled. Obviously, this “fake” constraint also changes the price levels at
each node so that some network users could take advantage of this. However,
if we take into account the costs and benefits of the congestion management
mechanisms currently used in the Nordic power market, the incentives from
putting up a “fake” capacity limit may become much higher.

Indeed, we assume that this simple network represents in fact two markets
linked with the cross-border lines (2,4) and (3,5). One of the markets, market
N, is made up of nodes 1, 2 and 3, whereas the other one, market S, is made
up of nodes 4 and 5. Consequently, the capacity limit of line (1,2) is now

Journal of Network Industries, Volume 4 (2003), No. 3 279

 Mette Bjørndal, Kurt Jørnsten, Virginie Pignon

internal to market N, and the one on line (4,5) belongs to market S. This
configuration can be seen as a stylized illustration of the Norwegian-Swedish
grid. If the congestion management method used to relieve internal congested
paths is different from the one used at the cross-border lines, then the
incentives to “fake” a constraint on line (2,4) may be greatly increased. The
costs for the system operator resulting from counter purchases would then
disappear and be replaced by grid revenues resulting from zonal pricing. In
the above example, the costs from counter purchasing would at least be 2.41
monetary units13, whereas the grid revenue when the capacity of line (2,4) is set
to 10 energy units is equal to 22.73 (summing price-differences times energy
flow for all lines in the network).

We have already noticed that the “fake” constraint entails a congestion cost,
measured by the reduction in social surplus, much higher than when we
manage congestion on line (4,5) and (1,2) by putting capacity limits directly
on these lines. However, the calculations made up to now are based on a nodal
pricing scheme for electricity. It would be interesting to see how the results
are affected when zonal rather than nodal prices are computed. Zonal prices
are currently computed for the Nordic power market, and their implementati-
on in order to allocate scarce transmission capacity in the rest of Europe is
seriously considered.

4. IMPLEMENTING ZONAL PRICING IN THE EXAMPLE

Theoretically, zonal pricing is an approximation of a full nodal pricing regime

and results from the aggregation of nodes into zones, thereby reducing the
number of different prices in the market. Stoft (1997) shows that the partition
of the network into zones is generally not obvious, however, he states that it
should be based on price differences. Yet, Bjørndal and Jørnsten (2001) show
that a zone allocation mechanism based on optimal nodal price differences
does not necessarily lead to a zone system with maximal social surplus. In
practical zonal implementations, the nodes at the endpoints of a congested
line would typically be allocated to different zones 14.

13
This assumes that the system operator can both increase and decrease prices indiscriminately
and that the original bid curves are used. It is the minimal costs from counter purchases, since
in reality the bids on the energy market and on the adjustment market may differ. Ignoring
market power, the prices offered on the adjustment market are generally higher because a quick
change in the quantity generated is costly, especially for the rather inflexible nuclear and coal
generation units.
14
Although this is not always optimal when we take into account more than one congested line
(Bjørndal and Jørnsten (2001)).

280 Intersentia
 Congestion Management in the Nordic Power Market

We consider again the simple example of figure 1 / table 1, and assume that
the boundary between the zones cut lines (2,4) and (3,5) vertically, like in
figure 4. Thus, the allocation of nodes into zones is fixed, which is more or
less also the situation in the Nordic power market where the national
boundaries form fixed zonal interfaces, and Norway may be split into a few
semi-fixed zones15. With only two zones, we have two sub-markets, the first one
consisting of market N (nodes 1, 2 and 3) and the second one of market S
(nodes 4 and 5).

Obviously, the unconstrained dispatch remains the same as with nodal pricing
(figure 1), since we ignore the grid constraints. In relation to the constraints
of lines (1,2) and (4,5), the chosen zonal division does not cut any of the
congested lines, and it is in fact not possible in this example to find two prices
that clear the sub-markets and result in a feasible flow. The problem must then
be resolved by counter purchasing. However, what we are interested in here
is what happens now if we “fake” a constraint on one or more of the cross-
border lines. It turns out that the results are completely different compared
to the case of nodal pricing. Even with the same supply, demand and network
parameters, the way the prices are computed changes everything. Thus, as we
did before, we put a capacity limit of 10 units on line (2,4) in order to decrease
the flows on the real congested lines (1,2) and (4,5). The results are shown
in figure 4.

