Embed Size (px)
Citation preview
California’s cap-and-trade program and emission
leakage: an empirical analysis
Chiara Lo Prete
The Pennsylvania State University
Ashish Tyagi
Frankfurt School of Finance and Management
Cody Hohl
The Pennsylvania State University
35th USAEE/IAEE North American Conference
November 14th, 2017
Houston, TX
0
100
200
300
400
500
600
Mil
lio
n m
etr
ic t
on
s o
f C
O2
eq
.
California’s GHG emissions
Peak:
487.63
2020 target:
431
2030 target:
256
2
2014:
441.53
Data source: California Air Resources Board
3
➢ California’s emission trading scheme is the first multi-sector cap-and-
trade program in North America
➢ It covers about 85% of the state’s GHG emissions (as of 2015)
➢ Status: compliance obligations began in January 2013, sunset in 2020
➢ Target: approximately 17% below 2013 emissions by 2020
➢ First deliverer approach: in-state electricity generators and electricity
importers are the point of regulation
California’s GHG cap-and-trade program
North
20.8 Nevada Oregon Border (NOB)
6.8
Utah
4.3
South
26.5
4
Annual net power flows
into California, Jan-Nov 2015
(Million MWh)
Data source: California ISO
5
0%
20%
40%
60%
80%
100%
CA in-stategeneration
0%
20%
40%
60%
80%
100%
Northwestimports
0%
20%
40%
60%
80%
100%
Southwestimports
Natural gas
Data source: California Energy Commission, 2015 data
Large hydro
Coal
Nuclear
Renewables
Unspecified power
6
«Repeal and replace» after 2020?
➢ The design of the existing cap-and-trade program has been the subject
of much debate. Concern has been voiced about:
• Emission leakage
• Contract shuffling
• Laundering
7
➢ Incomplete environmental regulation is likely to lead to emission
leakage (Bushnell et al., 2008; Fowlie, 2009; Goulder and Stavins, 2011;
Goulder et al., 2012)
➢ Emission leakage in regional CO2 cap-and-trade markets has typically
been examined ex ante (Wing and Kilodziej, 2008; Chen et al., 2011;
Bushnell et al., 2014; Caron et al., 2015)
➢ Empirical analyses are less common (Aichele and Felbermayr, 2015;
Fell and Maniloff, 2015)
Incomplete environmental regulation
and emission leakage
➢ This study empirically investigates the leakage effects of California’s
cap-and-trade program on electricity generation in the Western
Interconnection
➢ We examine whether the policy affected production and efficiency of
coal and natural gas fired power plants in California differently from
other regions in the WECC
8
Objective
9
➢ We construct a plant-level dataset including monthly observations from
2011 to 2015 on:
• Net generation
• Fuel consumption
• Operating capacity
• Power control area and location
• CO2 emission rates
• Emission abatement control types and number
• Fuel cost, as delivered to the plant (or state average fuel cost for
electric power generation)
• CO2 prices
• Electric load in the plant’s planning area
• Renewable and nuclear generation in the plant’s state
• Heating/cooling degree days and drought indices in the plant’s
climate division
• CAISO net power imports on major transmission interfaces
Data
➢ We use a differences-in-differences approach to examine impacts of
California’s cap-and-trade program on capacity factors and heat rates
of coal and NGCC power plants in the Western Interconnection
