Table of Contents
Go Back
Tradable green certificates: The dynamics of coupled
electricity markets
Klaus Vogstad, Ingrid Slungard Kristensen and Ove Wolfgang!
Norwegian University of Science and Technology (NTNU)
1. Sintef Energy Research
SEFAS Sem Seelandsvei 11/N-7465 Trondheim
phone +47 73597644 fax +47 73597250
klausv@stud.ntnu.no
1 Abstract
Liberalisation of markets previously under regulatory control require new instruments for
environmental policy making, because subsidies and regulatory intervention does not
conform to trans-national, liberalised markets. This is the case for newly regulated elec-
tricity markets. An arrangement of Tradable Green Certificates (TGC) as a market-based
subsidy for renewable energy has been proposed in several countries and already imple-
mented in a few. However, introduction of TGCs have been postponed and delayed
mainly due to the uncertainties involved for suppliers of renewables. Several studies have
been undertaken using economic static comparative analysis and partial equilibrium mod-
els. However, few of these analyses address the dynamic price formation process or the
mechanisms that are important in the design of a well-working stable market. To analyse
the stability of a TGC market, we construct a system dynamic model of the TGC market
coupled with the Nordic electricity market (Nord Pool). A set of trading strategies for the
participants under various marked designs is examined. These trading strategies were de-
duced from laboratory experiments.
The results showed that the proposed TGC market designs are likely to become unstable.
These instabilities arose endogenously from the trading strategies. Some crucial design
parameters such as banking and borrowing can reduce these instabilities. In particular,
the proposed banking arrangement intended to avoid price fluctuations caused by the
yearly stochastic variation of renewables. As a side effect, banking opts for some trading
strategies that cause even more harmful price fluctuations followed by price crashes.
These undesirable instabilities can be reduced by allowing borrowing and limit banking.
The conclusions from previous theoretical studies on the TGC market is examined and
compared with our findings. This case study shows how system dynamics can be com-
bined with experimental economics to address issues that cannot be dealt with within the
framework of partial equilibrium models and standard economic theory.
2 Introduction
Liberalisation of European markets requires new tools and instruments for environmental
policy making. Utilities and public services (i.e. electricity, waste management, telecom-
munications) previously under regulatory control are now subject to deregulation. Under
these conditions, traditional national environmental policy instruments do not necessarily
work as intended (Morthorst, 2000). As an example, the main goal of developing wind
power in Denmark was to reduce national CO2-emissions, but under the operations of the
liberalised Nordic electricity market, these CO2-reductions take place in Finland and Ger-
many rather than Denmark (Vogstad et al, 2000). National subsidy schemes also distort
competition in transnational markets. This points to the need of environmental policy in-
struments that are compatible with open markets.
Tradable green certificates (TGC) have been proposed in Denmark, Sweden and within
EU to achieve their goal of adding 340 TWh renewables within 2010. However, introduc-
tion of TGCs to replace the renewable subsidy scheme has been delayed several times in
Denmark. A mandatory TGC market will start from May Ist, 2003, in Sweden. A TGC
market has been in operation in the Netherlands since 2001, but differs from the proposed
arrangements of the Nordic countries and EU by not being mandatory on the demand side.
Langniss and Wiser (2003) discusses the experiences of renewable energy portfolio stand-
ards! (RPS), where the Texas RPS has given promising results in developing renewables.
In the case of Texas, however, wind power received favourable subsidies in addition to
the TGC’s so that the system has not really been put to test yet.
Favourable feed-in schemes for wind power in Denmark, Germany and Spain have indeed
been successful in developing the industry from being an alternative energy source to be-
coming a competitive energy technology. One disadvantage about this feed-in scheme is
the large amount of costs it inflicts on the authorities as the renewable generation grows.
As a result direct obligations on the consumers were proposed, coordinated by a TGC
market.
3 The principle of tradable green certificates
The main purpose of tradable green certificates is to increase the share of renewable gen-
eration at minimum costs.
TGC’s are financial assets issued to producers of certified green electricity and can be re-
garded as a market-based environmental subsidy. An issuing Body (IB) issues green cer-
tificates at the moment a producer registers the production of actual green electricity.
They are later withdrawn from circulation at when customers account for their obligations
by presenting the certificates to the registration authority, or if the certificates period of
validity expire. Between issuing and withdrawing, the certificates are accounted and can
be traded. The certificates function as an accounting system to measure the amount of
electricity produced from renewable energy sources.
Figure I shows in principle how a TGC market will work within the Scandinavian elec-
tricity market (Nord Pool). In the Nord Pool market, electricity and its derivatives are trad-
ed in double-auction markets. The spot market is used for hourly production scheduling.
The Balance market coordinates short-term regulation”. Futures contracts are used for
electricity trading up to 3 years ahead, and are hence used for /ong-term planning and in-
vestment planning. A TGC market values the environmental benefit of renewables as a
service. The authorities define a mandatory share of demand for renewable generation and
the TGC market then finds the price needed to reach this target.
It should be noted that all these markets work independently of the physical transmission
1. Renewable Portfolio Standards is another term for Tradable green certificates
2. The balance market provide capacity for available for regulation within a period of 15 minutes.
Figure 1 The principle of a TGC market. TGC’; are financial assets that can be traded
independent of electricity generation. The value of a certificate reflects the cost of
providing the additional amount of new renewables needed to fulfil the obligation.
‘TGC obligation (in % of sales)
l
‘Wholesale 7
Distributor
Consumer
‘Supplier
i.
Metering
Metering
Physical
transmission
and that all necessary metering for accounting is made at the supplier and consumer.
4 Implications of TGC markets
One of the main lessons from standard economic theory is that instrument must be direct-
ed directly towards its purpose in order to be efficient. Hence, a TGC system will typically
be a cost-efficient way to increase the share of renewable electricity in consumption be-
cause the market will find the price needed to reach the predefined target, and the cost of
renewables are directly paid by the consumers. Similarly a TGC system is not an efficient
instrument of reducing CO2-emission. Efficient CO2-reduction can be obtained by a
CO2-quota market. However, both reducing CO2-emissions and increasing the share of
renewables are present environmental policies, which justifies the use of both in environ-
mental policy making. Jensen & Skytte (2003) provides an analysis on the simultaneous
attainment of CO2 and renewables targets.
Subsidy schemes require authorities both to set renewable targets, and to find the suffi-
cient level of subsidies that will ensure targets to be met. In a TGC market system, the
authorities can focus on the renewable target, leaving the price setting to the market.
A large market is preferred to obtain the real benefits of a TGC market. Firstly, resources
are unevenly distributed across countries. The EU-project Renewable Burden sharing
(ReBUS,) identified a total 15% cost reduction potential of achieving the EU targets of
additional 340 TWh new renewables within 2010. Some of the countries could however
reduce their costs by 40%.
Secondly, a larger number of participants will reduce the possibilities for market power.
However, opinions differ among EU countries with respect to which type of technologies
that can be defined in the TGC portfolio. Large-scale hydropower and waste incineration
remains an issue as to whether these sources should be included in the TGC market. Al-
lowing large-scale hydropower, is in conflict with the intention of TGC’s. Hydropower,
undoubtedly renewable, is a competitive source of generation, and most of its potential
has already been utilised. The technology is mature, and projects that do not conflict with
environmental interests are limited. Allowing hydropower in a TGC system would do lit-
tle more than generate additional income to hydropower utilities until TGC prices drop to
zero.
‘Waste incineration, can in some cases be considered as renewable, in some cases not - for
instance when plastic is incinerated. In many cases, waste incineration is profitable due to
high deposition costs. Such controversies must be sorted out to take advantage of the pos-
sible benefits of a TGC market.
A real disadvantage of TGC markets is that the less competitive sources such as PV and
wave energy will not be able to compete against cheaper alternatives of wind energy and
bio. Such technologies will still be in need of subsidies.
5 A system dynamics analysis of the TGC market
The implications of a TGC market have been the subject of studies in several reports,
mainly in the form of comparative static analysis or using partial equilibrium models e.g.
Figure 2 Model hierarchy. In this approach, we model develop submodels of the
electricity spot market, the TGC market and TGC Trading strategies before the models
are fully interconnected.
é
.
Electricity spot market (chapter 6) \
the Swedish white paper on TGC’s (SOU 77:2001), the EU-project REBUS and a series
of studies at Riso and under the Nordic research project Nordleden! (Risg, 2002). To our
knowledge, TGC markets have been simulated in the Markal energy model, plus Econs
power market model and the Balmorel energy model (Hindsberger, 2003). These models
are all partial equilibrium models and can be used to simulate the development of different
sources of renewables the TGC price, and their substitutes. They do not however, address
1. For reports on from the Nordleden project on TGC’s, see http://www.nordleden.nu
the consequence of time lags involved in construction of new capacity, or possibilities for
strategic behaviour of purchase and sales of TGC’s. To which extent these characteristics
are important for the price and for the design of a TGC market is the subject of this study.
Figure 2 shows our stepwise approach of constructing a system dynamics model of the
electricity spot market including a TGC arrangement. The purpose is twofold. First, to
study the price formation in the TGC market under various designs. Second, to analyse its
interactions with the electricity market.
In chapter 6 we develop a simplified model of the Nordic electricity market where renew-
able generation is traded in the spot market, with the present feed-in scheme for subsidis-
ing renewables.
In chapter 7, we develop the TGC market that replaces the feed-in scheme from 2003 on.
There are uncertainties and different opinions concerning stability of the proposed TGC
market designs (Schaeffer & Sonnemans, 2003; STEM 2002; Krohn 2001).
In chapter 10, we elaborate the TGC market model further by representing trading strate-
gies of buyers and sellers, enabling us to address the issue of price stability under various
market designs. Using the system dynamics approach, we explore some common trading
strategies to study the impact on price dynamics under the proposed TGC designs. The
trading strategies were deduced from a laboratory experiment with a group of players.
Based on these simulations and experiments, we identify some crucial design parameters
for a well-working TGC market. Finally, we connect the TGC market model with the
Electricity spot market model to study the interaction of those.
