CO2 Charges:

1
CO2 Charges

• How can we assess the impact on electricity
  prices?

Dr Anthony Downward, Prof. Andy
Philpott, Electricity Power Optimization
Centre, University of Auckland, New Zealand.
EPOC Winter Workshop 2012, University of Auckland
2
Overview

• Background
• New Zealand Emissions Trading Scheme
• New Zealand Electricity Market
• NZEM historic prices
• Initial analysis assuming perfect competition
• SDDP Model
• Results
• Discussion of Assumptions
• Imperfect Competition model
• Impact of Transmission
• Fixed-point hydro mark-up
• Cournot Game
• Results
• Conclusions

3
Background NZ ETS

• Emissions Profile
• Electricity Production (2010) (20 of emissions)
• Transport Fuels (2010) (40 of emissions)
• Agriculture (2015) (10 of emissions)
• Industrial Processes (2010) (10 of emissions)
• Waste (2013)
• Forestry (2008)
• Emissions Trading Scheme
• It is aimed to be an all fuels all industries
  scheme.
• Agriculture will not be included until 2015.
• There is a price cap of 25 / t CO2 until the end
  of 2012.
• Until 2013 there is a 1 for 2 surrender rate.
• At the end of 2011 a review of the NZETS was
  completed which proposes increasing the price cap
  by 5 / t CO2 each year from 2013.

4
Background NZ ETS
https//www.commtrade.co.nz
5
Background NZEM

• Electricity Market Structure
• New Zealand operates a real-time nodal pool
  market with vertically-integrated gentailers.
• There are five main generation companies, three
  of which are currently entirely state owned,
  although they operate as independent companies.
• Generation dominated by hydro, however, the peak
  load must be met by thermal generation and there
  is always a risk of a drought.
• Electricity Pricing
• Offers are submitted to the pool every half-hour
  and are cleared against demand.
• Offers do not have to reflect marginal cost.
• Electricity prices are computed at a nodal level,
  based on the marginal offer/bid for electricity.

6
Background EAF

• Carbon Leakage
• The carbon intensive industries may head offshore
  to countries without carbon taxes.
• This can have two effects the country loses jobs
  and export earnings, and global emissions can
  rise.
• Electricity Price Mark-up
• One of the main concerns is the impact that a
  carbon charge will have on electricity prices.
• Quantifying the increase in prices in a hydro
  dominated market is difficult, since the hydro
  opportunity costs increase as a result of the
  carbon charge.
• Emissions Allocation Factor
• In order to avoid compromising the
  competitiveness of New Zealand firms, it was
  decided to compensate certain trade-exposed
  industries.
• To do this, the electricity price mark-up must be
  computed.

7
Background Electricity Prices
8
Is there a price impact?

• Examine bidding behaviour
• Construct a regression model for electricity
  prices
• Assume a competitive market and analyse the
  prices resulting with and without carbon charges.
• If you assume imperfect competition, you could
  construct a Cournot or Supply Function model and
  examine the equilibrium prices under different
  carbon charges.

9
Initial Analysis

• In 2008, Dr. Tom Halliburton was contracted by
  the Ministry for the Environment to estimate the
  mark-up in electricity prices due to a charge on
  CO2 emissions.
• This analysis was performed using SDDP, which
  assumes a risk-neutral central planner will
  manage the hydro reservoirs so as to minimize the
  total system cost (including costs of outages).
• This model was run over a 23 year time-horizon
  (2009-2032) for different carbon costs.
• The model was very comprehensive, using an
  investment model (GEM) to predict the installed
  capacity of various technologies in the future.

10
Initial Analysis

•  

www.mfe.govt.nz
11
Initial EAF

• Based on the initial analysis, an EAF of 0.52 t
  CO2 / MWh was put in place from July 2010 with
  the condition that it be reviewed by the end of
  2012.
• The use of a perfectly competitive model was very
  contentious, at the time, since the a report for
  the Commerce Commission had recently stated that
  market power was being exercised (and there have
  been other recent examples).
• The Major Electricity Uses Group contracted Prof.
  Andy Philpott to examine the how the prices might
  change in a market with imperfect competition.

12
Imperfect Competition

•  

13
Cournot Dispatch Problem

• We now present the dispatch problem, associated
  with a Cournot game over a network.
• where A is a node-arc incidence matrix defining
  the topology of the network
• L is a matrix containing loop-flow data, to
  ensure the flow complies with Kirchhoff's laws.

14
Cournot Welfare

• We compute welfare in the following way

Generation
15
Cournot Nash Equilibrium

• At a Nash equilibrium, each generator solves the
  following problem
• The above problem is not necessarily convex.
• This makes it difficult to prove any general
  results governing how the equilibrium may change
  after the introduction of a carbon charge.

16
Cournot Single Node

• If we are dealing with a single-node network, or
  unlimited capacity network, it can be shown that
  increasing the cost of emissions is guaranteed to
  lead to equilibrium prices which are
  non-decreasing and non-increasing levels of
  emissions.
• However, the increase in price depends on a
  number of factors, such as which generators are
  marginal and the level of price elasticity in the
  market.
• Question
• Does this result hold when line capacities are
  introduced?

17
Cournot Two Node

•  

Plant Marginal Cost (c) Emissions (?)
Coal 40 / MWh 1.0 T / MWh
Gas 50 / MWh 0.4 T / MWh
Downward A. The Energy Journal, 31(4)159166
(2010)
18
Cournot Two Node

• The coal plant is situated at node 1, and the gas
  plant is at node 2.
• The two nodes are joined by a transmission line
  with a capacity of 125 MW.
• The demand curves at each node at shown below.

