Table of Contents
Effects of Regulation on the Dynamics of Liberalised Power Sectors:
A Cost Benefit Analysis of the Capacity Payment in Hydro Based Systems
Gabriela Elizondo-Azuela
The World Bank and Imperial College
Dr. Matthew Leach
Imperial College
Dr. Abhijit Mandal
London Business School and Warwick Business School
Key Words: liberalisation, privatisation, regulation, power sector, capacity payment
Corresponding Author:
Gabriela Elizondo-Azuela
Finance, Private Sector and Infrastructure
Latin America and the Caribbean Region
The World Bank
1818 H Street NW
Washington DC 20433, USA
gazuela@wordlbank.org
Please do not copy or quote without the permission of the authors
ABSTRACT
In this paper, a cost benefit analysis associated to the application of the capacity payment
in the Colombian power system is reported.
The analysis results in two relevant conclusions. The first one is that the capacity
payment as it is designed today will not maintain acceptable levels of reliability in the
long run as it does not succeed in restoring private investments in the short to medium
terms. The second conclusion is that an increase in the value of the capacity payment
today, which succeeds in effectively attracting the required investments in capacity to
maintain minimum levels of reliability, has higher net benefits when long terms effects
are taken into consideration
I INTRODUCTION
Over the last two decades, the power sectors of many Latin American nations have been
privatised and subsequently liberalised with various degrees of success from both the
technical and economic points of view. The extent to which the reforms have succeeded
across the region is however still under analysis.
Without a doubt, governments have benefited from privatisations and fiscal burden relief.
Various analysts have in fact demonstrated the success of the reform with measurable
results in terms of lower electricity prices, lower transmission and distribution losses and
improved technical and economic efficiencies of privatised enterprises, among others.
However there are still doubts regarding the sustainability of these improvements,
specially with the fall of private investment flows after 1997. Indeed, there is widespread
concern that increased uncertainties and pool prices do not provide the long term signals
required to attract private investment in infrastructure projects. Conversely, private firms
are neither committed to the maintenance of minimum levels of security of supply nor
with the need to expand the service to poor or isolated areas.
For countries mainly based on hydroelectric generation the addition of firm capacity to
maintain minimum levels of reserve margin during dry seasons as well as to lower price
volatility have become issues of major concern (e.g. Brazil, Colombia).
As the liberalisation of power systems progressed across the region, the difficulties in
designing and applying regulatory mechanisms were exposed. The performance of
regulatory commissions has been poor in cases where regulators do not have experience
in dealing with the complexity associated to both, the economics of regulation and
company strategic behaviour. One of the most relevant weaknesses associated to the
performance of regulatory commissions is the fact that the design of mechanisms is rarely
the result of an analysis that considers the likely long term effects associated their
application. As regulators understood the technical and economic effects of applying
specific instruments and rules, a variety of adjustments and sudden modifications were
carried out, increasing the overall perception of risk. But even now, after two decades of
learning in the region, regulators do not analyse the long term effects of their decisions in
terms of costs and benefits.
In this paper, a cost benefit analysis associated to the application of the capacity payment
in the Colombian power system is reported.
The analysis throws two relevant conclusions. The first one is that the capacity payment
as it is designed today will not maintain acceptable levels of reliability in the long run as
it does not succeed in restoring private investments in the short to medium terms. The
second conclusion is that an increase in the value of the capacity payment, which
succeeds in effectively attracting the required investments in capacity to maintain
minimum levels of reliability, has higher net benefits when long terms effects are taken
into consideration. In other words, contrary to the expectations, the analysis demonstrates
that higher capacity payments have the potential to result in higher net benefits in the
long run.
II THE BASELINE
The model seeks to represent the system structurally and behaviourally considering the
key variables driving the system. Annex I shows the causal loop diagrams characteristic
of the system. A simplified diagram is depicted below in Figure I.
Figure I A Simplified Causal Loop Diagram
Electricity
Price
+ Profits
‘ 1 3
Electricity Demand = __
aft
q Capacity Addition
Reserve Margin oi gees
+
2 —
Desired Reserve Gap in Reserve
- Margin
Margin
Se
In the model, the investment behaviour of three types of generating firms is considered:
public utilities, multinational utilities, and independent power producers.
The baseline simulation shows that with a conservative electricity demand growth rate
and given the behaviour and constraints imposed on the participating private and public
firms the reserve margin of the system will lower year by year leaving the system highly
vulnerable to critical seasonality changes (see Figures II and III below). By 2015 the
system cannot ensure minimum levels of reliability leading to a rationing event that lasts
about 10 months during an intense drought or ENSO event (i.e. El Nifio Southern
Oscillation) starting December 2015. As seen, no new thermal plants are built in the
period 2000-2017. Only after the rationing crises, the participating agents have an
incentive to build new greenfield capacity.
Figure H Evolution of Supply and Demand (GWh/month)
9,000}
4
1
6
. 5
4
6,000} ON ws
a _,—Available_Hydro_Capacity_GWh_Month
—y~Avallable_Wind_GWh_Month
6
- ~3~Available_Thermal_GWh_Month
eo Pe 4 ~4-Total_Available_Capacity_GWh_Month
Ls —— © a
sao VL 1 —g~Peak Electricity Demand
3
es g Peak_D_RM_20
Various factors explain the reason why both private and public firms do not invest in
greenfield or new thermal based capacity in the period 2000-2016. In this system,
participant generators are constrained by different types of regulations as well as by their
particular financing capacities and strategies
' Rather, some old coal and fuel oil based power plants are retired over the period of the simulation.