6 The spot market for electricity
We start our analysis by establishing a stock & flow model of the supply and demand side
of the Nordic Power Exchange (Nord Pool). The time horizon was set to 20 years, as re-
newable targets are part of long-term energy and environmental planning. Capacity utili-
sation is in turn determined by the spot prices, and a numerical time resolution of 3 days
is sufficient to adjust spot prices according to the changes in demand and supply!. We
start with a description of the power market with a feed-in tariff subsidy scheme for re-
newables to be used as a reference for simulations with a TGCs market. Figure 3 shows
the causal loop diagram (CLD) of the spot market. For simplicity, we only distinguish be-
tween thermal generation, hydropower and renewables. Hydropower is indeed renewable,
but for the purpose of certificates, existing large-scale hydro is not included for reasons
mentioned in chapter 4. On the supply side, two balancing loops are involved in the price
formation. The loop B/ adjusts generation of electricity by the short-run marginal cost
curve of existing capacity (see Figure 6 for details on the supply curve of thermal gener-
ation). If spot prices sustain at a higher level than the long-run marginal costs of genera-
tion (LRMC), new capacity will be added. The capacity acquisition loop B2 and B3 is
similar for both thermal and renewable generation, except that subsidies and tax policies
affect their profitability. We assume hydropower capacity to be fixed in this simplified
1. To represent market spot prices as a goal-seeking process require small time constants in com-
parison to the other time delays in the system, i.e. time delays for capacity acquisition and life-
time of installed capacity. An alternative way is to find market equilibrium prices using a search
algorithm within each time step.
Figure 3 Spot market dynamics CLD.
- _~LRMC thermal
Expected LRMC
profitability fossil ‘ renewables
Fossil capacity. aay aecuite
SRMC thermal B2- Capacity ARES Dene
\ acquisition fossil renewables
Price elasticity of Capacity factor ” Subsidies
demand ft sien ssi hema pay
Spot price commitment / acquisition Renewable
+A > B0- Demand c sal renewables capacity
Darand betas +f YX _ Total generation pacity
renewabl
Grid losses generation
hydro generation
A
Avg full load brs Hydro capacity Avg full load hrs
hydro renewables
model.
6.1 Market dynamics
The Nord Pool electricity market is a double-auction market that clears every hour. Ap-
proximately 30% of all electricity is traded through the spot market, the remaining share
is traded through bilateral contracts or long-term contracts. The time constant for the spot
market is set to 1 week - enough to give a good estimate of how the capacity factor (ca-
pacity utilisation) changes over a year. The market dynamics formulation is given in equa-
tion set (1) where spot price is a level that adjusts in proportion to the fractional demand/
supply balance.
(1) Market dynamics
1.1 Spot price, = Spot priceg + { chg in price; dt [NOK/MWh]
1.2. chg in price, = Spot price:(demand-generation tot)/demand:1/Market adjustment
time [NOK/MWh/da]
1.3 Market adjustment time = 7 [da]
1.4 Spot pricey = 200 [NOK/MWh]
The Nord Pool futures market represents the joint expectations of market participants,
where contracts for electricity can be traded up to 3 years ahead. This market is used as
an indicator when investment decisions for new capacity are being made (see section 3.4).
The expected future spot prices are modelled as an adaptive trend extrapolation (1.5) of
prior average spot market prices (1.6), where the smoothing time horizon is 3 years, and
Figure 4 Spot market dynamics stock & flow diagram
aa guio rum
ew capacty ee
Reference pce depreciation rate
‘v9 futiogd Wes hy
Figure 5 Nord Pool spot market prices and Forward contract prices (2 years ahead)
from 01/98 to 12/2000. Source: Nord Pool
Nord Pool market prices
Electricity prices [SEK/MWh]
0 40 0 90 ao a2 38 0
Week 1 to 156 during 01/1998 - 12/2000
the forward time horizon is 2 years.
1.5 Futures market price = FORECAST(Yearly avg price, 3,2) [NOK/MWh]
1.6 Yearly avg price = SLIDINGAVERAGE(Spot price, 1) [NOK/MWh]
6.2 Demand side
The demand side is kept simple in our model, as the main focus is on the supply side. De-
mand is modelled using a Cobb-Douglas function in equation set (2) with a price elasticity
of demand equal to -0.3 on a yearly basis, although the reported estimates vary from -0.2
to -0.8 (NOU 1998, 99; Econ 1999, 11; Groenheit & Larsen 2001, 46). Simulating over
20 years, demand and price elasticity’s will change significantly, but in this model we will
only address consumer prices influence the demand, and keep the reference demand con-
stant over the simulation period. All reference values refer to data from the year 2000.
(2) Demand side
2.1 demand = Demand ref (Spot price/Reference price)P1e elasticity of demand rryyp/yy]
2.2. Demand ref! = 420 [TWhAr]
2.3 Reference price = 200 [NOK/MWh]
2.4 Price elasticity of demand = -0.3 [1]
6.3 Unit commitment
Electricity is not a commodity, and cannot be traded as such. Electricity is a service and
share many common features of service sectors. In service sectors, such as the airline in-
dustry, services must be produced in a timely manner. In the same way as airlines cannot
store flights, electricity as a service cannot be stored.
For this reason, the generation capacity of electricity must be flexible to meet consump-
tion at all times. The units are scheduled after increasing marginal operational costs, as
can be seen from Figure 6. Normalising the below graph with total installed capacity
yields the capacity factor, CF varying between 0 and 1, which is the maximum capacity
utilisation.
The stock & flow equations for the unit commitment are presented in equation set (3)
(3) Unit commitment
3.1 generation th = CF-Max full load hrs [TWhAr]
3.2 CF=GRAPH(Spot price, 0,50, {0,0.014,0.11,0.58,0.82,0.914,0.94,0.98,1,1//
Min:0;Max: 1//}) [1]
3.3 Max full load hrs = 8000 [hr/yr]
The marginal operational costs of hydropower are neglible, and hydropower units with
reservoirs use some different strategies” in production planning. For simplicity, hydro-
power is represented with constant capacity utilisation. Renewable generation however,
1. Reference demand differs from observed demand in 2000. We deliberately chose a demand that
assured the model initially be in long-term equilibrium, to ease our analysis.
2. Hydropower units are usually scheduled using the water value method, which represents the
“marginal costs’ of hydropower. The water value is the marginal change of expected future
cumulative profits from releasing water from reservoirs for generation. This problem can be
solved as a stochastic dynamic optimsation. A simplified system dynamic representation of the
water value method is implemented in the Kraftsim model (Vogstad et al. 2002)
has an element of stochasticity.
3.4 generation hydro = Hydro-Avg full load hrs hydro [TWhAr]
3.5 Avg full load hrs hydro = 4800 [hr/yr]
3.6 Capacity hydro = 41600 [MW]
3.7. generation re = Renewable capacity-Avg full load hrs renewables: (Stochastic gen-
eration share:Stochastic variation+(1-Stochastic generation share)) (1) [TWh/yr]
3.8 Avg full load hrs renewables = 3350 [hr/yr]
3.9 Stochastic generation share = 0.6 [1]
C.1 Stochastic variation of wind was generated from wind energy series collected for
Norway. See Tande & Vogstad (2000) for further details
Figure 6 Capacity factor based on marginal production costs of thermal units
_x2ot_Marginal costs of generation, thermal units
“1 packlup coal
-- Trak ead castes m1?
Cumulative thermal capacity MW]
So asa 8000 38 at a0
Marginal costs of generation [NOK/MWh]
Here average full load hours represent the average of hydropower units, and for renewa-
bles they represent the weighted average of present bio energy and wind power. Total gen-
eration is the sum of generation from each technology minus grid losses:
3.10 generation tot = (generation th + generation hydro + generation re):(1-Grid losses)
[TWhhyr]
3.11 Grid losses = 0.1 [1]
An important difference between renewables and thermal generation is the inability to
control generation according to prices. Some bio/waste incineration units or small-scale
hydropower (defined as new renewable) with reservoirs do operate after marginal costs,
but it is a good approximation to regard short-term renewable generation as inelastic. In
other words, renewable technologies lack the Unit commitment loop B1. The level of re-
newable generation is therefore determined by the long-term capacity acquisition loop
B3, in combination with the stochastic properties of wind and water, which is not included
in the simplified model. As we will see later in 7, this has important implications for the
TGC market.
6.4 Profitability assessment and capacity acquisition
In the short term, electricity generation is adjusted by the processes described by the Unit
commitment feedback loop (chapter 6.3), in response to short-term demand variations. In
the long term, expectations of future prices govern the investment of new capacity. If the
expectations of future spot market prices are significantly higher than the long-run mar-
ginal costs (LRMC) of new generation capacity, the utility sector will invest in new ca-
pacity. Holding the futures market price (1.5) up against LRMC for thermal generation
(4.2) in equation (4.1) indicates the effect of profitability on investment rate shown in Fig-
Figure 7 Effect of profitability on investment rate
effect of profitability on investment rate
effect of profitability on investment rate [1]
0 05 1 15
Futures market price/LRMC [1]
ure 7. When futures market price equals LRMC, the effect of profitability on investment
rate returns |, at which the investment rate is in dynamic equilibrium with the depreciation
rate (see section 3.4). When the futures market price significantly exceeds LRMC, the in-
vestment rate increases, up to a certain limit that corresponds to a maximum 45% growth
rate. Growth within the power industry is limited by the availability of service and mate-
rial from other industrial sectors. The shape of the curve in Figure 7 can be recognised as
a cumulative probability density function that represents the aggregate of a large number
of possible profitable projects, which would differ in costs. The long-run marginal costs
can be represented by a more disaggregated net present value calculation including prof-
itability requirements, capacity factor, investment costs and operational costs, which is
implemented in the Kraftsim model (see Botterud et al. 2002; Vogstad et al. 2002).