➢ To limit selection bias due to changes in group composition across
time, we restrict our sample to plants that are present pre and post
treatment
➢ We present results for two model specifications focusing on capacity
factors
10
Empirical strategy
➢ Specification 1 considers WECC sub-regions outside of California as
potential leakers, and MRO, SPP, and TRE as controls
11
Empirical strategy
12
➢ Specification 1
where i refers to power plant of generation type g and t denotes the
month (January 2011 to December 2015)
is capacity factor (in %) for plant i of type g
is equal to 1 if plant i of type g is in CA, and month t is
January 2013 or later
is equal to 1 if plant i of type g is in sub-region h, and
month t is January 2013 or later
Empirical strategy
CA
igtTREAT
h
igtLEAK
igtY
CA h '
igt CA hY g=1,....Gigt igt it i t m y t igt
h
TREAT LEAK X SONGS
0.00
0.10
0.20
0.30
0.40
0.50
0.60
0.70
0.80
0.90
1.00
2011B1 2011B2 2012B1 2012B2 2013B1 2013B2 2014B1 2014B2 2015B1 2015B2
Coal capacity factors
CAMX
NWPP
RMRG
SRSG
13
0.00
0.10
0.20
0.30
0.40
0.50
0.60
0.70
0.80
0.90
1.00
2011B1 2011B2 2012B1 2012B2 2013B1 2013B2 2014B1 2014B2 2015B1 2015B2
Coal capacity factors
CAMX
MRO
NWPP
RMRG
SPP
SRSG
TRE
14
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
2011B1 2011B2 2012B1 2012B2 2013B1 2013B2 2014B1 2014B2 2015B1 2015B2
NGCC capacity factors
CA
NWPP
RMRG
SRSG
15
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
2011B1 2011B2 2012B1 2012B2 2013B1 2013B2 2014B1 2014B2 2015B1 2015B2
NGCC capacity factors
CA
MRO
NWPP
RMRG
SPP
SRSG
TRE
16
17
➢ Specification 1
Xit includes:
✓ Log of electric load in the plant’s planning area
✓ Input price ratio (levels and square) for the plant
✓ Log of renewable generation in the plant’s state
✓ Heating/cooling degree days in the plant’s climate division
Empirical strategy
CA h '
igt CA hY g=1,....Gigt igt it i t m y t igt
h
TREAT LEAK X SONGS
18
➢ Specification 1
are plant fixed effects
are period fixed effects
are month-by-year fixed effects
are year fixed effects
is equal to 1 in California starting from February 2012
Empirical strategy
tSONGS
i
t
m
y
CA h '
igt CA hY g=1,....Gigt igt it i t m y t igt
h
TREAT LEAK X SONGS
19
➢ Specification 2 considers selected balancing authorities within WECC
as potential leakers, and rest of WECC as control
Empirical strategy
CISO
NEVP
BPAT IPCO
WACMPACE
WAUW
NWMT
PNM
PSCO
AZPS
WALC
BANC
EPETEPC
CFE
IID
LDWP
TIDC
DEAA
GRMA
HGMA
GRIF
PACW
PGE
SCL
TPWR
GRID
CHPD
GCPD
AVAPSEI DOPD
WWA
GWA
SRP
20
➢ Specification 2
where i refers to power plant of generation type g, and t denotes the
month (January 2011 to December 2015)
j, k, l refer to selected balancing authorities north, east and south
of California, respectively, which:
• include a significant share of generation from plants of type g
• are connected to California via major transmission lines
Empirical strategy
CA j k l
igt CA j k l
'
Y
g=1,....G
igt igt igt igt
j k l
it i t m y t igt
TREAT LEAK LEAK LEAK
X SONGS
21
Preliminary results
Treatment effects – capacity factors
Coal steam Gas CC
Specification 1 Specification 2 Specification 1 Specification2
CA - - -0.04** -0.04
CAMX 0.04*** - - -
NWPP 0.04*** - 0.07** -
SRSG 0.01 - 0.02 -
RMRG 0.03 - -0.001 -