Figure III Evolution of Operative Reserve Margin
0.8+
gin_Available
°
bi
2
=
Reserve_Mar,
°
»
0.04
2,000 2,005 2,010 2,015 2,020
Time
The evolution of contract prices as estimated by the model is shown in Figure IV below.
Figure IV Evolution of Contracts Price (USD/kWh)
0.6,
0.54
0.34
S_Contracts_Price
0.24
2000 2005 2010 2015 2020
In the contracts market, the volatility characteristic of spot electricity prices is minimised.
Essentially, the prices negotiated in bilateral contracts function as a medium term price
signal which partially replaces the long run marginal cost generally calculated under
centralised or non-liberalised power markets to plan future investments in capacity
additions. Still, bilateral contracts cover periods of only 2-3 years, for which the contracts
price does not necessarily provides with a signal for investment in the long term.
In the model, the contracts price is a function of the spot price (e.g. simulating a typical
contract for differences). Under the assumptions and conditions established in this
scenario, independent power producers (IPPs) and multinational utilities (MNUs) do not
invest in new capacity mainly due to the following aspects of the market:
¢ Low wholesale electricity prices in the system which result in unsustainable
project debt service coverage ratios” (i.e. lower than one) as well as in lower than
expected returns to investment.
¢ High cost of capital and short maturity periods, which result in an
unsustainable coverage of the debt service.
¢ Low rationing cost which was determined through a contingent valuation
analysis conducted at the beginning of the 1990s (i.e. its value is about one tenth
of the British Value of Lost Load (VOLL), Concha 2002, Benavidez 2002)°. This
affects the electricity price (e.g. the pricing curve associated to the cost of water
which is a function of the reservoir volume, is also a function of the rationing
cost).
¢ Low load factors for thermal based generation given the high share of
hydroelectric capacity in the system which also affect the expected profitability of
gas or even coal based facilities. In fact, thermal based capacity is seldom
dispatched in Colombia‘.
? For the particular case of independent power producers (IPPs), these are calculated considering the
conditions of the Colombian capital market, which imposes short maturity periods and high interest rates.
> The Energy Planning Unit (UPME) only conducts a monthly adjustment to inflation of the cost of
rationing (Concha 2002).
* Average utilization capacities for gas and coal based power plants in Colombia are between 20 and 50%
according to the National Dispatch Commission (CND). The baseline simulation confirms this range.
In fact, in Colombia the majority of thermal generators operating in the market have
signed long term contractual agreements that not only hedge the price, but ensure the
allocation of higher percentages of their output, either through real operation and dispatch
of their plants or through commercialisation (i.e. trading or buying the committed output
in the spot market when the pool price is lower than their variable costs).
The model has been designed assuming that public utilities (PUs) seek to balance
conservative levels of profitability with the intention of maintaining minimum required
levels of operating reserve. However, state owned and municipal PUs are limited not only
by the maximum market share imposed by the regulation (i.e. 25% as established in
CREG Resolution 128 of 1996) and financing constraints®, but also by their own
minimum demands of investment return.
Firm Capacity Payments
In Colombia, according to CREG Resolution 116 of 1996, a capacity payment is
distributed among generating plants that provide with firm capacity given the threat of
loss of load -or rationing- during critical hydrologic conditions. The capacity payment or
capacity charge (CxC) in Colombia intends to provide with both a long-term price signal
and a compensation allocation mechanism. The CxC is in fact an additional source of
income for plants that are needed as available to the system but that are infrequently
dispatched (i.e. peak load). The basic idea of the capacity payment is that, when there are
periods of excess capacity and the reserve margin is high, the probability of loss of load
is relatively low. Under these circumstances there is little incentive to invest.
Alternatively, when there is heavy demand relative to available capacity, the reserve
margin is low and the probability of loss of load increases which triggers investments in
capacity. In Colombia the CxC is also seen as a price floor in the spot market, as it is
‘Three large public utilities are considered in the model. It is assumed that EEPPM, the largest municipal
utility, is able to finance capacity additions to the extent that the company has no more than 21% of the
market share (Navarro, 2002). Although ISAGEN and CORELCA, State owned utilities, are in the process
of being sold (so far unsuccessfully) the baseline scenario assumes that these two could have the same
financing capacities as EEPPM (i.e. a rather optimistic assumption).
provided to generators even when the opportunity cost of stored water is zero during wet
seasons’ or economic structure of the merit order.
The CxC is estimated considering the fixed payment associated to the avoided capital
cost of the next cheapest generation addition’, the peak demand and the total amount of
available firm capacity. According to CREG Resolution 116, only 105% of peak demand
is paid as firm capacity. Based on this, the value of the CxC increases if -as the demand
raises- the available firm capacity does not increase. Conversely, if there is enough firm
capacity, the distribution of the total amount of resources is allocated among more
generating plants lowering the CxC. The following Figure provides with the baseline
estimation of the evolution of the value associated to the CxC.
Figure V Evolution of the Value of the Capacity Charge
0.0124
0.0104
)_per_kwh
' 0.008}
CxC_USD.
0.006+
1,995 2,000 2,005 2,010 2,015 2,020
Time
The graph in Figure VI on the other hand plots the value of the CxC as a function of the
reserve margin showing the ranges under which this measure has been designed.