(4) Profitability assessment
4.1 effect of profitability on investment rate th= GRAPH(Futures market price/LRMC
thermal ,0,0.25, {0,0.03,0.06,0.3,1,2.6,4.3,6.2,7.86,8.7,9//Min:0,;Max:10//}) [1]
4.2. LRMC thermal = 200 [NOK/MWh]
The same structure of profitability assessment applies to renewables except that subsidies
are included:
4.3 effect of profitability on investment rate re = GRAPH((Futures market price + Sup-
port scheme)/LRMC renewables ,0,0.25,{0,0.03,0.06,0.3,1,2.6,4.3,6.2,7.86,8.7,9})
[1]
4.4 Support scheme = 188 [NOK/MWh]
4.5. LRMC renewables = 300 [NOK/MWh]
It should be pointed out that the Nord Pool power market is not in long-term equilibrium.
A long-term (economic) equilibrium exists when the spot price equals the long-run mar-
ginal costs of new generation. For our case, the market is in long-term equilibrium when
the futures market price equals the long run marginal costs of the generation technologies,
that is
Futures market price = LRMC thermal = LRMC renewables+subsidies (0)
If this is not the case, the installed capacity of thermal and/or renewables and thereby
long-term prices will change. To simplify our study, we assume the electricity market to
be initially in equilibrium at a spot price of 200 NOK/MWh (which is the current observed
futures price in the Nord Pool market) and by letting LRMC thermal equal 200 NOK/
MWh while LRMC renewables is set to 300 NOK/MWh, requiring 100 NOK/MWh in
subsidies to maintain present installed capacity.
The futures market price is now approaching long run marginal costs of new generation,
while recent years have shown average market prices of around 157 NOK/MWh, which
is far below LRMC for new generation. Noteworthy, the futures price history from 1998-
2000 (Figure 5) showed a declining trend. Expectations of lower prices during the first
years of deregulation can be attributed to the expectations of increased competition, effi-
ciency that more than compensates for reduction of overcapacity. Most partial equilibri-
um models with endogenous investments assume markets to be in long-term equilibrium,
although this is rarely the case in real world.
Much of the considerable variation, which distorts the price signals, is subject to the large
variations of hydro inflow from year to year. Hydropower generation can vary as much as
+/-40% in a system where hydropower accounts for 61% of electricity generation during
normal years of hydro inflow. This problem is not encountered here in our simplified
model, as we use normal years of hydrological conditions, omitting the stochasticity of
hydropower. The lifetime of thermal units is set to 30 years and renewables to 20 years.
Thus the equilibrium fractional investment rate sufficient to match this rate is 3.33 %/year
and 5 %/yr respectively. Initial thermal capacity of 37360 MW and renewable capacity of
5685! correspond to the year 2000 situation for the Nord Pool market.
(5) Capacity acquisition
5.1 Capacity thermal, = Capacity thermal + { (new capacity th, - depreciation rate th,)
‘dt [MW]
5.2 new capacity th, = Equilibrium fractional investment rate th-effect of profitability
on investment rate th*Capacity thermal [MW/yr]
5.3 depreciation rate th, = Capacity thermal/Lifetime th [MW/yr]
5.4 Equilibrium fractional investment rate th= 3.33 [%/yr]
5.5 Lifetime th = 30 [yr]
5.6 Capacity thermalg = 37360 [MW]
5.7 Capacity renewables, = Capacity renewables + | (new capacity re; - depreciation
rate re,) ‘dt [MW]
5.8 new capacity re, = Equilibrium fractional investment rate re-effect of profitability
on investment rate re:Capacity renewables [MW/yr]
5.9 depreciation rate re, = Capacity re/Lifetime re [MW/yr]
5.10 Equilibrium fractional investment rate re = 5 [%/yr]
5.11 Lifetime re = 20 br]
5.12 Capacity renewablesy = 5685 [MW]
6.5 Simulation run with subsidy scheme
There has been a significant growth in renewables throughout the last years. On average,
the growth in renewables has been around 10% from 1999 to 2003 for the Nord Pool area.
We set the subsidy level to 188 NOK/MWh, which is the level of subsidies needed that
can maintain this growth rate. Two simulation runs are presented in Figure 8. Thin lines
show a simulation run with constant demand. The bold lines show the response of 10%
step increase in demand from 2003. In Figure 8a), the effect of subsidies is shown on con-
sumer price’, which coincide with spot price in this case. Demand grows slightly in re-
sponse to price reductions. Renewable generation increase its share of capacity and to a
lesser extent generation while thermal generation must reduce both installed capacity and
1. Renewable capacity is calculated as the sum of wind power and biomass installed in 2001, with
a corresponding average full load hour utilisation.
2. The increased taxes on customers from governmental spending on feed-in tariffs is not included
here.
capacity utilisation (Figure 8b)-d)). | The second simulation run (bold lines) underlines
Figure 8 Two electricity spot market simulation runs with 188 NOK/MWh subsidies
for renewables. Run I (thin curves): constant demand. Run 2 (bold curves): 10% step
increase of demand in 2003.
a)In both runs, Spot price and consumer price coincide b)In both runs, demand increases (coincide with total
(no taxes). Price decreases from increased supply of generation) from price reductions. Run 2: Thermal gener-
renewables. Run 2: Price responds rapidly from demand ation responds rapidly by increasing capacity factor
increase.
200: as [+ Demand
we [sense ee = generation ot
bs se price 200+ 2 =| "generation tot|
95: |4 *generation re
2
és
—.
c)Capacity development - renewables grow by 10% per d)In the long term, capacity utilisation is reduced from
year at the expense of thermal capacity in the long run increased share of renewables. Run 2: In the short term,
Run 2: Thermal capacity is more sensitive to electricity thermal generation can respond quickly to price changes,
prices than renewables with subsidies. but capacity adjustments of thermal generation will in the
long term deteriorate from increased renewable genera-
tion.
FFCapacity thermal
| *capacity thermal
|> Capacity renewables
|2 *Capacity renewables
Ficapacity factor thermal
Capacity factor thermal
a major difference between renewables and thermal generation. Thermal generation has
larger operational costs (i.e. fuel costs), and generation is scheduled according to their in-
creasing marginal operational costs. Capacity utilisation is therefore governed by the spot
price shown as the B/ unit commitment loop in Figure 3. In contrast, renewable generation
depends on the intermittent source of wind and water, whereas biomass in most cases gen-
erates electricity as a by-product of heat and is therefore not sensitive to electricity prices.
Renewable generation can only be adjusted by capacity acquisition (loop B3 in Figure 3)
which involves significant time delays. With respect to demand/supply balance and price
stability in the electricity market, stochastic generation from renewables does not yet rep-
resent insurmountable problems for the operation of the electricity market!, But how
would prices in a TGC market form, knowing where supply/demand balance must be met
on an annual basis, but where adjustments in supply involve long time delays in capacity
acquisition? This will be our concern in the following chapter.
1. In the Jutland area of the Nord Pool market, stochastic wind power now comprise more than
40% of total generation, which can still be absorbed by the power system, but not without coun-
termeasures.
7 The tradable green certificate market
A TGC certificate is a financial asset that can be traded independently of the physical gen-
eration of electricity. The physical supply of electricity is traded in the electricity spot
market so that a renewable supplier receives the spot price plus the TGC price per MWh
generation.
Figure9 CLD of TGC market mechanism
Price elasticity of Futures market
demand ree price
% TGC target +4 nee a 4
Demand 2 ;
\s ‘A % Baie Sab
Demand TGC : renewables A
B4- TGC demand B6- Capacity Renewable
balance ‘acquisition from TGC capacity
price P
TGC price
" pe Bele
7 wa generation
las - 76C supply TGC sales
balance +
+
+/ CF renewables
B7-Volume =
adjustment — TGC volume
stored
stochastic variation
Discrepancy
Desired volume
coverage
The main idea is to stimulate capacity acquisition of renewables through the TGC price
(loop BO, Figure 9) in order to fulfil the renewable share targets. This mechanism replaces
the subsidies (compare with Figure 3) in the sense that authorities can set a renewables
target, from which the TGC market will find the sufficient certificate price necessary to
reach the target. The difference between the spot market mechanism and the TGC market
mechanism, is the lack of short-term regulation by the unit commitment loop B/. As we
will see, this makes the price formation in the TGC market sluggish, because the only way
to adjust the supply goes through the capacity acquisition loop B6 (Figure 9). The exog-
enous stochastic variation of, primarily wind will also increase the price fluctuations sig-
nificantly, and this is the main motivation for allowing banking of certificates. The
compliance period of the Swedish TGC market is 1 year. If TGC obligations are not met
during this period, consumers are charged by a penalty fee that exceeds the TGC price.
Figure 10 shows the stock and flow diagram of the TGC market, where the linkage to the
electricity market is indicated. The TGC market sector is similar in structure to the spot
price - except that a maximum and a minimum price is introduced in eq. (6.1), and the
TGC market does not need to clear as frequent as the electricity market. A major differ-
ence between conventional sources and renewables is the ability of controlling genera-
tion. Wind turbines and small scale hydropower are not operated by marginal production
costs, but by wind and rainfall, whereas bio and waste generate electricity as a by-product
and are rarely scheduled by electricity prices, but by heat demand.
The purpose of the TGC market is however give long-term price signals for development
of new capacity, where market-clearing obligations should be met annually. In the ab-
Figure 10. TGC market SFD. Interactions with “Electricity spot market “indicated on
the figure. In the following simulations, spot market price is constant, so that the futures
market price is constant. TGC consumption is therefore only influenced by changes in
TGC price at this stage. TGC price does however influence consumer price and thereby
demand.
Electricity
Electricity
spot market
Electricity
spot market
sence of a short-run price adjustment process, long-term price formation could turn out to
be problematic.
(6) TGC market
6.1
6.2
6.3
6.4
6.5
6.6
6.7
6.8
TGC price = MIN(Max price, MAX(Indicated TGC price,Min price)) [NOK/MWh]
Max price = 350 [NOK/MWh]
Min price = 0 [NOK/MWh]
Indicated TGC price, = Indicated TGC pricey + | chg in TGC price; dt [NOK/
MWh]
chg in TGC price, = Indicated TGC price:(TGC demand - TGC sales rate,)/TGC AT
[NOK/MWh/da]
TGC AT =2 [wk]
TGC pricey = 178 [NOK/MWh]
Yearly avg TGC price = SLIDINGAVERAGE(TGC price, 1) [NOK/MWh]
The TGC target must be set for some time horizon ahead, preferably a rolling time hori-
zon. The (almost) linear curve in Figure 11c) shows the demand for TGC certificates re-
sulting from a linearly increasing TGC target measured as the percentage TGCs of total
generation. For our model, we start with the present share of 6 % renewable generation
(2003), to reach 24 % in 2020.