Figure 4: Zonal Pricing with C24 = 10

The flow on line (1,2) is effectively reduced, but the flow on line (4,5)
increases, thereby creating a potential conflict of interest between the system

15
The zonal definitions within Norway are fixed beforehand, but can be redefined periodically if
there is a special need for it.

Journal of Network Industries, Volume 4 (2003), No. 3 281

 Mette Bjørndal, Kurt Jørnsten, Virginie Pignon

operators of the different zones, as regards to the capacity announcements

of the cross-border lines16. In reality, and highlighted by the European
Commission (2003), the smooth cooperation between the transmission system
operators is one of the central points for an effective allocation of scarce cross-
border transmission capacity. This simple example shows that the probability
of smooth cooperation between transmission system operators may be affected
by the congestion management methods adopted. If we put a capacity limit
(for instance of 5 energy units) on line (3,5) rather than (2,4), the same
potential conflict of interest appears with 49.760 units on line (1,2) and 14.540
units flowing on line (4,5). Thus, using a zonal approach to computing prices
may make things completely different, and with only two price-areas, a “fake”
constraint on a cross-border line is in our example not enough to manage both
the real internal constraints.

Therefore, let us consider that market N is in fact split into three areas so that
each node represents one zone. Market S is still considered as a single area.
We could expect that this further split in the market should give more similar
results as in the case with optimal nodal prices. However, even if only nodes
4 and 5 are aggregated now, the flows in the grid resulting from a “fake”
constraint are quite different from the case where nodal prices were compu-
ted. The flows of lines (1,2) and (4,5) are displayed in table 2. Once again,
“faking” a constraint on the cross-border lines succeeds in reducing the flow
on line (1,2) below the threshold of 51 units, but fails to manage the congesti-
on on the internal congested line in market S, although flow has been reduced
over line (4,5) when the limit is put on line (3,5).

Table 2: Flows on line (1,2) and (4,5) with “fake” constraints on (2,4) or (3,5)

C24 = 10 C35 = 5
line (1,2) 50.472 48.936
line (4,5) 15.517 13.738

Let us consider now that the zone-boundary is flexible, and can be moved so
that it cuts the congested lines. Accordingly, lines (1,2) and (4,5) are inter-
zonal. Taking into account the real capacity limits of these lines for the
dispatch of generation and loads, there are some interesting results, which
are displayed in figure 5. Indeed, the social surplus of this alternative two-zone
solution amounts to 3153.81, which is quite close to unconstrained dispatch,
and the congestion cost amounts to 4.08 monetary units. Besides, the capacity

16
The social surplus is now 3133.73, implying congestion cost of 24.17. However, the congestion
costs are not really comparable with those cited previously, since the constraints are not fully taken
care of in this solution. In general, aggregating nodes into a limited number of zones increases
the congestion costs.

282 Intersentia
 Congestion Management in the Nordic Power Market

limit of line (4,5) is not binding anymore. Thus, the management of the real
congestion on the inter-zonal line (1,2) alleviates the congestion on line (4,5),
and although there are still only two prices, the allocation of the southern parts
of markets N and S to the same area, gives results that are very close to the
optimal dispatch, taking into account the true capacity limits of the transmissi-
on lines.

Figure 5: Alternative Zones with C12=51 and C45=11

These simple numerical examples point to the fact that computing zonal prices
makes things completely different, even if we consider competitive markets
without gaming or related strategic behavior17. Zonal pricing is not a mere
simplification consisting of reducing the number of different prices; zonal
pricing also changes the allocation of social surplus among the market
participants through averaged prices in each zone. Besides, fixing the
boundaries between the price areas, considering mainly the national borders,
irrespective of the electrical reality of the grid, does not seem to be a consistent
way of implementing zonal pricing. As we have shown in the last example,
fixing the zones according to the location of the congested lines could give
much more effective results, and the efficient use of scarce transmission
capacity implies that the boundaries between the price areas should be fixed
according to the effective physical constraints, not the political ones.

Even if zonal pricing is from an institutional point of view an appealing

method for Europe, because of the liberty it allows in the design of the
transmission pricing rules within zones, it will incite the development of
efficient transactions only if the boundaries between areas correspond to the
effective constraints. One of the practical consequences is that zonal pricing
should also enable the allocation of scarce transmission capacity within

17
For a study of market power with zonal pricing, see Harvey and Hogan (2000).

Journal of Network Industries, Volume 4 (2003), No. 3 283

 Mette Bjørndal, Kurt Jørnsten, Virginie Pignon

countries, and not only between countries, and that the boundaries between
price areas should follow the physical constraints.