° The capacity payment is paid by the consumers through the electricity tariff. The regulator then collects
this portion of the price and distributes it among generators with firm capacity.
7 As of today an OCGT plant whose capital investment is discounted at 11% to produce a fixed payment of
5.25 USD/kW-month.
Figure VI The Value of the CxC as a Function of the Reserve Margin
rgin_Available
Reserve_Mai
0.007 0.008 0.009 0.010 0.011 0.012
CxC_USD_per_kwh
The CxC, as it designed today, has not provided with an effective incentive to trigger
capacity additions. The following Figure shows an aspect of the Colombian electricity
market that is becoming common in the liberalised power system of developing
countries, specially in those with high shares of hydroelectric capacity. Investments in
capacity do not respond to sustained increases in peak demand, only after the price
reaches a high level, and investments are triggered, available capacity increases.
10
Figure VII Peak Demand vs Total Available Capacity Period 2003-2020
|
7,000+ |
Period 2003-2015 |
|
z
<
£ |
€ 6,000}
o) |
a |
>! I Period 2016-2020
3 |
G |
2 5,0004 |
a |
x! I
g |
2 |
a
|
4,000} |
5,000 6,000 7,000 8,000 9,000
Total_Available_Capacity_GWh_Month
Il EFFECTS OF OWNERSHIP STRUCTURE
In this section different scenarios of ownership share have been conducted to assess the
effects of the behaviour of different types of firms on reserve margin’s sustainability. The
intention is to demonstrate the behaviour of both private and public enterprises as
hypothesised in the model. In Section IV a cost benefit analysis of different values of
capacity payment are assessed considering a mixed ownership (as in the baseline).
3.1 Investment Under Public Ownership
Scenario A Pre-liberalisation. This scenario simulates an ideal expansion plan in which
capacity additions are fully financed by a centralised public utility (i.e. the government)’.
Results are depicted below in Figures VIII to XI and quantitatively reported in Annex II.
* The simulation mimics the least cost planning exercise which is performed with the Super Olade Bids
model under centralised non liberalised power systems.
11
Figure VII Evolution of Supply and Demand (GWh/month)
8,007 6
a soe
—1—Avallable_Hydro_Capacity_GWh_Month
yin pee ~~ Available_Wind_GWh_Month
8 ee =3—Available_Thermal_GWh_Month
4,000$ 4 Total_Available_Capacity_GWh_Month
eo / IF ee coe —g~ Peak Electricity Demand
g Peak_D_RM_20
2,000+ fpf
0. 2 + ? 4
2000 2005 2010 2015 2020
Time
Figure IX Evolution of the Value of Capacity Payment
<
0.0124
z
cay
5
a
g 1
2
I Mo. .
2 0.0094 4 V2
Oo 2
a
0.006 + + + + ‘
1995 2000 2005 2010 2015 2020
Time
Note: Line | is the baseline, Line 2 is Scenario A
12
Figure X Evolution of Thermal Based Generation (GWh/month)
6,000
=
5 =—
= 5,000}
=
=
9,
© 4.0004 H
o
2
im
2
® 3,000]
$ fT ’ 1 4
<
2,000 : ; . t
2000 2005 2010 2016 2020
Note: Line | is the baseline, Line 2 is Scenario A
Figure XI Cumulative Reserve Margin
gin
\
\
\
bai 2
Cummulative_Reserve_Mar.
\
2000 2005 2010 2015 2020
Time
Note: Line | is the baseline, Line 2 is Scenario A
13
As seen, under this scheme, the rationing event is completely avoided, electricity prices
are lower, and although the government has to invest in the installation of about 2,184
MW of thermal based capacity (in addition to the 660 MW hydro plant considered in the
baseline), ultimately the net benefits reach 2,934 USD Million over the period 2000-2020
(i.e. mainly due to elimination of rationing and associated costs).
Indeed, Scenario A would require resources from the federal budget in the amount of
2.12 billion USD over the 20 year period, which would reduce the availability of
budgetary resources for other more pressing priorities (e.g. education and health).
Discouraging the participation of the private sector, on the other hand, would only
diminish the overall amount of resources available for infrastructure development and
lower the efficiency of national resources allocation.
The cumulative reserve margin’ is lower in scenario A than in the baseline (see Figure
XI). This is explained by the investment behaviour exhibited by the various firms. In the
baseline simulation, investments by independent power producer (IPPs) in the years after
the liberalisation responded to their expectations regarding the evolvement of a
competitive profitable market rather than to high prices. After this transitional period
(1995-2000), private firms —in the model- do not invest until electricity prices are high
enough to obtain minimum returns to investment. In Scenario A, a minimum reserve
margin (20%) is always maintained. Figures VIII and XI show how under a least cost
ideal expansion plan, the timing of investments to maintain a minimum reserve margin
provides with a more efficient system in terms of reliability of supply management (i.e.
which is the rationale the behind ideal expansion plan).
The evolution of carbon emissions under Scenario A is closer to the one calculated by
UPME with the Super Olade Bids model to produce an ideal expansion plan.
° This index measures the degree to which the evolution of investments contribute to the maintenance of the
reserve margin.
14
3.2 Investment Under Private Ownership
Scenario B. This test simulates an scenario in which public utilities have no resources to
invest in capacity additions and the sustainability of the system is only dependent on
private initiatives. The intention is to investigate whether a private ownership structure
alone (as opposed to a mixed ownership structure) would ensure the long term
sustainability of the system.
