Download Carbon Costs Pass-through Rates (using volume

Carbon Pass-Through Rates on Spot Prices in
Australian Electricity Market
Marjan Nazifi, Stefan Trück, Liangxu Zhu
Faculty of Business and Economics
Macquarie University
The 40th IAEE International Conference
Meeting the Energy Demands of Emerging Economics
18-21 JUNE 2017, SINGAPORE
1
Background
• Under the Paris Agreement, Australia has an emissions reduction target of 26 to 28 per
cent below 2005 levels by 2030. This would require the intensity of carbon emissions for
the economy to fall by around 65 per cent, and the emissions per capita being halved.
• Australian emission per unit of GDP have been extremely high: 0.78kg of carbon dioxide
equivalent in comparison to 0.43kg of carbon dioxide equivalent in OECD (IEA, 2009).
• Electricity generation accounts for about 35% of all CO2 emissions in Australia.
• The average emission intensity of the NEM in Australia has been around 0.91 tCO2/
MWh over the last 5 years in comparison to 0.4 tCO2 / MWh in EU over the last 5 years.
• Intensities vary significantly across states, e.g. 1.20 tCO2 / MWh for Victoria, less than
0.1 tCO2 / MWh in Tasmania.
2
Figure 1: NEM Generation by Fuel Type
3
Background (cont.)
• In February 2011, the Labour government proposed the socalled Clean Energy Bill: a two-stage carbon policy mechanism –
commencing with a fixed price carbon period from July 1, 2012
and transitioning to an emissions trading scheme (ETS) on July 1,
2015.
• One carbon permit allowed the discharge of 1 tonne of CO2 in a
compliance year.
• A fixed price carbon (introductory) phase came into effect on
July 1, 2012 with an initial price of $23. The tax was planned to
increase to $25.40 for the financial year 2014–15.
• The carbon tax was abolished with effect of July 2014 by the
Liberal-National coalition government that came into power in
4
September 2013.
Background (cont.)
• Experience from EU-ETS suggests that expectation about carbon price was
added to electricity futures prices.
• Mixed results on relationship between carbon and electricity spot prices (Bunn
and Fezzi, 2009; Nazifi and Milunovich, 2010; Gulli and Chernyavska, 2013).
• Strong relationship between returns of EU-ETS carbon futures and electricity
futures prices (Gronwald et al, 2011).
• For Australia, simulation studies on carbon pass-through suggest a range from
17% (McLennan Magasanik Associates 2008), 100% (ROAM Consulting 2008),
up to more than 300% (Simshauser & Doan 2009).
• Recently a few empirical studies on the pass-through rate of the CPM on
wholesale electricity prices in Australia (Gulli and Chernyavska, 2013; O’Gorman
and Jotzo 2014; Apergis and Lau, 2015; Nazifi, 2016).
5
Motivation
•
We extend the relatively sparse literature on the impact of the Australian CPM on wholesale spot
electricity prices
•
This paper adopts an empirical approach in order to estimate actual CPTRs in the NEM. Almost
unique environment to examine the impact of an introduction and abolishment of a Carbon Tax
on wholesale electricity prices:
i.
ii.
iii.
it considers regional markets where generation is almost entirely based on brown coal;
black coal; and gas fired power plants,
no significant increase in renewable energy for most of regional markets between 2009
and 2015 (apart from South Australia and Tasmania),
hardly any demand side elasticity in Australia, since NEM operates as a gross pool
market, i.e. all electricity has to be traded through the exchange (no bilateral contracts)
•
It develops a framework that examines carbon pass-through rates over a long sample. This
includes the period before the carbon tax became effective, the period of its lifetime from July
2012 to June 2014, as well as the period after the tax was repealed from July 2014 to June 2016.
•
It improves the estimates of the CPTRs by relaxing some of the assumption made by previous
studies (e.g. possible technological change).
•
It allows for a polynomial relation between electricity demand and spot prices rather than a
6
linear model.
The Australian National
Electricity Market (NEM)
• NEM is an interconnected grid comprising
several regional networks (NSW, QLD, SA,
TAS, and VIC).
• It operates as a mandatory wholesale pool
under the management of the Australasian
Energy Market Operator (AEMO).
• The wholesale market of the NEM is a real
time energy market and it is settled at a 30minute period.
• The NEM weighted average emissions
intensity is estimated 0.94 tCO2/MWh.
• Limited amounts of Electricity can also be
transferred across regions through so-called
interconnectors, e.g. between QLD-NSW,
NSW-VIC, VIC-TAS, etc.
7
Electricity Prices in Australia
• Before introduction of the carbon tax, Australian wholesale
electricity prices – although extremely volatile - were among the
lowest of all OECD countries (ABARE 2008).
• Australian average household spend (only) 2.4% of income on
electricity bills.
• Half-hourly Electricity spot prices can range from -$1,000 / MWh to
$13,800 / MWh.
• Australian electricity wholesale prices are among the most volatile
in the world (Mayer and Trück, 2015) what results in substantial
futures risk premiums (Handika and Trück, 2014).