However, the implementation of variable boundaries between zones may turn

out to be quite complex. Thus, Hogan (1999) computes the number of zones
needed to implement zonal pricing in PJM18. The results are worth taking into
account for a practical implementation of zonal pricing in Europe. As in
continental Europe, the network in PJM is large and highly meshed. The
immediate consequence of a meshed network is that electric flows may use
a larger number of lines compared with a radial network, i.e. a network without
loops. This high number of lines, injection and withdrawal nodes, makes the
relief of a constraint easier since it increases the number of possible adjust-
ments. On the other hand, it makes the physical paths of electric power more
variable, introducing more uncertainty concerning the likely location of the
physical constraints. Hogan (1999) uses actual data on nodal prices in April,
May, June, July, August and September 1998. The criterion considered to
select a zone is that the standard deviation of prices across the zone should
be less than 10% of the average prices. It results in 94 zones in May, 75 in June,
57 in July, 52 in August and 64 zones in September, needed to cover all the
major physical constraints. The zones are not the same in each month and
sometimes, locations that should belong to one zone are not contiguous!

Returning to our stylized examples, implementing zonal pricing in a network

with only 5 nodes is not realistic. Besides, with a small network, matters are
not so simple since there is indeed a lack of flexibility to aggregate nodes into
zones with a uniform price. In the Nordic Power market today, there are
normally 6 price areas (2 in Norway, 1 in Sweden, 1 in Finland and 2 in
Denmark), and we may expect at least about 15 areas in the whole of Europe.
In the next section we will consider a single constraint in an extended network.
This will provide more flexibility in determining area-boundaries and area-
prices.

5. AN EXTENDED NETWORK

We consider a slightly larger network consisting of 8 nodes, 4 in market N and

4 in market S. The configuration of this example (see figure 6) is also inspired
by the existing boundary between the Norwegian and the Swedish grids. We
still assume that there is both production and consumption in every node, with
quadratic cost and benefit functions, i.e. linear supply and demand curves.
Compared to the smaller example, we now assume different parameters for

18
PJM is an area in the US where electric power flows are controlled by a single system operator.
This area covers the major part of Pennsylvania, New Jersey, Maryland, Delaware, the District
of Columbia and a small part of Virginia.

284 Intersentia
 Congestion Management in the Nordic Power Market

the cost and demand functions for each location. Implementing different bi‘s
can be interpreted as varying the size of the (nodal) markets19. The chosen
parameters, given in table 3, seem well suited to the relative differences in the
Norwegian and Swedish sub-markets.

Table 3: Cost and Demand Parameters

Node Consumption Production

I ai bi ci
1 20 0.02 0.8
2 20 0.05 0.1
3 20 0.1 0.6
4 20 0.25 0.4
5 20 0.02 0.8
6 20 0.05 0.5
7 20 0.02 0.3
8 20 0.25 0.2

The unconstrained dispatch of the generation units needed to meet the

demand, results in the flows of figure 6. In this case, the social surplus is equal
to 4779.57, and the system price is 17.70. This non-congested situation reflects
the market potential. We assume, however, that the unconstrained dispatch
of the generation plants violates the transfer capacity limit on line (7,8), which
amounts to 90 energy units. The management of the congestion will imply
costs, thereby reducing the social surplus, and again, the way this congestion
is managed affects the final results for the involved parties.

Figure 6: Unconstrained Dispatch

19
Low b-values represent larger markets in terms of the quantity consumed for a given price.

Journal of Network Industries, Volume 4 (2003), No. 3 285

 Mette Bjørndal, Kurt Jørnsten, Virginie Pignon

We consider first the constrained optimal dispatch, where generation units

and loads are re-dispatched so that the flow on line (7,8) falls below 90 units.
Nodal prices and flows are displayed in figure 7. Now, the social surplus equals
4759.24, and the minimum congestion cost resulting from adjustments in
generation and demand, is therefore 20.34. For the generators and consumers,
the relative benefits and losses compared to the unconstrained situation,
depend on their location on the grid, since the single congestion on line (7,8)
provokes different prices for electric power at each node. The constrained
optimal dispatch represents an upper boundary on the social surplus that can
be attained from any congestion management method, and this solution could
be obtained either by nodal pricing or by a “perfect” counter purchase
arrangement, where the least cost re-dispatch induces a cost of 20.3420 for the
system operator.