As shown quantitatively in Annex II and below in Figures XII to XVI, this scenario does
not result in positive net benefits. Not only the rationing crises is worse than the one
exhibited in the baseline simulation, but the electricity prices are also higher under this
scenario. The reason behind this outcome relies on the behaviour of private firms. In the
model, the investment of private firms respond to high electricity prices and for this
reason the maintenance of tight reserve margins work out in their favour. Indeed, the
sustainability of a minimum reserve margin to protect the system from rationing events
does not form part of the strategic behaviour of private firms.
Figures XII to XVI illustrate the nature of private firm’s investment in liberalised
systems. Only after some years and when the reserve margin is close to the peak demand
(2010-2014), independent power producers (IPPs) and multinational utilities (MNUs)
invest in thermal based generation. Without the participation of public firms, the
operative reserve margin decreases sooner than in the baseline, triggering earlier private
investments in thermal capacity (see Figure XIV). The reserve margin however is never
above minimum required levels (e.g. 20%). For this reason the system is vulnerable to the
ENSO event of 2016 and the rationing crises reaches a deficit of 3,077 GWh.
15
IW_Real
pacity_M!
Total_Available_Ca,
6,000-
Figure XI Evolution of Supply and Demand (GWh/month)
6 5
“4,
—y~Available_Wind_GWh_Month
ys 6
5
aa a Wy —,—Available_Hydro_Capacity_GWh_Month
6 oe
es
¥ he pies ‘
aI
3,000} 3
2 a “rt
—3~ Available_Thermal_GWh_Month
4 Total_Available_Capacity_GWh_Month
—g— Peak Electricity Demand
sa ae
Peak_D_RM_20
3
t t t 2
2000 2005 2010 2015 2020
Time
Figure XII Evolution of Available Capacity (MWs)
11,000,
ite
10,000+
9,000+ x
, 4 Mi
4
8,000} e
2
7,000+
6,000+
2000 2005 2010 2015 2020
Time
Note: Line | is the baseline, Line 2 is Scenario B
Figure XIV Evolution of Thermal Based Generation (GWh/month)
5,0007
4,500+
4,000+
3,5004
3,0007
Available_Thermal_GWh_Month
2000 2005 2010 2015 2020
Note: Line | is the baseline, Line 2 is Scenario B
Figure XV Cumulative Reserve Margin
ae
8 pee
<
5 1
i
= 4
o
ce =
a we
o
g,
= a
3 2
€
=
3
is)
24
2
2000 2005 2010 2015 2020
Time
Note: Line | is the baseline, Line 2 is Scenario B
17
Figure XVI Rationing Crises 2015-2017
7,500+
Month
7,000+
Pal
snot
pacity_GWh.
g
|
® =
2 \Y 2
3 2 pa
= ee
g
5,500]
3
ko}
-
5,000 ' t + : 1
2015 2016 2016 2017 2017 2018
Note: Line | is the baseline, Line 2 is Scenario B
IV COST BENEFIT ANALYSIS OF THE CAPACITY PAYMENT
In Scenarios C to E, changes in the design of the capacity payment will be tested to assess
the effects of higher payments on system’s reliability and associated costs and benefits.
Indeed, one would expect that an increase in the value of the capacity charge would only
result in an increase to the total costs of supplying electricity or the use of more public
resources. After all the capacity charge would increase the electricity price in order to
promote a higher reserve margin.
What would then be the costs and benefits associated to an increase in the value of the
capacity payment? Or would an increase in the value of the capacity payment prevent the
gradual lowering of the reserve margin and ensure the sustainability of the required
optimal reserve margin?
The following changes to the design of the capacity payment will be tested in three
different scenarios:
18
Scenario C: An increase in the value of the capacity charge through an increase in the
discount rate used to calculate the monthly payment associated to the same technology
(i.e. an OCGT)
Scenario D: An increase in the reserve margin policy
Scenario E: A combination of the previous. The idea is to find out whether these two
measures are additive, synergic or neutralised among each other.
Scenario C. This scenario considers an increase in the discount rate considered in CREG
Resolution 116 to a apply a more realistic value that reflects the country risks. To find an
“optimum”, a test has been performed with a range of values that go from 0 to 16%. The
quantitative results are provided in Annex III and depicted in Figures XVII to XIX.
The test shows that the net benefit to the system is maximised when applying a capacity
payment estimated with a discount rate of 14%. The maximisation of the net benefit
stems from two important effects on the dynamic nature of the system: a) avoided
rationing (see Figure XVII) and b) lower electricity prices due to earlier investments in
greenfield capacity (see Figure XVII)
Figure XVII Total Available Capacity During Rationing Event
7,5004
7,000} ®
—4-Total_Available_Capacity_GWh_Month
Total_Available_Capacity_GWh_Month
6,500} -9-
—3— Total_Available_Capacity_GWh_Month
=4- Total_Available_Capacity_GWh_Month
—g~ Total_Available_Capacity_GWh_Month
6,0004
_g~ Total_Available_Capacity_GWh_Month
5,500} _.Peak_Electricity Demand
“T
5,000- + + + + 1
2,015.0 2,015.5 2,016.0 2,016.5 2,017.0 2,017.5
Time
Note: Line | is r=0, line 2 is r=11, line 3 is r=12, line 4 is 13, line 5 is r=14, line 6 is r=15
19
Figure XVII Evolution of Thermal Based Generation Period 2010-2020
5,000,
a
a
3
8
hi
°
=)
3
5
| 3,000} a8
eS a
2,500+ ql
Available_Thermal_GWh_Month
¥
8
3
2,000-
2,010 2,015 2,020
Note: Line | is r=0, line 2 is r=11, line 3 is r=12, line 4 is r=13, line 5 is r=14, line 6 is r=15
Figure XIX Baseline and Scenario C
0.0144
<= 0.0124
& 2 2
5 2
a
1 0.0104
CxC_USD.