(7) TGC demand
7.1 TGC demand = TGC target-demand [TWhAr]
7.2. TGC target = GRAPH(TIME , Starttime, 17, {6,24}) [%]
The profitability assessment is the same as defined in equation set 4. /-4.3, except that
TGC prices replace the feed-in tariff in the subsidy scheme so that equation 4.4 changes
to:
4.4 Support scheme = TGC price [NOK/MWh]
Due to the problems of intermittency of wind and small-scale hydro, some kind of storage
possibility is desirable to ensure a smooth price formation. The Swedish TGC system al-
low unlimited lifetime of certificates. Another issue is the borrowing of certificates from
year to year. We include the possibility of banking in equation set (8). Certificates are is-
sued upon registered generation (8./). A simple rule of selling TGC’s is introduced in eq
(8.6). Sale of certificates depends on the expected generation, and the price of TGC’s in
the market using a Cobb-Douglas function. With this formulation, we assume traders to
change their sales rate by 0.8 per cent for each per cent TGC price change relative to the
reference value. The reference value of TGC price is formulated as an adaptive smoothing
of approximately TGC prices during the last three years. Alternatively, in case of rational
traders - the price elasticity of supply would be very high, and a reference price that would
achieve market equilibrium.
(8) TGC trading
8.1 TGC volume, = TGC volumeg + | (TGC issued,- TGC sales rate) -dt [TWh]
8.2 TGC volumey = 25 [TWh]
8.3 TGC issued, = generation re [TWhAr]
84 TGCsales rate, = MAX(expected generation effect of TGC price on sales rate, expi-
ration rate) [TWhAr]
8.5 Min sales time = 1 [wk]
8.6 effect of TGC price on sales rate = generation re-(TGC price/Yearly avg TGC
price, yyPrice elasticity of TGC supply [TWhhr]
8.7 Price elasticity of TGC sales = 0.8 [1]
8.8 expiration rate = DELAYMTR(TGC issued, Valid lifetime, 3) [TWhAr]
8.9 Valid lifetime = 1e13 [yr]
71 TGC market simulation results
The below simulations show the dynamics of the TGC market defined in the equation sets
(1)-(8). As can be seen in Figure 11a), The TGC prices oscillates. This is mainly due to
the time delay in the expectation formation. We did not explicitly model the time delays
involved in the capacity acquisition!. However, these time delays are implicitly modelled
in the “effect of profitability on investment rate” function (see Figure 7).b).
Figure 11 TGC market dynamics introduced in 2003. Constant wind and constant
electricity spot price. Reference curves are market with * and corresponds to the
development of renewable generation and capacity with a fixed feed-in tariff of 178
NOK/MWh.
a)TGC price b)TGC volume. Initial volume equals 1 year of demand
Fee pea 2
Znnaxprce
Le vaitteos —apartone —rtars —arapboan snuteoe —vathnos —anitowe arbors —wirhoon
c) TGC demand, sales rate and physical generation, Ref- d)Renewable capacity. Reference (*) shows capacity
erence (*) shows generation development under the sub- development under the subsidy scheme
sidy scheme.
Twhiye ww
ad Eyee demana | ||7°0°
esreeton ||| 022 Ea Tanevabien
2 Bree sate te Bearers |
i/hoos a/ftoos iyavtoio aavtors —1a/bo20 2 atoay Sayaploosy ayaystae! Sayajsorsi apajuezo:
Figure 11a) show long-term fluctuation of TGC prices. This behaviour is attributed to
the time delays involved in forming expectations about prices and the time delays for new
capacity. The declining price trend is explained by the price rise from increasing obliga-
tions to purchase TGC’s, which makes consumer prices higher. In Figure Jc) we can
observe how the physical generation of renewables deviate from the TGC target in the (al-
most) linear curve starting from 2003. Sales rate and TGC demand matches perfectly, be-
cause deviations when storing and forwarding certificates are possible as in Figure 1b).
The development of renewables under a subsidy scheme is also shown for comparison as
a reference case. With one-year average clearing time of the market, the TGC sales rate
also deviates from the TGC demand. A smaller TGC Adjustment time (6.6) results in
smaller deviations of TGC demand and TGC supply, but TGC prices fluctuations will in-
crease, both in amplitude and frequency. Similarly, other delays along the capacity ac-
quisition loop will influence the prices and TGC demand/supply balance, such as the
expectations of future prices and the time delays of capacity acquisition.
The supply and demand obligation at the beginning of the TGC period will be important.
If there is a big gap between generation of renewables and TGC obligations, prices start
rising (or down if there is overcapacity) because it will take at least two years before suf-
1. Time delays for building new renewable capacity amounts to 1.1 year for applications and at
least half a year for construction.
ficient capacity can become available. This effect is more dramatic with restrictions on
borrowing and banking. This indicates the importance of setting realistic targets for
TGC’s in the beginning, and to allow flexibility in terms of banking and borrowing. Con-
straints on prices are also important to protect consumers from having prices skyrocket-
ing. It would also be important to secure or guarantee prices during the first years of the
TGC market. This issue has also been a major concern in the discussion of the implemen-
tation of TGCs in Denmark and Sweden. Both countries have some subsidy scheme for
various technologies. Current proposals are to establish TGC price caps and price floors
that gradually will be phased out during the first years.
As a result of the price fluctuations, capacity develops in more or less pronounced boom
and bust cycles (see Figure 11d).
Figure 11b) shows the variations in TGC volume during the planning period. The main
purpose of banking is to cope with the intermittency of renewable generation. Various
options on banking and borrowing could alter the TGC sales strategy previously defined
in equation set (8). Suppose unlimited banking is allowed, but no borrowing. If TGC sup-
pliers decided to reduce their sales of TGC’s prices would rise while there is few possi-
bilities to increase the supply of TGC in the short run until sufficient capacity is added.
To study the possibilities for market power and trading strategies under various market
designs, we develop a more detailed description of buying and selling in the TGC market.
8 Market design
The intermittency of renewables has been the main concern in the discussion of price vol-
atility of the TGC market.
Price volatility has been discussed qualitatively in Nielsen & Jeppesen (2003), Bye et al
(2002) and in the Swedish white paper SOU 2001:77. The discussions take a traditional
economic point of view, and the main concern in this discussion is the inelastic curves of
supply and demand. As discussed in chapter 6.3, renewables do not operates after mar-
ginal costs, and the adjustments of the supply curve comes from investments in new ca-
pacity, involving time delays of permits and construction. Price will therefore fluctuate
with the variability of generation from year to year for the consumers to meet their yearly
obligations. Several options are considered to reduce the price volatility of TGC’s:
Figure 12 Price volatility in the TGC market
a)Inelastic demand and supply b)Elastic demand and supply from banking/borrowing
TCG Price LRMC renewables TOG Piece LRMC renewables
Max price|__ Demand ' Max price|Demand
calm yr] J windy yr
price * banking,
<= © borrowing
Min price|
TWhiyr
+ Include different technologies with operational costs.
By including biomass that operates after marginal costs, the problem of price volatility
can be reduced. However, the potential for such technologies are questioned. Biomass
and waste incineration are usually cup! plants generating electricity as a by-product of
their heat generation. Furthermore, income from electricity generation would come from
both the spot market and the TGC market, so the likelihood that such units would be sen-
sitive to TGC prices are questionable.
+ Maximum and minimum prices
Price caps and price floors secure the consumer/producers against high/low prices (see
Figure 12a). This arrangement is especially important in the introduction phase to reduce
tisk for investors.
+ Banking and borrowing
To improve the stability of supply, certificates can have a valid lifetime longer than the
yearly compliance period of consumers. This mechanism is referred to as banking, where
producers and consumers can choose in periods of low prices (i.e. windy years) to store
certificates for later years. Banking is assumed to have a price smoothing effect, and in
the proposed Swedish TGC market, certificates have unlimited lifetime. According to the
above-mentioned studies, the price elasticity of the supply and demand curves increase as
1. Combined heat and power generation units.
illustrated in Figure 12b). Similarly, allowing participants to fulfil their obligations in the
future (similar to futures contracts), would also increase the elasticity’s of demand and
supply. The disadvantage of borrowing is that some regulation must secure that these fu-
ture obligations are met, for instance by imposing penalties. Several arrangements of bor-
rowing mechanisms are possible, but the principle remains the same, that is to increase
flexibility of supply by allowing trading with future TGC production.
9 Laboratory experiments of TGC trading
The European Renewable Electricity Certificate Trading Project (RECeRT) took on the
experimental approach to study the influence of price caps, banking and borrowing on
price volatility ina TGC market. The first experimental economics study, reported in
Schaeffer & Sonnemans (2000) - showed that unlimited banking in combination with high
price caps could induce price crashes and increased volatility rather than the opposite.
Price caps and borrowing and banking all had an influence on the price volatility. The
best results were obtained when only borrowing was allowed. A larger internet based ex-
perimental study involving over 140 participants was also conducted under the same
project, and the resulting price history is shown in Figure /4b). Unfortunately, the market
turned out to be short most of the simulation period, and the TGC prices naturally settled
one the maximum price, which makes this experiment inconclusive with respect to price
volatility. Another initiative, the RECS project has been trading TGC’s at an internet-
based exchange for several years, but they did not report on price formation.
Figure 13 Laboratory experiments on TGC trading at NTNU
At NTNU, we set up a network simulation game and invited (wind) power engineers and
energy policy administrators for a laboratory experiment on TGC trading. The model
used is the same as the TGC market model shown in Figure /0, except that 5 buyers and
5 sellers interactively controlled the purchase and sales rate for each year through a user
interface with relevant information on price, TGC’s issued, volume etc. (see Figure 13).