Figure 7: Constrained Dispatch C78 = 90

Zonal prices, in the same way as nodal prices, should reflect congestion cost
so that the energy flowing on line (7,8) does not exceed the capacity limit of
this line. With two zones, i.e. zone N, consisting of nodes 1,2,3,4, and zone S,
consisting of nodes 5,6,7,8, there are now three cross-border lines linking the
markets. The resulting prices and flows, displayed in figure 8, have the effect
that the congestion cost increases to 34.80, which is an increase of 71%
compared to the case with nodal prices. Thus, zonal pricing is a price signal

20
In such a setting, the nodal prices are interpreted as nodal marginal values, and apply only to the
marginal quantities that are injected or withdrawn from the nodes.

286 Intersentia
 Congestion Management in the Nordic Power Market

that alters generation and consumption in each node, in a way that may
noticeably increase the congestion costs for market participants 21.

What is also interesting in this example is that if we take into account an

incentive to “move” the limit from the intra-zonal line (7,8) to one of the cross-
border lines, it is possible to manage the internal constraint. Let us consider
a “fake” constraint of 24 energy units on line (3,7). The resulting prices and
flows for each line are displayed in brackets in figure 8. Thus, “faking” a
constraint on the cross-border line (3,7) relieves the internal congestion22, and
the zonal solution with the original constraint is replicated, giving the same
flows as the real constraint, and the same social surplus.

Figure 8: Zonal Pricing (two zones)

However, since the constraint (7,8) is internal to price area S, it should have
been resolved by counter purchases, and even if we assume competitive
regulation markets with adjustment bids equal to generation marginal costs,
this mechanism entails costs for the network operator23. Resolving the
constraint by zonal pricing on the other hand, provides a grid revenue of
18.36, putting the limit of 24 energy units on line (3,7). Therefore, there is

21
The social surplus of the optimal dispatch can be attained with zonal pricing only if the grid is
zonable with respect to the given constraints, i.e. if the internal transactions within the zones have
no consequences on the flows of the cross-border lines between zones. This can be true under
very special grid configurations, and we have shown that even with our rather small meshed
networks, zonal pricing entails social surplus lower than with the constrained optimal dispatch
(Crampes and Laffont (2001)).
22
In fact, this could also be achieved by putting a limit on the flow of line 1-5.
23
For a study of gaming with a counter-purchase arrangement, refer to Stoft (1998).

Journal of Network Industries, Volume 4 (2003), No. 3 287

 Mette Bjørndal, Kurt Jørnsten, Virginie Pignon

an incentive to “move” the internal constraint on line (7,8) to one of the cross-
border lines, so that the management mechanism would be zonal pricing with
market splitting, and not counter purchases. Hence, the real congestion is
replaced by a replica, which is managed through the change in supply and
demand resulting from the zonal prices in each area.

Another way to cancel the costs of counter purchases would be to split market
S into different zones, so that line (7,8) would become an inter-zonal
transmission line. Hence, we assume now that zonal prices are computed with
four zones, two in area N and two in area S, as depicted in figure 9. The
boundaries cut lines (4,8), (3,7) and (1,5) as before and, furthermore, line
(3,4) in zone N and (7,8) in zone S. The results from moving the capacity limit
to line (3,7) are displayed in brackets in figure 9. In that case we have again
assumed that the grid is operated as if there were a transfer capacity limit of
24 energy units on the flow of line (3,7).

Figure 9: Zonal Pricing (four zones)

When we consider four zones rather than two, the differences in flows and in
prices between the case where line (7,8) is limited, and the case with a capacity
limit on line (3,7), are substantial. A capacity limit on line (3,7) cannot
replicate the solution for the real constraint on line (7,8). The results
concerning the social surplus and the grid revenue are also different, as
displayed in table 3. By using the actual constraint on (7,8) there is a
considerable reduction in congestion cost when moving from 2 to 4 zones
(from 34.80 to 22.26 monetary units, i.e. very close to the optimal dispatch),

288 Intersentia
 Congestion Management in the Nordic Power Market

but hardly any improvement when the network is operated with the constraint
on line (3,7).