G
0.0087
0.006+
1,995 2,000 2,005 2,010 2,015 2,020
Note: Line | is the baseline, Line 2 is scenario r=14
20
This increase in the value of the capacity charge raises the annual payment by the
regulator from about 527.5 to 646 million USD in 2003, or in the order 20-25% every
year. This measure would indeed increase the reserve margin to minimum required levels
of reliability (i.e. 20%) after 2013 and even, avoid the rationing after 2015 that results
from the rain pattern scenario used in the baseline and in this scenario (see Figure 6.17
below)"”.
Figure XXI compares the baseline scenario and Scenario A (discount rate at 14%) in
terms of total available capacity.
Figure XX Evolution of Capacity and Reserve Margin, Scenario C
=
—4—Available_Hydro_Capacity_GWh_Month
4
LM "6 5
«000 IN ag vw
6
Pe _Available_Wind_GWh_Month
6 =
— Available_Thermal_GWh_Month
. 3
Le a 1 {<q Total_Available_Capacity_GWh_Month
3,000. AL ways wg Peak_Electricity Demand
3
g Peak_D_RM_20
0. 2 ———
2000 2005 2010 2015 2020
Time
'° Recall that the rain scenario was chosen to follow the same pattern as in the two decades 1980-1990 and
1991-2000.
21
Figure XXI Evolution of Capacity (Baseline and Scenario A)
12,000}
3 2
a
bal
= 11,000]
=
2
& 10,0004
o 2,
&
bat
o 1
9,000] 42
s
: ;
| 8.0004 ¥
3 2
to}
‘sy
7,000. ' : ' 1
2,000 2,005 2,010 2,015 2,020
Note: Line | is the baseline, Line 2 is Scenario C
Most important is however to find out what is the total balance of costs to the system and
the evolution of investments to analyse in detail the effects of this change to the value of
the capacity payment.
Increasing the discount rate used to estimate the capacity payment has in fact three
notorious consequences:
1. Investments are anticipated by about three years, as depicted in Figure XVIII
2. Rationing is avoided (see difference in Figures XX and XXI)
3. Thanks to earlier investments, total costs to the system are ultimately 2,860 USD
million lower if avoided costs of rationing are considered (see Annex III).
System dynamics models are concerned with long term behavioural patterns and the
dynamic tendencies of complex systems. Indeed, precise quantitative estimates are not
the focus of the analysis. This exercise demonstrates a counter-intuitive result. While an
increase in the discount rate of the capacity payment signify millions of dollars in extra
annual costs to the consumer, ultimately, early investments in capacity avoid price spikes
22
as well as the high costs associated to the rationing event, resulting ultimately in
important savings to the system. This benefit however can only be estimated when.
considering long term patterns. Indeed, the provision of a higher capacity payments has
the potential to avoid rationing in the long term. In effect, this instrument allows the early
investment of most cost effective efficient technology, lowering not only the total cost to
the system in terms of electricity price (i.e. the diffusion of more efficient capacity lowers
the electricity price in the wholesale market as this capacity displaces inefficient most
costly technology), but the rationing threat. The idea that an increase in capacity charge
(CxC) would only result in higher electricity prices has been falsified. Indeed, allowing
an increase in revenues results in earlier investments in efficient capacity which results in
high positive net benefits to the system when considering the avoided cost of rationing.
Scenario D This scenario considers an increase in the amount of firm capacity paid
considered in CREG resolution 116 established at 5% above the peak demand through
changes in the reserve margin policy. Different tests have been carried out to find the
optimum value. Results are shown in Annex IV and in Figures XXII and XVI below.
23
Figure XXII Total Available Capacity During Rationing Event
7,500
7,000+
—1—Total_Available_Capacity_GWh_Month
8.500) =y~Total_Available_Capacity_GWh_Month
=3- Total_Available_Capacity_GWh_Month
6,000+ —=4— Total_Available_Capacity_GWh_Month
=g- Total_Available_Capacity_GWh_Month
— Peak Electricity Demand
5,500
5,000 + + + + 1
2015 2016 2016 2017 2017 2018
Time
Note: Line 1 is the baseline, lines 2, 3, 4, 5 correspond to CxC provided to 10, 15, 20, 30% of peak
demand.
According to the results, the net benefits are maximised when the regulator pays 130% of
peak demand as firm capacity to generators that provide with this service. This is indeed
an interesting result since a reserve margin of 30% has been always considered an
optimum almost as a rule of thumb. The results of the test confirm this hypothesis
showing the same effects than in Scenario C, the increase in the cost of the CxC is
covered by a decrease in the total costs of electricity to the system and most importantly
to the elimination of rationing. Again, the reason being investments triggered four years
in advance (see Figure XXIII).
24
Figure XXIII Evolution of Thermal Based Generation Period 2010-2020
£ 5,000}
<
S
=
<
3
=! 4,000
o 1
E
2
r
© -
2 3,000 a
3 23:45)
g 123
4
2,000: t ,
2010 2015 2020
Time
Note: Line 1 is the baseline, lines 2, 3, 4, 5 correspond to CxC provided to 10, 15, 20, 30% of peak
demand.