The model made investment decisions endogenously.
Figure 14 Laboratory experiment results of TGC markets
a)Price crash from the NTNU experiment. Low sales rate b)Price formation in the ReCERT internet experiment
in the first years lead to persistent high prices and over involving 140 participants. Prices stabilised on maxi-
investment in new capacity, which caused the price crash_mum price as the market was short throughout the whole
in the subsequent years. simulation period. With respect to price stability, this
experiment was therefore inconclusive (Source:
TGC price
——o ne pect ReCERT, 2001)
-1-TGC_price
\ -2-Maximum_price
-g-Minimum price
scale, lowes!
Loves oeaty ae
] Green Cato
|
|
|
Traded pri
Jay Yay tn Tan Tan Jay an an an Jaan
near 2002 200s Boos BME 20200720820
c)Price crash in a computer laboratory experiment. 100% d)Same experiment as in d) but with a lower maximum
banking and 50% borrowing allowed. Equilibrium price price and only borrowing This market design lead to a
indicated by lower horizontal line. Upper horizontal line price formation close to equilibrium.
indicates maximum price (penalty). (Source: ReCERT,
Schaeffer & Sonnemans, 2000)
high. 100% banking. 50% borrowing 7a
‘a ©
10 Modelling trading in a TGC market
Trading is constrained by the valid lifetime of a certificate (banking) and the possibilities
of borrowing. In addition price caps represent constraints on the TGC price and buyers
need to meet their TGC obligations defined as a percentage of electricity sold to the con-
sumer. Trading with TGC certificates can be viewed as a problem of profit maximisation
for the seller and a cost minimisation of the buyer. The optimisation problem is, however
not simple, depending on the market design. If the TGC’s are valid for one year, we
should expect all certificates to be sold that year, and the price will then depend upon the
conditions of wind and rainfall. If on the other hand, certificates have unlimited lifetime
- it is possible to hold back certificates over longer periods of time, which opts for various
trading strategies. The price of certificates in the long run should converge to the long-
run marginal costs of new renewables, but since developing new capacity takes time and
TGC obligations must be met every year, it is possible to hold back certificates to stimu-
late price increases and prices can persist far from equilibrium price. If borrowing is pos-
sible, holding back certificates would lead to more borrowing if prices exceed the
expected long run marginal costs of new generation and thus reduce market power from
suppliers.
Figure 15 shows the CLD of buying and selling in the TGC market. The market consist
Figure 15
Trend sensitivity
__Trend adjustment elle
time i)
Trend sensitivity bu ——~ —Pttect of price trend
onsaleg
£R1-Demand fom TGC price trend
‘nen followers 2 Banking fom
= A, trend followers,
‘end atue trader time horizon
: Price elasticity Expected
| Price elasticity of Pree elastcity of sa sie
% TGC target i Reference value oo Renewable
= Demand Purchase + get
, PCT arin aire |, ‘generation
- + Ge ae comet
Effect of price off Efet of price on
Demand TGC
purchase | 7 sales
\A rose ; obs
89 -Demand balance 4 We 1510-Supply balance TGC sale rate TGC Volume seller
{ from rele trading from vale ring “+ B8- Volume aust
27-Velume begs ‘Stem borrontg, eine
\ adjusiment from TOC Purchase land desired coverage
4 serrening etine : toler
and desired coverage by ie
TOC volume buyer +
buyer Max borrowing
buyer, a Max b
ssired coverage -” ction
— buyer -_
Desired coverage
seller
of buyers and sellers, each constrained by possibilities of borrowing and banking. These
regulations constrain their TGC volumes represented by Loop B7 and BS. Their respec-
tive sales and purchase strategy is represented by a combination of two loops: We hypoth-
esize Value trading (B9,B/0) and trend following (R/,R2) to be the trading strategy of
buyers and sellers in the TGC market. Taking the seller as an example, value trading
means that you have a reference value to which you compare the market price. If the mar-
ket price is higher than your evaluation of the value of the certificate, you will sell more,
otherwise you will sell less. On the contrary, trend trading is a reinforcing process. The
steeper the trend, the less you will sell because you can probably get a higher value for the
certificate later on, which will reinforce the trend further.
The stock and flow diagram is shown in Figure 16. Equation set (9) defines the sellers
TGC volume management. Issuing of certificates equals the generation from renewables,
whereas the sales rate of TGC’s take the expected generation from renewables as a refer-
ence (eq. 9.4), adjusted by the effect of price, the effect of trend and the constraints im-
posed on borrowing, banking and volume control due to risk aversion. If certificates
expire, they will be sold immediately according to eq. 9.4 and 9.10. When borrowing lim-
its are exceeded (eq. 9.5- 9.9), the sales rate is adjusted to keep within limits. The maxi-
mal borrowing fraction is defined as a fixed percentage of the expected generation (eq
9.8).
(9) TGC trading seller
9.1 TGC volume seller, = TGC volume sellerg + J (TGC issued, - TGC sales rate, ‘dt
[TWh]
9.2 TGC volume = 12.5 [TWh]
9.3 TGC issued, = generation re [TWhAr]
9.4 TGC sales rate, = MAX(expected generation effect of TGC price on sales rate: effect
of TGC trend on sales rate-adj for borrowing seller+TGC vol adj seller,expiration
rate) [TWhAr]
9.5 adj for borrowing seller = MAX(borrowing margin/Borrowing AT,0) [TWhAr]
9.6 Borrowing AT = 1 [mo]
9.7 borrowing margin = (max borrowing seller - TGC volume seller) [TWhAr]
9.8 max borrowing seller = Fraction max borrowing: expected generation [TWh/yr]
9.9 Fraction max borrowing = -0.5 [TWh TWhiyr)]
9.10 expiration rate = DELAYMTR(TGC issued, Valid lifetime, 3) [TWhAr]
The same structure applies to buyers of TGC’s defined in equation set (10). The buyer
must control his TGC volume through purchases, whereas the demand obligations control
the consumption rate of TGC’s. We assumed that the buyer adopted the same strategy as
Figure 16 Stock and flow diagram representing buying, selling and trading strategies of
TGC certificates.
Trend fotlower strategy
ese:
rec target
Desired coverage
¢— 6%»
the sellers, that is value trading and trend following.
(10) TGC trading buyer
10.1 TGC volume buyer, = TGC volume buyerg + | (TGC purchased, - TGC consump-
tion rate, ‘dt [TWh]
10.2 TGC volumeg = 0 [TWh]
10.3 TGC issued, = TGC purchase rate [TWhAr]
10.4 TGC purchase rate, = MAX(TGC demand-effect of TGC price on purchase
rate-effect of TGC trend on purchase rate+adj for borrowing buyer+TGC vol adj
buyer,expiration rate) [TWhAr]
10.5 adj for borrowing buyer seller = MAX(borrowing margin purchase/Borrowing
AT,0) [TWhhyr]
10.6 borrowing margin buyer = (max borrowing buyer - TGC volume buyer) [TWh/yr]
10.7 max borrowing buyer = Fraction max borrowing: TGC demand [TWhAr]
10.8 expiration rate buyer = DELAYMTR(TGC purchased, Valid lifetime, 3) [TWh/yr]
In the following we will discuss the trading strategies more closely
The rational expectations paradigm assumes that traders have complete knowledge of all
the economic relationships, they have access to all available information that needs to be
taken into consideration, and they have enough time and resources to do so in order to
make optimal decisions on buying and selling. Any price fluctuations are exogenously
caused by new information of fundamentals (i.e. breakthrough’s in technology or excess
generation of TGC from last month) The rational expectations paradigm can be a suffi-
cient approximation in many cases, but this assumption does not always hold.
A more realistic assumption is the paradigm of bounded rationality, where traders are re-
stricted in terms of resources, time and cognitive capacity to make optimal decisions on
buying and selling. Their decisions are based on a limited, selective set of information
available to them. System dynamics and cognitive science provide us with theory to mod-
el boundedly rational agents by capturing their decision rules. Heuristic rules for trading
could be inferred from analysing data from the laboratory experiments on TGC markets.
Unfortunately, the quality of the conducted experiment at NTNU was not sufficient in or-
der to use it for estimating decision rules, and there were too few experiments. Still, some
observations and experience can be used to hypothesize their decision rules.
Price dynamics of common trading strategies in asset markets has been studied in emerg-
ing fields of economics (see for instance Farmer & Joshi, 2000 and Gaunersdorfer, 2000).
Their studies show that simple and commonly used trading strategies based on adaptive
belief endogenously generate price fluctuations and statistical behaviour of prices as those
observed in real world markets. This indicates that representation of simple decision
rules can capture characteristic behaviour of markets that in turn can be utilised for anal-
ysis and design. In the following, we will describe two trading strategies, namely value
trading and trend following.
10.1 Value traders
Value traders make subjective evaluation of the “fundamental” value of the asset and be-
lieve the market sooner or later will adjust to this value. They attempt to make profits by
selling if they think the market is overpriced and buying if they believe the market is under
priced. These traders are called “fundamentalists” in the sense that they make an assess-
ment of the value of a TGC asset from the “fundamentals” of the market. In a TGC mar-
ket, fundamentals are information about new project developments, permits, contracts,
cost of new technologies etc. The “fundamentals” of renewables are fairly reliable in
comparison to stock markets. However, such analyses would require time and resources.
The most influential source of information is perhaps recent prices and thus we can rep-
resent the fundamental value denoted as Perceived TGC value as an adaptive expectation
of recent prices in eq 11.2., where the Value trader adjustment time is the average smooth-
ing time, which should be of the same magnitude as the time delays involved in construc-
tion of new capacity, because only new capacity can adjust price in the long run. The
price elasticity of TGC supply indicates how many per cent a seller would change his sales
in response to price changes. The effect of TGC price on sales rate function is a multiplier
used in the sales and purchase policies.
(11) Value trading
11.1 effect of TGC price on sales rate = (TGC price/Reference price)?" elasticity of TC
supply [TWhAr]
11.2 Reference TGC price = DELAYINF(TGC price,Value trader AT) | [NOK/MWh]
11.3 Value trader time horizon = 3 Lr]
11.4 Price elasticity of TGC sales = 1.5 [1]
The same structure applies to buyers of TGC’s as well, except for a change in parameters.