Table 4: Social Surplus and Grid Revenue with Four Zones

C78 = 90 C37 = 24
Total Social Surplus 4757.32 4745.33
Social Surplus Zone N 2601.45 2613.54
Social Surplus Zone S 2018.72 2092.52
Grid revenue Zone N 17.65 -10.72
Grid revenue Zone S 120.51 29.15
Grid revenue lines (1,5) (3,7) (4,8) 42.97 28.75

With four zones, the incentive to “move” the real constraint to line (3,7)
vanishes. Indeed, the congestion on line (7,8) is managed through the price
difference between the internal areas in zone S, thereby canceling the costs
of counter purchases. Moreover, employing the real constraint brings about
much higher grid revenues. Thus, putting the boundaries between the price
areas where the effective physical constraint lies substantially limits the
incentive to “cheat” on the congestion management rules, at least for the
transmission system operators. Therefore, we could expect that “faking” a
constraint would be much less interesting if the same mechanism was adopted
to manage internal as well as cross-border congestion.

Once again, these figures point to the great interactions in an electric network,
and the importance of operating a power system according to its real
constraints. “Moving” a constraint, even with the same pricing rules and zones
is not a neutral decision. This is very much illustrated by the differences that
the agents experience as regards to the prices for the different solutions. The
prices for the producers and consumers in node 8 for different solutions are
given in table 5.

Table 5: Different Prices for Node 8

Pricing rule and Constraint Price

Nodal pricing, C78 = 90 16.83
Nodal pricing, C37 = 24 17.84
2 zones 18.17
4 zones, C78 = 90 16.75
4 zones, C37 = 24 17.85

The different distributive consequences of the congestion pricing rules, as well

as Kirchhoff’s laws, that make the management of a true constraint with a
“fake” one possible, may raise incentive problems, as the great interactions
between the uses of an electric network allow “cheating” on the rules to benefit

Journal of Network Industries, Volume 4 (2003), No. 3 289

 Mette Bjørndal, Kurt Jørnsten, Virginie Pignon

from the distributive effects of a particular congestion pricing method. This

strategic behavior affects the prices for the all market participants, sends the
wrong pricing signals, and may result in inefficient transactions between the
consumers and generators. Eventually, it may put a threat on the economic
efficiency of the market, which is the main goal pursued by the creation of
integrated regional markets.

6. SUMMARY AND CONCLUSIONS

In this paper, we have illustrated the congestion management mechanisms

used in the Nordic countries, i.e. zonal pricing or market splitting on the one
hand for the inter-zonal constraints, and counter purchases or counter trading
for intra-zonal congested paths on the other hand. We have seen that the
simultaneous use of these methods may give incentives to “cheat” on the rules,
so that the congestion would be managed with one method rather than the
other. The physical rules that direct electric flows make the replacement of
a real intra-zonal congestion by a “fake” constraint on an inter-zonal line
possible. The incentive may be quite clear for a revenue-regulated network
operator that does not have to pay for the costs of counter purchases. The
incentive exists also, even if it is less clear-cut, for some generators and
consumers that would benefit from more profitable prices. They could face
a decreased transmission tariff resulting from the lack of counter purchases’
costs, and also (perhaps above all) a change in the zonal prices resulting from
the “fake” constraint. The solution hence might be in proper regulation, that
the network operator should at least enforce the same pricing rules for intra-
zonal and inter-zonal transmission lines. The incentives to “fake” constraints
are all the more important as the boundaries between the zones are fixed,
since it is easier then to implement such strategic behavior. We have indeed
shown in the examples that the incentives may be greatly reduced when the
boundaries are where the congestion really lies.

We have also shown that zonal pricing may produce completely different
results, as regards the prices of course, but also as regards the flows on the grid,
the congestion, the social surplus and the grid revenue. Hence, zonal pricing
is not a mere simplification of nodal pricing; the aggregation of nodes into
zones with uniform energy prices does substantially change the allocation of
social surplus among the agents, thereby bringing about winners and losers
in the market, with different and conflicting incentives. Therefore, the
consequences of choosing zonal rather than nodal prices are extremely
difficult to anticipate, and they should be carefully assessed before implemen-
ting zonal pricing in Europe. Moreover, the differences in the optimal dispatch
and different zonal solutions as regards congestion cost for “real” and “fake”
constraints, indicate that there might be a reduction in social surplus from

290 Intersentia
 Congestion Management in the Nordic Power Market

managing a constraint through a replica. However, in order to assess the exact

cost for society, we need to take into account in detail how counter purchases
are carried out, and how this secondary market functions. Modeling this
market interaction is a topic for future research. Finally, as the prices vary
considerably according to which constraint is considered in the solution, the
prices that result from managing congestion by a “fake” constraint may be
misleading if used as a signal of the usage of scarce transmission resources.
The incentive to “fake” constraints may therefore substantially reduce the
economic efficiency of the market, which is the main goal of the integration
of regional markets for electric power.