19
Cournot Two Node

• Without a charge on carbon, the gas plant is more
  expensive to run than the coal plant.
• Furthermore the demand at node 2 is larger than
  at node 1.
• These factors mean that the gas generator has
  incentive to withhold generation and congest the
  line towards node 2, at equilibrium.

20
Cournot Two Node

• Once a charge on carbon (a 26) has been
  applied, the situation changes.
• Now the gas plant is cheaper to run than the
  coal, due to the coal plants higher emissions.
• This leads to a different equilibrium outcome. At
  equilibrium, the line is not constrained and the
  generators compete as in a single node situation.

21
Cournot Two Node

• We will now consider the impact that this carbon
  charge has had on generation, prices, welfare and
  emissions.

Nodal Prices a 0 a 26
Node 1 102.03 99.83
Node 2 118.75 99.83
Generation a 0 a 26
Coal 198.5 175.89
Gas 137.5 205.01
Welfare a 0 a 26
Consumer 18,071 23,566
Producer 21,766 14,032
Carbon a 0 a 26
Emissions 253.5 t 257.9 t
Revenue 0 6,705
22
Cournot Example Summary

• Electricity Price Mark Up
• The previous example shows that in a market with
  transmission there are complicated interactions
  that mean the equilibrium prices do not vary
  smoothly with CO2 charges.
• In the example, we saw prices drop as a result of
  the CO2 charge (corresponding to a negative EAF).
  We could have constructed a similar example where
  the CO2 charge causes congestion, leading to a
  price increase much larger than the increase in
  marginal costs.
• In fact, depending on the shape of the residual
  demand curves, and the capacities of generators,
  the possible equilibrium mark-ups can vary widely.

23
Cournot Game for New Zealand
In this Cournot model, we use the following fuel
prices The plants that we model are as
follows
24
Approximating Hydro Offers
25
Computing the mark up

• Electricity Price Mark Up
• It is simple to compute the change in cost for a
  thermal generator for a given carbon tax.
  However, understanding how the hydro generators
  react to a carbon charge is much more
  complicated.
• In a competitive setting, hydro generators
  water-value functions incorporate
• the costs of marginal thermal generators, and
• shortage costs.
• We approximate this relationship by marking up
  historical hydro offer stacks by the expected
  change in price.
• This has feedback effect on the equilibrium
  prices, and so a fixed point must be found.

26
Hydro Mark up
26
27
Hydro Mark up
27
28
Price Mark up
29
Methodology
30
Methodology
31
Computing the mark up

• Computing the Fixed Point
• In the simplest case, when there is only one
  period type, we wish to find the mark up, K, such
  that
• K E(K),
• where E(K) is the average equilibrium electricity
  price mark up given a hydro mark up K.
• However, a single value of K is not enough, since
  there is a lot of variation in electricity market
  states over short horizons we must consider the
• types of periods (p) peak, shoulder, off peak,
• and over longer timescales
• hydro availability (i) wet, dry, uncertain,
  normal.

31
32
Fixed Point
Price Mark up
33
Effect of Hydrology
34
Calibration
35
Computing the mark up

• Computing the Fixed Point
• To account for this variation we must extend the
  notation.
• Let E(Ki,i,p) be the equilibrium price mark up
  due to carbon charges for market state i in
  period type p, with hydro mark up Ki.
• Then we compute the hydro mark up to be
• Kj Sip(rijp E(Ki,i,p)),
• where rijp is the probability that a extra unit
  of water in state j will be used in state i, and
  period p.
• Solving the above system, we compute a fixed
  point, K.

35
36
Cournot Game
We model the game as Cournot the firms in the
market own multiple plants, each with constant
marginal costs. A quantity, q, is injected for
each plant. The profit function for firm i
is where cj is the marginal cost of the
generator, which changes depending on the carbon
charge. We computed the equilibrium for 300
different periods. (25 periods for each of the
hydrology states, and time of day).
36
37
Results
Electricity price mark up in 25 normal, offpeak
period, K0
38
Results
38
Electricity price mark up in 25 dry, shoulder
periods, K0
39
Results

• Average Mark-ups due to carbon charge with no
  hydro mark up
• Converged EEFm values

Off Peak Shoulder Peak
Wet 0.00 0.00 0.39
Normal 0.46 0.45 0.43
Uncertain 0.82 0.23 0.28
Dry 0.75 0.00 0.00
Off Peak Shoulder Peak
Wet 0.00 0.00 0.39
Normal 0.64 0.67 0.68
Uncertain 0.68 0.66 0.70
Dry 0.85 0.85 0.85
39
40
Results

• Based on these individual EAFm values, we compute
  an overall EAFm of 0.65 t CO2 / MWh.
• If we perform a sensitivity analysis around
  relative frequency of wet to normal years, we
  compute EAFm values between 0.61 and 0.69 t CO2 /
  MWh.
• These figures are slightly higher than the 0.52 t
  CO2 / MWh computed using SDDP, which assumed a
  competitive market.
• There is currently a consultation document on the
  governments ETS website proposing increasing the
  EAF to 0.537 or 0.606 t CO2 / MWh.

40
41
Conclusions

• Emission allocations to industry are a
  significant expense to the government (and hence
  the taxpayer), and they therefore need to reflect
  the true additional costs that industry faces.
• An assumption of a perfectly competitive market
  provides neither an upper- or lower-bound on the
  electricity price increases due to carbon
  charges.
• Imperfect competition is much more difficult to
  model and the presence of hydro generation means
  provides that such models need to include the
  hydro generators anticipation of the thermal
  price increase.
• Our methodology provides a framework whereby we
  can compute such a mark up under imperfect
  competition.

41