We assume the same parameters as for sellers, except for the price elasticity of demand
Price elasticity of TGC demand = -1.5 [1]
(For the full set of equations for value traders, see appendix)
10.2 Trend followers
Trend followers have shorter time horizon than those of value traders. They believe that
prices will fluctuate but that the market has some inertia that can be exploited. A seller
would then hold his position of TGC’s if the trend is positive, and sell when the trend is
negative. Conversely, a buyer would hold his position when prices are falling, and buy
when prices are rising:
(12) Trend followers
12.1 effect of price trend on sales = (1+ TGC price trend) Tend elasticity f1]
12.2 Price trend = TREND(TGC price, Trend AT) [Ar]
12.3 Trend adjustment time = 1 br]
12.4 Trend elasticity = 2 [1]
The trend is observed over some time interval. It is not likely that short-term price fluc-
tuations would occur in the TGC market, as demand obligations must be met once a year,
but the trend horizon must be less than the expected average time to add new capacity.
The trend adjustment time is therefore set to 1 year in eq (12.3).
The NTNU experiment strongly suggested that the trend strategy to be dominating, per-
haps because this is the simplest strategy and only requires information on previous pric-
es. We hypothesize that traders would both look to the value and to the trend when trading
in a TGC market. If the price is high and the trend is positive, the seller would probably
sell less than if the trend is pointing downwards. Similarly, if the price is low and the trend
is pointing up, the seller would be more inclined to wait than if the trend is going down.
This trading strategy is represented through eqs 9.4 and 10.4:
TGC sales rate, = f(expected generation re-effect of TGC price on sales rate effect of TGC
trend on sales rate, ... other factors ... )
In asset markets, trend trading and value trading is separated by their difference in time
horizon. In a TGC market, however, the slow dynamics of new capacity makes these
strategies relevant on the same time scale.
10.3. Managing risk by controlling the TGC volume
Using the trend and value strategies, TGC volumes could grow very large, which repre-
sent large risks if the market should crash. Neither the buyer nor the seller should keep
large volumes of TGC’s over a long period. This risk aversion can be taken into account
by adjusting the TGC volume, with an adjustment time of up to 3 years as in equation set
(13)-(14). The buyer would like to have some coverage of TGC’s
(13) Volume adjustment seller
13.1 TGC vol adj seller = (SLIDINGAVERAGE(TGC vol seller, 1) - desired volume sell-
er)/Volume adj time seller
[TWhhyr]
13.2 desired volume seller = expected generation’ Desired coverage time [TWh]
13.3 Desired coverage time seller = 6 [mo]
13.4 Volume adj time seller = 3 [yr]
(14) Volume adjustment seller
14.1 TGC vol adj buyer = (SLIDINGAVERAGE(TGC vol buyer, 1) - desired volume buy-
er)/Volume adj time buyer
[TWhhyr]
14.2 desired volume buyer = TGC demand: Desired coverage time buyer [TWh]
14.3 Desired coverage time buyer = 6 [mo]
14.4 Volume adj time buyer = 3 br]
11 — Simulation results
Consider the simplest case withouth borrowing and | year certificate valid lifetime. We
assume a price elasticity’s of -1 and | respectively for the TGC price elasticty of buyer
and seller. No trend following strategy is applied. Figure 17 shows a Monte Carlo sim-
ulation based on these assumptions, where wind energy accounts for the stochasticity
from the renewables (see Figure 17e) Prices increase during the first years, peaks and de-
cline well below the equilibrium price of 178 NOK/MWh. In the calmest years, prices hit
the price cap of 350 NOK/MWh. The average price development however, can be com-
pared with that of the TGC market model in chapter 7 (see Figure //). This market design
does not allow for banking and borrowing, as can be seen by buyers’ TGC volume and
sellers TGC volume in Figure 17c)-d)
When we allow for banking, Figure 18 shows almost the same behaviour during the first
years, since there is shortage of renewable capacity in proportion to the TGC target. After
2010, buyer’s and seller’s start banking. No trend strategy is used in this simulation. It
appears that banking does not reduce price volatility during the first years, due to the ini-
tial capacity deficit during the first years When there is a surplus of TGC’s, banking tend
to smooth prices to some extent (compare end periods of Figure 17a Figure 18a).
In the next simulations, we include the trend strategy. Traders now consider both the
Figure 17
TGC 1 year valid lifetime, no borrowing. No trend followers
a) Price b)Capacity development
HY 60,000
[rat pace rary Sapa einaml pa
ro \ ee es Pre scape rpeveies 8 Peres)
ss ree pak erga) ooo Ec cpuer eresolcs ieee)
ree wer ca enc [enpeey remus Gi Ponta)
[Fcc pe tom) = |aohaveecmrcd
100- Baad sano toa
c) Buyers volume is restricted by expiration rate
d) Sellers volume is restricted by expiration rate
Fae vatume buyer High)
J Tec volume buyer (75% Percentile)
soo IEC volume buyer (Average)
IGE volume buyer (25% Percentite)
(GC volume buyer (Low)
[Tee volume seer (High)
J TGC volume seiter (75% Percentile)
oo 'GC volume sellar (Average)
[Tac volume sete (Low)
e)Normalised wind energy used to represent stochastic generation. Wind energy series adapted from Tande et. al &
Vogstad 1999)
—Soarastiewaraten (sh)
= stocnasticvanation (75% Percensle)
—stocnasticvanation (Average)
—stosrestc aration (25% Percentile)
= stocnastic venation (Low)
— stoznasticvanation (Stundord Deviation
Figure 18 Infinite banking, no borrowing. No trend followers
a) Price
b) Capacity development
nox
“| [lh
TN
prea (HM)
C pees (75% Percentile)
C pres (Average)
$C prea (25% Parcentie)
|—rac price (Low)
[capacity renewables (High)
— capacity renenbies (75% Percentile)
capacity renenabies (Average)
= capacty renewables (25% Percentile)
|— capacity renenabies (Low)
Figure 18 Infinite banking, no borrowing. No trend followers
c) TGC volume buyer d)TGC volume seller
ae, [Fae volume buyer (righ) 2am oa
x o
+—+—_+—_+—_ +++,
price and the trend of the TGC when buying and selling. The price elasticity of trend is
set to 2. Thus, | % price increase will increase the sales rate by 1 %, and 1% increase in
trend will reduce the sales rate by 2%. Similarly the buyer will increase his purchase rate
when the price is dropping, but if the trend is negative, he will delay his purchase to wait
for even lower prices. If we compare Figure 19d with Figure 18d, we can observe that
the seller is slightly more restrictive in selling, which results in a deficit on the buyer side
(compare Figure 19c with Figure 18c) - enough to drive up prices in Figure 19a, where
the average price nearly hits the price cap. The conditions of capacity deficit in the start
of the TGC period triggers the trend following strategy. In the case of initial overcapacity
(or TGC targets less ambitious), a downward price trend would not have caused a similar
problematic price drop, because the buyers are anyhow obliged to purchase certificates.
Just after the price peaks, buyers begin to accumulate certificates due to the negative
trend.
Figure 19 Infinite banking, no borrowing with trend strategy
a)Price b)Capacity development
c)TGC volume buyer d) TGC volume seller
am - 0a See volume seter (Hah)
5
To explore these strategies furhter, Figure 20 shows a typical simulation run with the
same market design presented in the Monte Carlo simulation. In Figure 20, sellers start
with an initial TGC volume of 25 TWh TGC’s, and they do not increase sales rate even
though the prices are rising. In 2006, buyers cannot fulfill their obligations, and their vol-
ume is negative, while the sellers in fact choose to bank TGC’s. The price hits the price
cap, and the sellers reduce their TGC volume towards 2010, but by this time, a significant
amount of capacity has been developed, and prices fall below the initial equilibrium price
of 178 NOK/MWh. Figure 20c show the relative effect of the trading strategies on sales
rate. In this simulation run, the TGC trend effect dominates the sales strategy, while the
value trading strategy serves to moderate the trend strategy.
Figure 20 Typical simulation run, infinite banking, no borrowing with trend strategy.
a) Price b)TGC volume buyer and seller
NOK/MWh wh
Tec price sol
= Max price
a/ijzooo_ 71/2005 _a/ayzo10 4/4/2015 4/3/2020 uysyz000 4/3/2005
c)Effect of value and trend strategies on sales rate d)Capacity development
[retiea or TGC pre on sales rate
| etrect or trend on sales rate
| total ettect on sates rate
sss/d000 wyafhoos waftoxe s7ho1s aa/bo20 1/3/2000 3/a/2008_4/3/2010 4/3/2015 4/4/2020
Now lets consider allowing 50% borrowing of certificates, that is - buyers and sellers can
borrow up to 50% of their respective yearly obligation and present TGC yearly genera-
tion. The results in Figure 21 shows that prices show less volatility in response to sto-
chastic variation from in renewable generation.
Figure 21 Banking, 50% borrowing trend
a) b)
QQ “— [—Capaaty renewables (Woh)
200. capacity renewables (75% Percent)
— — capacity renewables (verge)
_ — | Feapsetyrenewabes (25% percenie)
soo —capactyrenewabies (Low)
c) d)
we ites ‘volume buy ome [—TGC volume seller (High)
of ° —
The trend strategy in a market with up to 50% borrowing does not have the same impact
as when only banking is allowed. In Figure 22b, buyers now borrow certificates as prices
rise. Comparing with Figure 22c, the seller tries to reduce his sales rate during the first
years, but this strategy does not have a sufficient impact on the price any more. And after
some years, the value trading strategy becomes the dominant one.
Figure 22 Typical simuation run, infinite banking, 50% borrowing with trend strategy.
a)Price
ee
ears
teen
c)Effect of value and trend strategies on sales rate
b)TGC volume buyer and seller... Max borrowing limits
for buyers and sellers shown as curve 3 and 4.