North - -0.05 - 0.10***
East - 0.04*** - 0.06***
South - 0.01 - -0.03
➢ Robust standard errors are clustered at the balancing authority level
22
Preliminary results
Treatment effects – capacity factors
Coal steam Gas CC
Specification 1 Specification 2 Specification 1 Specification2
CA - - -0.04** -0.04
CAMX 0.04*** - - -
NWPP 0.04*** - 0.07** -
SRSG 0.01 - 0.02 -
RMRG 0.03 - -0.001 -
North - -0.05 - 0.10***
East - 0.04*** - 0.06***
South - 0.01 - -0.03
➢ Robust standard errors are clustered at the balancing authority level
23
➢ Preliminary results from the capacity factor models suggest some
evidence of carbon leakage from California’s emission trading program
➢ NG combined cycles in California had a statistically significant
reduction of ~4% in capacity factors since January 2013
➢ Displaced generation was primarily made up for by increased coal and
NGCC generation in central WECC
➢ Implication: implementing border adjustments to address leakage post
2020 is critical
➢ Future work: a) model effects on efficiency more accurately; b) refine
empirical approach to deal with unobserved time-varying confounders;
c) estimate the scale of leakage
Summary and future work
Summary stats
Coal steam 2011-2012 2013-2015 2011-2012 2013-2015
pre-treatment cap factor
post-treatment cap factor
pre-treatment heat rate
post-treatment heat rate
# of plants
average size (MW) average SD average SD average SD average SD
CAMX_CA 1 57 0.74 0.08 0.67 0.08 42412 3379 36937 14110
CAMX_Non_CA 1 1800 0.72 0.21 0.77 0.15 9970 344 9653 125
NWPP 20 596 0.72 0.24 0.75 0.21 12567 7278 13203 8474
RMRG 16 490 0.75 0.19 0.76 0.19 11237 2504 10911 934
SRSG 9 1172 0.71 0.20 0.70 0.18 10843 1006 10784 1019
MRO 61 387 0.57 0.23 0.57 0.22 16282 9916 16849 10650
SPP 29 692 0.68 0.21 0.64 0.21 11279 2652 11300 2515
TRE 18 1057 0.72 0.24 0.68 0.24 10788 1339 11342 3107
NGCC 2011-2012 2013-2015 2011-2012 2013-2015
pre-treatment cap factor
post-treatment cap factor
pre-treatment heat rate
post-treatment heat rate
# of plants
average size (MW) average SD average SD average SD average SD
CAMX_CA 68 262.0 0.55 0.29 0.56 0.29 9174 2218 9389 2725
CAMX_Non_CA 2 272.0 0.43 0.28 0.42 0.31 8508 1104 8778 4479
NWPP 23 360.1 0.39 0.30 0.53 0.29 8654 3691 8379 2283
RMRG 8 331.4 0.26 0.21 0.27 0.21 10222 5078 9790 3882
SRSG 28 567.5 0.43 0.30 0.46 0.29 8684 2336 8483 1957
MRO 17 279.4 0.22 0.23 0.23 0.24 9715 3945 10095 4186
SPP 23 545.7 0.42 0.25 0.39 0.24 8154 1379 8292 1600
TRE 60 614.3 0.46 0.24 0.46 0.23 9095 2645 9014 2574
0
100
200
300
400
500
600
Mil
lio
n m
etr
ic t
on
s o
f C
O2
eq
.
California’s GHG emissions
Peak:
487.63
2020 target:
431
2030 target:
256
2
Electricity generation
(in state)
Electricity generation
(imports)
Other sectors
2014:
441.53
Data source: California Air Resources Board
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
2012 Net Generation Shares by Tech-Fuel Mix
Non-hydro Renewables Hydro Nuclear Coal Steam NG CC Other Fossil
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
2012 Generation Mix by NERC Sub-Region
Non-hydro Renewables Hydro Nuclear Coal Steam
NG Steam NG CC NG Combustion Turbine Other Fossil
5
CARB’s response
Source: Cullenward and Coghlan, 2016
2
Comparison of cap-and-trade programs in the U.S.