Figure XXIV Capacity Payment (Baseline and Scenario B)
$ oot
al 2 2
3 2
a
g Boul ™
1
2 0.0004 1 1
5 4 2
al
0.006. ' ' + 1 t
1995 2000 2005 2010 2015 2020
Time
Note: Line 1 is the baseline, Line 2 is Scenario B with 130% peak demand
25
The comparison between Scenarios C and D is depicted in Figures XXV and XXVI
below. Effectively the two measures result in almost equal effect in terms of net benefits.
Figure XXV Capacity Payment (Scenarios C and D)
r_kwh
O12 Prada ! /
0.009}
_pe'
CxC_USD.
0.006- + t t t t
1995, 2000 2005 2010 2015 2020
Time
Figure XXVI Evolution of Capacity (Scenarios C and D)
12,0005 1 |
Kz
11,0004
IW_Real
10,000}
9,000+ Ann’
yay
8,0004 \
2
7,000+
pacity_M'
ra
Total_Available_Ca
6,000-
2000 2005 2010 2015 2020
Time
26
Scenario E This scenario combines the two modifications carried out in tests C and D to
test weather the two policies are additive or synergic (positively or negatively) or on the
contrary neutralised. Results comparing tests C, D and E are provided in Figures XX VII
and XXIX. As shown, the two policies together are not additive neither in benefits nor in
costs in terms of evolution of reserve margin. Marginally, it does however increases the
additions of wind and gas based capacity and lowers investments in coal based capacity,
with the consequent lowering in carbon emissions.
Figure XVII Value of Capacity Payment (Scenarios C, D and E)
0.015} Nal 4
<
3
S
a 3
a
al h, 6, P cal
Q f 4 1
3, 4 4
3
2 oot04
5
ta_ 0
1995 2000 2005 2010 2015
Time
2020
Note: Line | is Scenario C, Line 2 is Scenario D and Line 3 is Scenario E.
The total net benefits for the system associated to Scenario E are 0.32% lower than those
of Scenarios C and D. Ultimately, in terms of benefits, the three scenarios are similar
despite the difference in capacity payment between Scenarios C-D and Scenario C.
27
» Capacity_MW_Real
Total_Available_Caj
Figure XXVIII Evolution of Total Available Capacity (Scenarios C, D and E)
12,000 1
d
3.2
11,000+
10,000+
1
‘a
3
7 ay,
\e
8,0004 LV Vy
wa
7,000- + + + +
2000 2005 2010 2015 2020
Time
Note: Line | is Scenario B, Line 2 is Scenario C and Line 3 is Scenario D.
Figure XXIX Evolution of Thermal Available Capacity (Scenarios C, D and E)
5,000,
s lf
= 4,500] J
°°
2,
= 4,0004 =
=
9 41
£ 3,500
o
FS
2
| 3,0004 acl 3
a 23
s
3B
$ 2,500]
¢
2,000 1 1
2010 2018 2020
Time
Note: Line 1 is Scenario A, Line 2 is Scenario B and Line 3 is Scenario C.
28
Considering both Scenarios C and D it can be established that the measure does not have
an additive result except for the additionality in terms of emission reductions and the
support to renewable energy. From the perspective of the Clean Development Mechanism
(CDM, a flexible mechanism under the Kyoto Protocol), a regulatory change such as the
one shown in Scenario C would prove additional and has the potential to reduce more
carbon emissions than the installation of a 80 MW run of river plant for a period of 21
years (see World Bank 2003).
V CONCLUSIONS
The following conclusions can be derived from the scenarios considered and reported
above.
The status quo will not keep acceptable levels of reliability in the long term
The results of the baseline simulation suggest that despite of the overcapacity exhibited in
the Colombian ESI today —with a 66% share of hydroelectric capacity- the reserve
margin of the system can lower gradually leaving the system highly vulnerable to
seasonality changes and ENSO events after 2010. Investments in capacity do not respond
to sustained increases in peak demand until after the price reaches a level required to
reach expected returns to investment. Indeed, investments are triggered by price spikes
which lead to waves of boom and bust in the construction of plants.
Under the assumptions and conditions established in this scenario private firms do not
invest in greenfield facilities before 2015, mainly due to low wholesale electricity prices
and low load factors associated to thermal based capacity. In addition, the financing
constraints imposed by commercial banks (i.e. high costs of capital, low maturity
periods) contribute to the lack of private investment in the sector.
The capacity charge today does not succeeds in restoring private investment flows
The capacity payment provided as designed by the energy regulatory body in CREG
Resolution 116 is not sufficient for two reasons: a) it is calculated with a discount rate of
11% while the minimum return to investment sought by a private investor is 15% due to
29
the risk premium demanded for the particular case of Colombia and b) it only pays 105%
of the demand in firm capacity, whilst the system needs at least 20% of reserve margin.
Regardless of the allocative inefficiency (which related to political economy issues)
associated to this instrument in the Colombian setting, the resources available to ensure
the availability of the necessary amount of firm capacity to avoid rationing events, is
insufficient.
Higher capacity payments have the potential to result in higher net benefits
While an increase in the discount rate of the capacity payment signify millions of dollars
in extra annual costs to the consumers, ultimately, early investments in capacity avoid the
high costs associated to rationing events, resulting ultimately in important savings to the
system. This benefit however can only be estimated when considering long term patterns.