[-F ice volume seer
2 max borroving seller
3 max borrowing buyer
J+ 13° volume buyer
ysy2000 1/3/2005
d)Capacity development
[revtea or TGC pre on sales rate
|2-etrect or trend on sales rate
[3 total ertect on sales rate
1/1/2000 1/3/2005 3/3/2030 1/4/2015 3/3/2020
If we remove the possibilities for banking (Figure 23), the results show are similar, except
Figure 23 No banking. 50% borrowing with trend strategy
a)Price b) Capacity development
|— TGC price (High) sone |— Capacity renewables (High)
200 |—ree pce (75% Percentie) = capacty renewables (75% Percent)
> [ee pac css Pecenutey ||| 40.90 ZZ: | capeatyrenenvies (2 Feeney
ue SS coe ten = ape reneeaes (om)
c) TGC volume buyer d) TGC volume seller
= [— TGC volume buyer (High) am [TGC volume seller (High)
Tee volume buyer (75% Perce) Te volume ele (75% Percentie)
200. [Tee votume buyer cass Percent) |] 100 ee voume seer (2s Percent
zz [Ftac atime boyeriom ronan — |mrccwenmetatrtowt
oe . | —~neen
that less TGC’s are stored at the buyer in the end of the TGC period. Price variations from
year to year do not seem to be significant compared with the simulation runs including
borrowing.
The results of these simulations suggest that allowing banking to reduce price volatility
from the stochastic variation of renewables could in fact increase price volatility that arise
from the strategic behaviour that becomes available. This effect does not appear if we as-
sume traders only to use the value trading strategy described in /0./. If, however, the trad-
ers also apply some trend following strategies as described in /0.2, price crashes will
likely to occur. Laboratory experiments strongly supports the hypothesis that traders to
some extent do apply trend strategies.
The problem can be avoided by allowing borrowing, with or without borrowing. Banking
seems to shift market power in favor of sellers, whereas borrowing adjust this assymetry
between buyers and sellers. These findings support the conclusions found in Schaeffer &
Sonnemans (2000).
12 Interactions of the Spot market and the TGC market
A TGC system with mandatory demand has two partial effects on the electricity market.
First, it produces extra revenue for producers of renewable electricity. This will increase
supply of electricity and reduce electricity demand from other sources. Secondly, there
will be a partial increase in the consumer price for electricity for given wholesale electric-
ity prices since the consumer also has to buy certificates. With some price elastic demand,
the wholesale prices for electricity are reduced. Hence, both the partial effects of the TGC
system tend to reduce the income for traditional power producers. The effect on the con-
sumer prices is, however, ambiguous: electricity prices net of certificates goes down, but
the additional costs of certificates increase the consumer price. The total impact of TGC’s
on consumer prices thus depend on price elasticity’s of demand, price elasticity of both
electricity supply and TGC supply, and the price on certificates. Jensen & Skytte (2001)
and Bye et.al (2002) reports that for some smaller share of TGC obligations, consumer
prices can actually be reduced.
In contrast to subsidies, TGC prices will influence consumers directly through the con-
sumer price of electricity, which is now both the payments from electricity generation and
the TGC market. The consumer price now consist of the electricity spot price, plus the
fraction of renewables that must be purchased (/5.2-/5.3). We can thus expect a reduced
demand due to price elasticity of demand (see loop B4 - TGC demand balance in Figure
24). A reduced demand will also reduce spot prices (loop BO - Demand balance, Figure
24). This means reduced generation from thermal units, because thermal generation is
sensitive to spot prices (loop B1 - Unit commitment). In the long term, capacity acquisi-
tion will also be influenced by sustained lower spot prices through loop B2 - Capacity ac-
quisition thermal and loop B3 - Capacity acquisition renewables. However, investment
in renewables are stimulated by B7 - capacity acquisition from TGC price, which more
than compensates for the reduced spot price. Finally, a more subtle interaction is discov-
ered through loop R1 in Figure 24. Investments in renewables, increase total generation,
which reduce the spot price. However, a reduction in spot price stimulates demand, which
also means an increase in the TGC demand leading to a higher TGC price and therefore
increased profitability of renewables and investment in new capacity, which increase gen-
eration and so forth. The importance of this reinforcing loop is not yet examined. A de-
tailed feedback dominance loop analysis could reveal the relative importance of these
previously mentioned loops, using the proposed method of David Ford (Ford, 1999). The
TGC market and the electricity market interacts through the consumer price. The subsidy
scheme is now replaced by the TGC price (see eq 4.4)
Figure 24 Causal loop diagram showing the loops between the interacting spot market
and the TGC market.
=f ERMC thea
Expected LRMC
profitability thermal, renewables
Thermal apacy 7, woe
SRMC thermal 82-Capaciy pected profitability
Pesrinare reneabls
Price las of
‘cmand yhacone —t
UA ie
ve omni ; ‘acquistion Renewable’
s
‘BO~ Demand ‘a ‘otal generation renewables capacity
Demand “hace S~—_Total genera vied
Theimal generatic
B3- Capacity
Int, n2,R3 -Price 7
reductions through
generation Renewable
BS TGC demand Consumer pric generation
ce TGC share Bt Demand |
reductions from TGC :
share CF renewables:
% TGC target
* nA B6- price +
Demand TGC Seen TOC TGC volume
stored
PTC sales #7-16CvoMme |
Toc wae aay
25,86,89 - demand -——
Increase from Increased
renewables
(15) Demand side including a TGC obligation (replaces equation set 4)
15.1 demand = Demand ref :(consumer price/Reference price)?"*¢ elasticity of demand
[TWh/yr]
15.2 consumer price = spot price + renewables share: TGC price [NOK/MWh]
15.3 renewables share = generation re/generation tot [1]
15.4 Demand ref = 420 [TWhAr]
15.5 Reference price = 200 [NOK/MWh]
15.6 Price elasticity of demand = -0.3 [1]
In the profitability assessment of renewables, the support scheme of 178 NOK/MWh in
subsidies (4.4) is now replaced by the TGC price:
4.4 Support scheme = TGC price [NOK/MWh]
12.1 Simulation results, coupled markets
Integrating the TGC market with the electricity spot market developed in chapter 6 yields
the results presented in the figures below. In Figure 25 there is no borrowing while TGC
certificates have unlimited lifetime. In Figure 26, 50% borrowing is allowed and certifi-
cates have unlimited lifetime. In both simulations, the trend strategy is used in addition
Figure 25 Banking, no borrowing and trend strategy included. Price and generation
shown as yearly averages
a)Consumer price, electricity spot price and TGC price
NoK/MWh
avg consumer price (High)
—avg consumer price (75% Percentile)
J+ ave consumer price (Average)
—avg consumer price (25% Percentile)
—avg consumer price (Low)
— Yearly avg price (High)
— Yearly avg price (75% Percentile)
J Yearly avg price (Average)
— Yearly avg price (25% Percentile)
— Yearly avg price (Low)
— Yearly avg TGC price (High)
— Yearly avg TGC price (75% Percentile)
| Yearly avg TGC price (Average)
— Yearly avg TGC price (25% Percentile)
—Yearly avg TGC price (Low)
°
Jan 01, 2000 Jan 01, 2005 Jan 01, 2010 Jan 01, 2015 Jan 01, 2020
b)Total generation/demand, thermal generation and renwable generation development. (Demand coincide with gen-
eration)
Twhy/ye 1 L
a eer SSS
400 —Yearty avg total generation (High)
— Yearty avg total generation (75% Percentile)
| Yearly avg total generation (Average)
— Yearty avg total generation (25% Percentile)
300 — Yearty avg total generation (Low)
— Yearty avg generation th (High)
i early avg generation th (75% Percentile)
* Yearly avg generation th (Average)
200 = |— early avg generation th (25% Percentile)
— Yearly avg generation th (Low)
— Yearly avg generation re (High)
early avg generation re (75% Percentile)
a” |=. Yearly avg generation re (Average)
5 —Yearty avg generation re (25% Percentile)
—vearty avg generation re (Low)
Jan 01, 2000 Jan 01, 2008 Jan 01, 2010 Jan 01, 2015 Jan 01, 2020
c)Capacity development
mw
— Capacity thermal (High)
— Capacity thermal (75% Percentile)
+ capacity thermal (Average)
— Capacity thermal (25% Percentile)
— Capacity thermal (Low)
— Capacity renewables (High)
Capacity renewables (75% Percentile)
Capacity renewables (Average)
— Capacity renewables (25% Percentile)
— Capacity renewables (Low)
30,000
20,000
10,000
Jan 01, 2000 Jan 01, 2008 Jan 01, 2010 Jan 01, 2015 Jan 01, 2020
to the value trading strategy.
From the preceeding discussion and in chapter 4, the results are as expected. When intro-
ducing TGCs in 2003, consumer prices increase (Figure 25a), and spot prices are low-
ered.
Figure 25b) shows how demand and generation under a TGC market with unlimited
banking. Surprisingly, the consumption remains fairly unaffected by the increasing costs
from TGC obligations, because electricity spot prices are suppressed by renewable gen-
eration. The relationship between the combined income of TGC markets and electricity
markets will oppose each other in cases of wet/windy years or dry/calm years. These bal-
ancing mechanisms also shown in the loop diagram in Figure 24 tend to stabilise varia-
tions in revenues of renewable suppliers. During windy years, the generation of wind will
be high, but electricity market prices decrease, as will the TGC price!. During calm years,
the number of TGCs issued decreases while prices on TGC’s rise, and electricity prices
rises as well.
Developers of renewable technology may experience periods of growth and stagnation in
the market for renewables, which is not desirable. A properly designed TGC market can
reduce the possibilities of price crashes that arise endogenously from the trading strategies
examined. Allowing borrowing (Figure 26) reduces the market power of suppliers in sit-
uations of capacity deficit. A higher and smoother development of renewables can be at-
tained by allowing borrowing of TGC’s.