California’s GHG cap-and-trade
program
Regional Greenhouse
Gas Initiative (RGGI)
Participating states CA 9 Northeastern and Mid-
Atlantic states (CT, DE, MA,
MD, ME, NH, NY, RI, VT)
Gases covered 6 major GHGs (CO2, CH4, N2O, HFCs, PFCs,
SF6), plus NF3 and other fluorinated GHGs
CO2
Sectors covered Electricity generators (including imports)
Industrial sources
Distributors of transportation fuel
Distributors of natural gas and other fuel
Fossil fuel electricity generators
(excluding imports)
Point of regulation Electricity generators (within CA)
Electricity importers
Industrial facility operators
Fuel distributors
Electricity generators (within
RGGI states)
Coverage threshold Emitters of at least 25,000 metric tons CO2e
annually
Fossil fuel plants with
nameplate capacity > 25 MW
Emission coverage 85% of GHG emissions (2015) 17% of GHG emissions (2012)
Status First auction on November 1, 2012;
compliance obligations began on January 1, 2013
Compliance obligations began
on January 1, 2009
2
Comparison of cap-and-trade programs in the U.S.
California’s GHG cap-and-trade
program
Regional Greenhouse
Gas Initiative (RGGI)
Participating states CA 9 Northeastern and Mid-
Atlantic states (CT, DE, MA,
MD, ME, NH, NY, RI, VT)
Gases covered 6 major GHGs (CO2, CH4, N2O, HFCs, PFCs,
SF6), plus NF3 and other fluorinated GHGs
CO2
Sectors covered Electricity generators (including imports)
Industrial sources
Distributors of transportation fuel
Distributors of natural gas and other fuel
Fossil fuel electricity generators
(excluding imports)
Point of regulation Electricity generators (within CA)
Electricity importers
Industrial facility operators
Fuel distributors
Electricity generators (within
RGGI states)
Coverage threshold Emitters of at least 25,000 metric tons CO2e
annually
Fossil fuel plants with
nameplate capacity > 25 MW
Emission coverage 85% of GHG emissions (2015) 17% of GHG emissions (2012)
Status First auction on November 1, 2012;
compliance obligations began on January 1, 2013
Compliance obligations began
on January 1, 2009
4
➢ Unspecified imports are assigned a default emission factor of 0.428 ton
CO2/MWh (representative of a fairly clean natural gas plant)
➢ Under this environmental design, the presence of a default emission
factor affects electricity importers’ decisions and may create perverse
incentives
➢ Two primary concerns:
• Laundering: electricity importers with long-term contracts with
higher-emitting out-of-state power choose not to report the
emission content of their imports
• Resource shuffling: electricity importers rearrange their financial
contracts by replacing higher-emitting resources with lower-
emitting resources
Default emission factor
7
➢ Because of its hybrid approach to measuring emissions from the
electricity sector, California’s emission trading scheme is vulnerable to
both leakage and reshuffling
➢ Physical relocation of electricity generation from the regulated region
(California) to the unregulated region (the rest of the Western
Interconnection) → leakage
➢ No change in electricity generation in California and the rest of the
Western Interconnection → reshuffling
Incomplete environmental regulation,
emission leakage and reshuffling
14
Preliminary results
Treatment effects – heat rates
Coal Steam Gas Steam Gas CC Gas turbine
CA -0.05***
(0.005)
-0.01
(0.012)
-0.01
(0.005)
0.01**
(0.004)
NWPP -0.01
(0.008)
-0.03*
(0.017)
-0.02
(0.014)
-0.03
(0.073)
SRSG -0.02
(0.015)
0.01
(0.015)
-0.01***
(0.005)
-0.02
(0.014)
RMRG -0.01***
(0.002)
-0.06**
(0.029)
0.02**
(0.011)
-0.01***
(0.004)
N 2,963 2,880 8,629 11,191
➢ Robust standard errors are clustered at the state level
➢ Xit include SONGS, logged capacity factor, CDD, renewable
generation, RPS state target
0
2,000
4,000
6,000
8,000
10,000
12,000
14,000
16,000
2011B1 2011B2 2012B1 2012B2 2013B1 2013B2 2014B1 2014B2 2015B1 2015B2
Coal heat rates
CAMX
NWPP
RMRG
SRSG
0
2,000
4,000
6,000
8,000
10,000
12,000
2011B1 2011B2 2012B1 2012B2 2013B1 2013B2 2014B1 2014B2 2015B1 2015B2
NGCC heat rates
CA
NWPP
RMRG
SRSG
7
➢ This paper empirically investigates the effects of California’s cap-and-
trade program on electricity generation in WECC
• Is there evidence of physical relocation of electricity generation
from CA to the rest of WECC?