Indeed, the provision of a higher capacity payment has the potential to attract on a
sustainable basis the financing needed over time to expand services to future consumers
and avoid rationing in the long term. In effect, this instrument allows the early investment
of most cost effective efficient technology, lowering not only the total cost to the system
in terms of electricity price (i.e. the diffusion of more efficient capacity lowers the
electricity price in the wholesale market as this capacity displaces inefficient most costly
technology), but the rationing threat.
The idea that an increase in capacity charge (CxC) would only result in higher electricity
prices has been falsified. Indeed, allowing an increase in revenues results in earlier
investments in efficient capacity which results in high positive net benefits to the system
It has been therefore being demonstrated that a simple regulatory measure has the
potential to solve a problem. Indeed, it has been extensively recommended that the
regulators of Latin American countries apply simple transparent regulations as opposed
to complex configurations that have not been tested or fully explored in other more
developed systems (e.g. auctioning options and futures).
The structure of the system in terms of ownership matters
A mixed ownership will deliver a more sustainable system when the appropriate
incentives to trigger investment in greenfield capacity are in place. Neither a centralised
30
State-owned nor a private-led market structure will allow the sustainable development of
a system with high shares of hydroelectric capacity and the need to sustain high reserve
margins. In terms of net benefits, it has been shown that a mixed ownership outperforms
the alternatives.
Recommendations
Regulatory frameworks have to be designed bearing in mind local capacities and
institutional approaches. The application of simple measures that contribute to restore
private investment flows, which can be easily implemented and monitored are therefore
recommended. For instance, the capacity payment suggested in Scenario C could be
assigned through transparent and competitive capacity auctions, conducted under the
purview of the regulator (i.e. as opposed to distributed among generators based on the
outputs of complex models), or even a parallel capacity market to the energy spot market
can be set up. Later on, and depending on the capacity of the system, a forward energy
trading market whose prices signal expectations about future supply/demand balances can
be developed.
31
Annex I. Causal Loop Diagram
Contracts Price CP
fies Rats ——_
. Return to
Investment, IRRj
System Marginal Fuel. Costs
Price SMP
Technology
Development Allowed Market
+ 5
Cost of Share, DMS
Rationing, COR
3
+
Market Share
2 Rain2. ‘Variable Costs, VCj Gap, MSG
Demand DOE Wo
+ f+
Desired IRR ; +4
Bidding Price BPj Projected Market SE sens 6 TAUERE,
Regulation on Share, He ae
Emissions, RE
Carbohi Projected Capacity Investment Capacity
Emissions, tCO2 Load Factor LFj per “— Addition ICA
as + ) a
é J Projected Registered <@— _ Capacity in
Comin. PRC + Construction, CC
Dispatched
Capacity DUj
# Reservoir x RL
Reserve Margin, RM vy; Velocity
- A Total Registered
Capacity, TRC
+ Available
Capacity Uj /
Projected on Reserve Margin
Demand PDOE Gipasiig Face / ; 4 Gap, RMG
Technology j, CFj
Projected Available
Capacity, PAC
Capacity Needed, CN Optimal Reserve
Margina, ORM
+
Projected Reserve
Margin, RM
32
Annex II Quantitative Comparative Analysis 2000-2020
Indicators Baseline Beenario.A Scenario}
Public Private
Total Registered Capacity by 2020 (MW) 19,699.16 19,312.95 16,717.54
Total Registered Capacity by 2016 (MW) 15,190.38 17,195.39 14,409.99
Total Registered Capacity by 2015 (MW) 15,323.39 16,785.00 14,518.69
Gas Based (MW) 6,746.50 6,884.86 6,045.79
Coal Based (MW) 1,276.26 991.69 521.05
Wind Based (MW) 960.20 720.20 94.49
LRMC (2000-2015) (USD/MWh) 24.30 32.30 19.20
LRMC (2000-2020) (USD/MWh) (1) 47.20 28.30 54.40
Total Cost to System as Electricity Sold (USD Million) (2) 180.86 101.95 418.36
Total Cost to System Considering both Electricity and Rationing Cost (USD Million) (3) 3036.01 101.95 4,735.41
Total Cost System (USD Million / year) (3) 9.043 5.10 20.92
Cumulative Reserve Margin 8.52 7.50 5.82
Cumulative Rationing (GWh) 2,009.25 - 3,076.99
Rationing Duration (months) 10.68 - 12.12
Rationing Period 2016.03-2016.92 - 2015.99-2017.00
Cumulative Carbon Emissions (Million Tons) 194.36 249.55 227.75
NET BENEFITS: Avoided Costs of Rationing - 2,934.06 -1,699.4
Note 1: After 2015 a drought forces the system into rationing and the LRMC increases. The magnitude of the COR influences very much this value.
Note 2: As total amount of electricity purchased at marginal spot price (does not consider rationing cost).