8
Figure 2: Spot Price Behaviour
9
Figure 3: Emission Intensities
10
Table 1: Average Emission Intensities by state
Entire Period
Pre Carbon Tax
Period
Carbon Tax Period
Post Carbon Tax
Period
NSW
0.9145
0.9377
0.9048
0.8895
QLD
0.8354
0.8342
0.8300
0.8424
SA
0.5536
0.6181
0.5028
0.5078
VIC
1.2043
1.2436
1.1746
1.175
11
Table 2: Emission Intensities and the carbon costs
Emission
Factor
Additional
Cost (2012-13)
CPTR=1
Output
(TWH)
NSW
0.937
$21.55
73.4
QLD
0.834
$19.18
59.3
SA
0.618
$14.21
14.3
VIC
1.243
$28.59
56.1
12
Figure 4: Merit Order in the NEM (pre-tax)
13
Figure 5: Merit Order in the NEM (post-tax)
14
Statistical Modelling Approach
•
We develop an OLS regression model (following Sijm et al., 2008; Nazifi, 2016) :
𝑓
𝑃𝑡𝑒 − 𝑃𝑡 = 𝛼 + 𝑏1 𝑑𝑡𝑎𝑥 𝐶𝑡 + 𝑏2 𝑑𝑝𝑜𝑠𝑡 + 𝛾1 𝐷𝑡 + 𝛾2 𝐷𝑡2 + 𝛾3 𝐷𝑡3 + 𝜀𝑡
for each of the four major regional markets NSW, QLD, SA, VIC.
•
We use daily average volume-weighted wholesale spot prices (both uncapped and capped)
and average daily emission intensities for each state for three sub-periods:
i.
ii.
iii.
•
We employ a mechanism to filter out extreme observations in the spot prices using the
recursive seasonal model (RM) suggested by Janczura et al (2013):
i.
ii.
iii.
•
when the tax was not yet implemented ( 1 July 2009 – 30 June 2012)
when the tax was implemented (1 July 2012 – 30 June 2014)
when the tax was abolished (1 July 2014 – 30 June 2016)
capped at 100$AUD/MWh,
capped at 200 $AUD/MWh, or
capped at 300 $AUD/MWh
We apply Newey-West (HAC) estimators to address autocorrelation and heteroscedasticity in
15
errors.
How to Measure Fuel Costs
•
•
•
•
•
•
The model assumes that fuel costs are fully and directly passed on to electricity prices.
Fuel costs are calculated based on half-hourly electricity demand and short-run marginal
costs (SRMC) for each individual power generation instalment in the NEM based on data
provided by ACIL Tasman (2009), ACIL Allen Consulting (2014), AEMO (2016) and CO2CRC
Limited (2016).
Generators are ranked in ascending order of their short-run marginal costs, known as the
merit order.
Using data on half-hourly demand for each market, we simulate the dispatch of individual
power plants to calculate a weighted average of fuel costs for each half-hourly interval.
Thus, for the carbon-tax period, we consider the SRMC inclusive of the additional costs of
carbon for the merit order.
The calculated fuel costs for electricity generation are then subtracted from the volumeweighted wholesale electricity prices to generate the ‘spread’ as represented on the lefthand side of the equation.
16
Econometric Specification
•
The intercept α represents some stable and fixed components of the spread, such as
the fixed-cost elements.
•
A dummy variable 𝑑𝑡𝑎𝑥 is created, which is equal to 1 from 1 July 2012 to 30 June 2014,
and 0 otherwise to statistically evaluate the potential impact of the carbon tax.
•
𝐶𝑡 denotes the cost of the carbon permits required to cover emissions from generating
a MWh of electricity, which is calculated by using the official carbon tax rate (AUD /
tCO2-e) multiplied by the region’s emission intensity (tCO2-e/MWh) on a particular day.
•
A dummy for the post-tax period, 𝑑𝑝𝑜𝑠𝑡 is added which is equal to 1 from July 2014 to
30 June 2016, and 0 otherwise to investigate whether spot prices have reverted to their
pre-tax levels (the ‘stickiness’ of post carbon tax prices).
•
A polynomial relationship (up to order three) between demand and wholesale spot
electricity prices is considered due to the increasing relationship in a non-linear fashion
between prices and demand for the electricity market.