These simulations are based on realistic marginal operational costs, long-run marginal
costs and price elasticity’s. Consumer prices did not reduce as a consequence of the TGC
market (see Jensen & Skytte (2002) and Bye et al. (2002) for a discussion). However, con-
1. Of course, some renewable suppliers will chose to store their certificates awaiting higher prices
on TGCs
sumer prices did not increase significantly either, but remain fairly unchanged
Figure 26 Banking, 50% borrowing and with trend strategy . Price and generation
shown as yearly averages
a)Consumer price, electricity spot price and TGC price
NoK/MWh
avg consumer price (High)
—avg consumer price (75% Percentile)
J+ ave consumer price (Average)
—avg consumer price (25% Percentile)
—avg consumer price (Low)
— Yearly avg price (High)
— Yearly avg price (75% Percentile)
J Yearly avg price (Average)
— Yearly avg price (25% Percentile)
— Yearly avg price (Low)
— Yearly avg TGC price (High)
— Yearly avg TGC price (75% Percentile)
| Yearly avg TGC price (Average)
— Yearly avg TGC price (25% Percentile)
—Yearly avg TGC price (Low)
°
Jan 01, 2000 Jan 01, 2005 Jan 01, 2010 Jan 01, 2015 Jan 01, 2020
b)Total generation/demand, thermal generation and renwable generation development (Demand coincide with total
generation)
Twhy/ye 1 1
400 —Yearty avg total generation (High)
— Yearty avg total generation (75% Percentile)
|+ Yearly avg total generation (Average)
— Yearty avg total generation (25% Percentile)
300 — Yearty avg total generation (Low)
— Yearty avg generation th (High)
— Yearty avg generation th (75% Percentile)
| vearty avg generation th (Average)
— Yearty avg generation th (25% Percentile)
— Yearty avg generation th (Low)
— Yearty avg generation re (High)
3 |— Yearly avg generation re (75% Percentile)
|-3 Vearty avg generation re (Average)
- — Yearty avg generation re (25% Percentile)
|— Yearty avg generation re (Low)
Jan 01, 2000 Jan 01, 2008 Jan 01, 2010 Jan 01, 2015 Jan 01, 2020
c)Capacity development
mw 1
— Capacity thermal (High)
— Capacity thermal (75% Percentile)
+ capacity thermal (Average)
— Capacity thermal (25% Percentile)
— Capacity thermal (Low)
— Capacity renewables (High)
— Capacity renewables (75% Percentile)
> capacity renewables (Average)
— Capacity renewables (25% Percentile)
— Capacity renewables (Low)
30,000
20,000
10,000
Jan 01, 2000 Jan 01, 2008 Jan 01, 2010 Jan 01, 2015 Jan 01, 2020
d)Capacity factor thermal units
We should however note that present elctricity market prices fluctuate significantly due
to daily load variations, seasonal variations in demand, and the stochastic properties of hy-
dropower generation in the Nord Pool. These considerations can be taken into account by
implementing the TGC market model in the more detailed Kraftsim model (see chapter
13 below)
13. Simulations of TGCs in the Kraftsim model
A more detailed system dynamic model of the Nordic electricity market has previously
been developed (Vogstad et al, 2002; Botterud et al. 2002). This model includes some
additional long-term feedback loops of , technology progress and resource availability.
Capacity acquisition includes a more detailed description of the application process and
the construction process, plus the vintage structure of capacity. The profitability assess-
ment includes a more detailed net present value calculation with feedbacks from technol-
ogy progress for the investment conts and feedback from expected capacity utilisation
concerning the operational costs and the expected profitability from sales of electricity.
Furthermore, the model distinguish between coal, nuclear, gas, gas peak load and gas with
CO2 sequestration; hydropower, wind power and bio energy plus imports/exports ex-
change. The supply side is still kept simple with an underyling growth of demand (1.6%
per year) and a price elasticity of demand with an adaptive reference price. Seasonal var-
iations in hydropower, wind energy and demand is included, and a simplified water value
method for hydropower scheduling is represented endogenously in the model.
By implementing the TGC market model in the previously developed Kraftsim model, it
is possible to assess the impact on various energy technologies and to which extent TGC
markets can be used as an instrument to transit from a fossil fuelled towards a renwable
power supply.
14. Summary of conclusions
In chapter 7 we developed a system dynamics model of the TGC market that pointed out
the possible problem of price formation from the lack of short-term regulation of supply.
Adjustments on the supply side of the TGC market can only be made in the long term by
investing in new capacity, which makes the dynamics of the market sluggish.
The main concern of price stability in previous studies have been the yearly variations of
renewables, which may cause large price variations from year to year. To circumvent this
problem, a TGC market with banking (i.e. unlimited lifetime of certificates) has been the
prefered solution. However, such an arrangement opts for strategic behavior that can in-
duce much larger long-term price variations. If traders use price trends in their trategies,
the reinforcing effect causes prices to crash when sellers withold their TGC’s over several
years before new capacity comes on line. Allowing borrowing of certificates will reduce
the impact of this strategy, as buyers can postpone their obligations, and developers can
sell TGC’s that will be produced in future years.
Partial equilibrium models and standard economics presently used to analyse TGC mar-
kets do not address these potential problems concerning price stability and trading strate-
gies. A combination of system dynamic analysis and experimental economics can analyse
the impact of various such trading strategies in order to avoid costly mistakes.
In chapter 13, we simulated the TGC market fully integrated with the electricity spot mar-
ket. The balancing feedback loops between these markets seem to reduce the variations
in investments of renewables that was observed in TGC market model. Consumer prices
were not significantly alterend after the introduction of the TGC market.
15 _— Bibliography
A. Gaunersdorfer 2001. Adaptive Beliefs and the Volatility of Asset Prices. Central Eu-
ropean Journal of Operations Research 9, 5-30
Botterud, A., M. Korpas, K. Vogstad and I. Wangensteen. 2002. A Dynamic Simulation
model for Long-term Analysis of the Power Market. Proceedings, Power Systems
Computation Conference, 25th-28th June, Sevilla, Spain.
Bye, T., O.J Olsen and K. Skytte 2002. Gronne sertifikater - design og funksjon. Report
no. 11/2002, Statistics Norway. (in Norwegian)
Econ 1999. Dokumentasjon av ECON’s kraftmarkedsmodell. Econ note 4/99, Econ cen-
tre of economic analysis.
Farmer, J. D. and S. Joshi 2000. Price Dynamics of Common Trading Strategies. Santa
Fe Institute Working Paper 00-12-069, to appear in the Journal of Economic Behav-
ior and Organization.
Ford, D.N, 1999. A behavioural approach to feedback loop dominance analysis. System
Dynamics Review 15:3-36
Forrester, J.W. 1961: Industrial Dynamics. MIT Press. Cambridge MA
Groenheit, P.E., H.V. Larsen 2001. Balmorel - Data and Calibration. Balmorel Project
Report March 2001. Available [online] http://www.balmorel.com.
Hindsberger, M., M.H Nybroe, H.F. Ravn and R. Schmidt 2003. Co-existence of Elec-
tricity, TEP, and TGC Markets in the Baltic Sea Region. Energy Policy, Vol 31(1).
Jensen, S.G, K. Skytte 2003. Simultaneous attainment of energy goals by use of green
certificates and emission permits. Energy Policy Vol 31(1)
Jensen, 8.G, K. Skytte, 2002. Interactions between the power and green certificate mar-
ket. Energy Policy Vol 30(5).
Krohn, S (2001). An Industry View of a Proposed Danish Green Certificate Market. The
Devil is in the Detail. Copenhagen Workshop on Tradables, 10-11 September 2001.
Available [online]: http://www.windpower.org/articles/busiview.htm
Langniss, O. and R. Wiser 2003. The renewables portifolio standard in Texas: an early
assessment. Energy Policy 31 p527-535.
Morthorst, P.E., 2000: “The development of a green certificate market” Energy Policy vol
28, pp 1085-1094, 2000.
Nielsen, L., and T. Jeppesen 2003. Tradable Green Certificates in selected European
countries- overview and assessment. Energy Policy 31(1).
NOU 1998. Energi og kraftbalansen frem mot 2020. Whitepaper, Norwegian Ministry
of Petroleum and Energy. (Norwegian)
REBUS 2001. Renewable energy burden sharing: Effects of burden sharing and certifi-
cate trade on the renewable electricity market in Europe. EU Report, May 2001.
RECeRT 2001. The European Renewable Certificate Trading Project Final Technical Re-
port. Available [online] http://recert.energyprojects.net/
Riso 2002. Green Certificates and Emission Trading. Riso report no. Risg-I-1936(EN)
Schaeffer G.J & J. Sonnemans 2000. The influence of banking and borrowing under dif-
ferent penalty regimes in tradable green certificate markets - results from an exper-
mimental economics laboratory experiment. Energy and Environment Vol 11(4).
SOU 2001: “Handel med elcertifikat - ett nytt satt att framja el fran fornybara energikal-
lor” Svenska offentliga utredning. SOU 77:2001 (In Swedish)
STEM 2002. Yttrande éver Elcertifikatutredningens betinkande, “Handel med elcerti-
fikat”, SOU 2001:77 Statens energimyndigheter, yttrande. pp5-7 (In Swedish).
Available [online] http://www.stem.se
Sterman, J.D., 2000: “Business Dynamics: Systems Thinking and Modeling for a Com-
plex World”, McGraw-Hill. Book website: www.mhhe.com/sterman
Vogstad, K., A. Botterud, K.M Maribu and S.G. Jensen. 2002. The transition from a fossil
fuelled towards a renewable power supply in a deregulated electricity market. Pro-
ceedings, System Dynamics Conference, 28th-1st Aug, Palermo, Italy.
Vogstad, K., H. Holttinen, A. Botterud and J.O. Tande 2000. System benefits of coordi-
nating wind power and hydro power in a deregulated market. Proceedings "Wind
Power for the 21st Century, 23-25 Sept. 2000, Kassel Germany.
16 = Note
The TGC market Powersim Studio .sip file is available from https://bsew.ntnu.no/pub/
bsew.cgi/d229462/TGCmarket.sip
The Kraftsim model Powersim Studio .sip is available from https://bscw.ntnu.no/pub/
bsew.cgi/d229462/Kraftsim1 |.sip
Back to the Top
CC BY-NC-SA 4.0