Note 3: Total Cost including rationing, per year (cost of rationing has been considered constant 100 USD/MWh for the calculation of LRMC and Rationing)
Note 4: Estimated considering the difference between the investment by PUs (government) in Scenario M and the baseline, which is 4,735 MWs. (see Table 6.1
for indicative capital costs)
Note 5: Considers the addition If the large hydroelectric plant Pescadero-Ituango (1600 MW)
33
Annex III The Costs and Benefits of Applying Different Values of Capacity Payment Period 2000-2020
Tests
No CxC | BLr=11 r=12 r=13 r=14 r=15 r=16
[Total Wind Capacity 2020 (MW) 461 960 1000.2 1000.2 1080 1080 1080
[Total Coal Based Capacity 2020 (MW) 1149 1276 1240 1270 1080 1080 1080
[Total Gas Based Capacity 2020 (MW) 5800 6750 6690 6700 6630 6630 6630
(Cost to System (only Electricity Price) (USD Million) 190.57 180.8 181.36 172.71 170.53 172.94 175.34
[Benefits in terms of Cost of Electricity Avoided 7 O77 9.21 17.86 20.04 17.63 15.23
[Total Cost to System Considering Rationing Cost (USD Million) | 5393.4 3036 1731 172.71 170.53 172.94 175.34
voided Rationing GWh 5,690 1996 1071 0 0 0 0
voided Cost of Rationing - 2357. 3,662 5,221 5,223 5,220 5,218
ICost of Capacity Payment 0 44.80 48.11 51.48 54.9 58.37 61.89
[Emissions 224.82 194.2 191.98 191.6 190.1 190.1 190
voided Emissions at Market Costs (USD Million) - 30.62 32.84 33.22 34.72 34.72 34.82
[TOTAL NET BENEFIT (USD Million) 0 2,343 3,646 5,202 5,203 5,196 5,190
ote: The Cost and Benefit Analysis is calculated against an scenario of no capacity payment or r=0, (as opposed to against the baseline). This however
Kdoes not changes the conclusions reached but only the magnitude of the quantitative results.
34
Annex IV The Costs and Benefits of Applying Different Values of Capacity Payment Period 2000-2020
Tests
BL R105 R110 R115 R118 R120 R125 R130 R135 R140
Total Wind Capacity 2020 (MW) 960 1000 1000 1000 840 1000 1080 1080 1080
Total Coal Capacity 2020 (MW) 1276 1244 919 919 1300 1269 1080 1080 1080
Total Gas Based capacity 2020 (MW) 6750 6693 6343 6343 6822 6705 6630 6630 6630
Cost to System (Electricity Price) (USD Million) 180.86 180.54 185.90 186.80 175.61 173.98 170.93 172.41 173.90
Total Cost System w/ Rationing Cost (USD Million) | 3,036.01 1,730.50 185.90 186.80 175.61 173.98 170.93 172.41 173.90
Total Cost Capacity Payment 44.80 46.94 49.08 50.36 51.21 53.34 55.48 57.61 59.74
Carbon Emissions (Million Tons CO2) 194.2 191.9 193 193 195.3 191.6 190.13 190.1 190.1
Avoided Emissions at Market Costs (USD Million) - 11.5 6 6 -5.5 13 19.5 20.5 20.5
TOTAL NET BENEFIT (USD Million) - 1,303.37 | 2,845.83 | 2,843.65 | 2,853.99 | 2,853.49 | 2,854.40 | 2,850.79 | 2,847.17
ote: The Cost and Benefit Analysis is calculated against an scenario of no capacity payment or r=0, (as opposed to against the baseline). This however does not
changes the conclusions reached but only the magnitude of the quantitative results.
35
Annex V Quantitative Comparative Analysis 2000-2020
Tiaisaters Baseline Scenario A Scenario B | Scenario C
CxC 11% CxC 14% CxC 130D Both
Total Registered Capacity by 2020 (MW) 19,699.16 19,507.48 19,507.36 19,080.01
Total Registered Capacity by 2019 (MW) 17,464.17 17,448.19 17,448.13 17,425.12
Total Registered Capacity by 2016 (MW) 15,190.38 15,984.12 15,984.11 16,123.70
Total Registered Capacity by 2015 (MW) 15,323.39 15,987.00 15,986.99 16,176.72
Gas Based (MW) 6,746.50 6,630.44 6,630.35 6,165.17
Coal Based (MW) 1,276.26 1,080.65 1,080.62 878.45
Wind Based (MW) 960.20 1,080.20 1,080.20 1,320.20
LRMC (2000-2015) (USD/MWh) 24.30 25.00 25.00 25.40
LRMC (2000-2020) (USD/MWh) (1) 47.20 22.80 22.80 22.80
Total Cost to System as Electricity Sold (USD Million) (2) 180.86 170.53 170.93 173.44
Total Cost to System Considering both Electricity and Rationing Cost (USD Million) (3) 3036.01 170.53 170.93 173.44
Total Cost System (USD Million / year) (3) 9.043 8.53 8.55 8.672
Total Cost CxC (USD Million) 44,80 54.90 55.48 59.08
Cumulative Reserve Margin 8.52 8.72 8.72 8.72
Cumulative Rationing (GWh) 2,009.25 0 0 0
Rationing Duration (months) 10.68 0 0 0
Cumulative Carbon Emissions (Million Tons) 194.36 190.14 190.13 181.88
Total Benefit = Avoided Cost of Rationing — Cost Regulatory Measure (Million USD) NA 2,855.38 2,854.40 2,847.28
Note 1: After 2015 a drought forces the system into rationing and the LRMC increases. The magnitude of the COR influences very much this value.
Note 2: As total amount of electricity purchased at marginal spot price (does not consider rationing cost).
Note 3: Total Cost including rationing, per year (cost of rationing has been considered constant 100 USD/MWh for the calculation of LRMC and Rationing)
36
Back to the Top
CC BY-NC-SA 4.0