The subject investigated in this paper is related to the specific methods for
relieving transmission constraints within the Nordic power market. However,
the analysis and problems posed are of major concern when considering the
integration of regional power markets with a meshed grid structure. Therefore,
it should be of great interest regarding the creation of a European electricity
market and the harmonization of various sub-markets within this. Even if the
pricing rules used in the Nordic power market seem appealing from an
institutional point of view, the potential efficiency problems raised should be
carefully taken into account. Of course, the interactions in the network are
all the more important as the network is highly meshed. The incentive
problem may therefore be particularly acute in Europe, and more severe than
in the Nordic market, where in reality only Norway and Sweden are strongly
electrically connected. This is a point that we should bear in mind when we
design transmission pricing rules for the European electricity market.

REFERENCES

BJØRNDAL M. (2000). Topics on Electricity Transmission Pricing. Dissertation

submitted for the Degree of Dr. Oecon., Norwegian School of Economics
and Business Administration, Bergen, Norway.
BJØRNDAL M. and JØRNSTEN K. (2001). “Zonal Pricing in a Deregulated Energy
Market”. The Energy Journal, 22, pp. 51-73.
BOUCHER J. and SMEERS Y. (2002). “Towards a common European Electricity
Market – Paths in the right direction…still far from an effective design”.
Journal of Network Industries, 3, pp. 375-424.
CRAMPES C. and LAFFONT J-J. (2001). “Transport pricing in the electricity
industry”. Oxford Review of Economic Policy, 17, pp. 313-328.
DE VRIES L. and HAKVOORT R. (2002). “An Economic Assessment of Congesti-
on Management Methods”. The Journal of Network Industries, 3, pp. 425-466.
ETSO (1999). “Evaluation of congestion management methods for cross-
border transmission”. www.etso-net.org.

Journal of Network Industries, Volume 4 (2003), No. 3 291

 Mette Bjørndal, Kurt Jørnsten, Virginie Pignon

ETSO (2002). “Co-ordinated use of Power Exchanges for congestion manage-

ment in Continental Europe: Market Design and Role of Power Exchang-
es”. www.etso-net.org.
European Commission (1996). Directive 96/92 concerning common rules for
the internal market in electricity.
European Commission (2003). Regulation No1228/2003 of the European
Parliament and of the Council on conditions for access to the Network for
cross-border exchanges in electricity.
HARVEY S. and HOGAN W. (2000). “Nodal and Zonal Congestion Management
and the Exercise of Market Power”.
http://ksgwww.harvard.edu/people/whogan.
HOGAN W. (1992). “Contract Networks for Electric Power Transmission,”
Journal of Regulatory Economics, 4, pp. 211-242.
HOGAN W. (1999). “Restructuring the Electricity Market: Institutions for
Network Systems”. http://ksgwww.harvard.edu/people/whogan.
KNOPS H., DE VRIES L. and HAKVOORT R. (2001). “Congestion management
in the European electricity system: an evaluation of the alternatives”. The
Journal of Network Industries, 2, pp. 311-351.
SCHWEPPE F., CARAMANIS M., TABORS R., and BOHN R. (1988). Spot Pricing of
Electricity, Kluwer Academic Publishers.
STOFT S. (1997). “Zones: Simple or Complex?”. The Electricity Journal, January-
February, pp. 24-31.
STOFT S. (1998). “Using Game Theory to Study Market Power in Simple
Networks”. Federal Energy Regulatory Commission.
WANGENSTEEN I. (1997). Deregulation of the Power Market – Implementation and
Experiences in the Norwegian and Nordic Market. Report EFI Sintef Group TR
F4562.
WESTRE E. (1995). “Transmission pricing in Norway”. Electricity Transmission
Pricing and Technology, Einhorn M., Siddiqi R. (eds.), Kluwer Academic
Publishers.
WU F., VARAIYA P., SPILLER P., and OREN S. (1996). “Folk Theorems on
Transmission Access: Proofs and Counterexamples”. Journal of Regulatory
Economics, 10, pp. 5-23.

292 Intersentia