17
Figure 6: Relationship between Volume and Price
Relationship between daily volume and prices across the NEM
pre-tax period (blue), carbon tax period (red), and post-tax period (green) 18
Table 3: Carbon Costs Pass-through Rates (using volume-weighted
prices capped at 300/MWh)
Market
𝜶
𝒃𝟏
𝒃𝟐
𝜸𝟏
𝜸𝟐
𝜸𝟑
R2
Adj R2
NSW
-1737.43***
1.31***
19.85***
65.23***
-0.81***
0.0034***
35.63%
35.50%
QLD
-2328.41*
1.86***
20.02***
121.94**
-2.14**
0.0126**
35.36%
35.24%
SA
-423.58***
2.90***
19.24***
76.95***
-4.80***
0.1042***
34.42%
34.29%
VIC
-1266.83***
0.97***
19.70***
70.01***
-1.28***
0.0079***
44.51%
44.40%
19
Figure 7: Basslink Interconnector
20
Table 4: Regression Results (using volume-weighted prices capped at
100/MWh)
Market
𝜶
𝒃𝟏
𝒃𝟐
𝜸𝟏
𝜸𝟐
𝜸𝟑
R2
Adj R2
NSW
-235.86*
1.28***
17.08***
8.73*
-0.11**
0.0004**
54.11%
54.02%
QLD
-1571.23***
1.62***
15.78***
79.76***
-1.35***
0.0077***
49.93%
49.84%
SA
-197.61***
2.51***
12.46***
32.65***
-1.84***
0.0369***
46.20%
46.09%
VIC
-569.93***
0.93***
16.49***
30.35***
-0.53***
0.0032***
61.48%
61.41%
21
Table 5: Regression Results (using volume-weighted prices capped at
200/MWh)
Market
𝜶
𝒃𝟏
𝒃𝟐
𝜸𝟏
𝜸𝟐
𝜸𝟑
R2
Adj R2
NSW
-1380.11***
1.31***
20.02***
51.39***
-0.63***
0.0026***
40.24%
40.13%
QLD
-1546.71*
1.75***
18.91***
80.46*
-1.40*
0.0083*
39.60%
39.49%
SA
-161.81***
2.79***
17.58***
22.95**
-1.12*
0.0213
33.60%
33.47%
VIC
-1106.32***
0.96***
19.26***
60.62***
-1.10***
0.0067***
48.71%
48.61%
22
Table 6: Regression Results (using uncapped prices)
Market
𝜶
𝒃𝟏
𝒃𝟐
𝜸𝟏
𝜸𝟐
𝜸𝟑
R2
Adj R2
NSW
-11706.7***
1.29***
19.57***
437.96***
-5.43***
0.0223***
34.19%
34.06%
QLD
-15494.87**
2.00***
25.37***
813.77**
-14.22**
0.0828**
16.97%
16.81%
SA
-4270.72***
1.60***
12.63***
880.81***
-59.73***
1.3344***
45.98%
45.88%
VIC
-3659.35**
0.92***
19.94***
208.91**
-3.94**
0.0248**
23.28%
23.13%
23
Table 7: Regression Results (using daily prices)
Market
𝜶
𝒃𝟏
𝒃𝟐
𝜸𝟏
𝜸𝟐
𝜸𝟑
R2
Adj R2
NSW
-1958.42***
1.31***
18.82***
73.5***
-0.91***
0.0038***
35.75%
35.62%
QLD
-1922.66*
1.84***
19.2***
101.31*
-1.79*
0.0106**
35.61%
35.48%
SA
-357.16***
2.85***
17.84***
63.18***
-3.86***
0.0827***
32.98%
32.85%
VIC
-1075.13***
0.97***
18.56***
59.26***
-1.08***
0.0066***
45.58%
45.47%
24
Table 8: Regression Results (assuming linear relationship between demand and
volume-weighted prices capped at 300/MWh)
Market
𝜶
𝒃𝟏
𝒃𝟐
R2
Adj R2
NSW
-50.78***
1.33***
19.59***
0.76***
26.54%
26.45%
QLD
-107.28***
1.86***
19.2***
2.09***
33.63%
33.55%
SA
-67.33***
2.9.***
17.88***
4.61***
31.39%
31.31%
VIC
-51.41***
1.00***
19.29***
1.31***
39.92%
39.84%
𝜸
25
Summary & Conclusions
•
•
•
•
The results demonstrated a high degree of carbon costs pass-through across the
considered markets with the increase in electricity prices typically exceeding
carbon costs in particular for the markets in New South Wales, Queensland and
South Australia.
Results were relatively robust based on different model specifications and using
different prices series (e.g. uncapped or capped at $100, $200; average daily
prices; and linear demand relation).
Interestingly, the estimated carbon pass-through rates seemed to be inversely
related to emission intensities for each of the regions (relatively low for Victoria
and the highest for South Australia).
Prices have not reverted to their pre-tax levels after the tax was repealed,
possibly leading to carbon-induced profit effects (so-called “windfall profits”) for
some generators.
26
Summary & Conclusions (cont.)
•
•
•
•
•
With regards to emission intensities, carbon tax did not have the effect of
(substantially) reducing Emission Intensities in the NEM (a minor reduction of
intensities by 3.4% for NSW, less than 1% for QLD, and 5.5% for VIC).
Given the substantial increase in spot electricity prices, our results suggested that the
impact of the CPM on investment into ‘greener’ generation assets has been rather
limited.
The extent of the estimated CPTRs for the considered markets reveal that the
incidence of taxation is typically borne by consumers rather than generators.
Therefore, the carbon tax has led to relatively unfavourable outcomes for consumers.
It seemed that the CPM rather resulted mainly in short-term and minor changes in
generation behaviour, since it was not considered to remain in place over relevant
investment horizons.
A key implication for environmental policy makers is that any future mechanism for
carbon pricing in Australia should be accompanied by a stable and long-term policy
framework.
27
Thank you!
28