• Is there evidence of “beneficial leakage”?
• How to account for reshuffling?
Research objective
9
➢ What has been the impact of California’s emission trading program on
capacity factors and heat rates of fossil fuel-fired baseload power plants
in WECC?
➢ Technologies considered: NGCC, natural gas and coal steam turbines
First step of the analysis
Actual Electricity OutputCapacity Factor =
Potential Electricity Output
Fuel Input for Power GenerationHeat Rate =
Actual Electricity Output
12
➢ Differences-in-differences model using monthly data
➢ The model will be estimated separately for plants of different
technology types
➢ Expectations on the sign of the treatment effects in the presence of
emission leakage
Differences-in-differences framework
CA NW SW '
it CA NW SWY it it it it t i itTREAT TREAT TREAT X
Capacity factor
specification
Heat rate
specification
CA αCA<0 αCA< or >0
NW, SW αNW, αSW>0 αNW, αSW<0
3
➢ First multi-sector cap-and-trade program in North America
➢ Covers about 85% of the state’s GHG emissions (as of 2015)
➢ Expected to drive about 22% of emission reductions needed to reach
the 1990 levels
➢ Status: compliance obligations began in January 2013, sunset in 2020
➢ Target: approximately 17% below 2013 emissions by 2020
➢ First deliverer approach: in-state electricity generators and electricity
importers are the point of regulation
California’s GHG cap-and-trade program
9
➢ Differences-in-differences approach
➢ Treated group:
➢ California
➢ NWPP, SRSG and RMRG (leaker regions)
➢ power plants in WECC that were in operation in January 2013 or
later
➢ We tested alternate control groups:
➢ Power plants of the same technology type in U.S. states that are
not subject to a cap-and-trade program (RGGI)
➢ Power plants of the same technology type in NERC sub-regions
that had similar pre-treatment trends and generation mix as the
treated group
➢ Show graphs?
Methodology
13
Preliminary results
Control sample sensitivity analysis – capacity factors
Coal Steam Gas Steam Gas CC Gas turbine
CA -0.18** -0.03*** -0.03*** 0.01***
NWPP 0.21*** 0.11** 0.14*** -0.002
SRSG 0.21*** 0.02 0.01 0.02
RMRG -0.02*** 0.01 0.03*** 0.01
The model includes plant and monthly fixed effects. Capacity factors are in % point. Standard errors are clustered at the
state level. *, **, *** denote statistical significance at the 10%, 5%, and 1% level, respectively. Controls include power
plants of the same technology type in U.S. states that are not subject to a cap-and-trade program (RGGI)
8
➢ Using a diff-in-diff approach, we evaluate the impact of California’s
cap-and-trade program on capacity factors and heat rates of coal and
NGCC power plants in WECC sub-regions (Specification 1) and
selected balancing authorities (Specification 2)
Empirical strategy
➢ Differences-in-differences model using monthly data from January
2011 to December 2015
The control sample includes plants of the same technology type in
NERC regions with similar pre-treatment trends and generation
mix as the treated group:
✓ SPP
✓ TRE
✓ DELTA, VACAR (SERC)
✓ FRCC
10
Empirical strategy
CA NWPP SRSG RMRG
igt CA NWPP SRSG RMRG
'
Y
g=1,....G
igt igt igt igt
it i t t igt
TREAT TREAT TREAT TREAT
X SONGS
4
20.8
North
NOB6.8
Utah4.3
South26.526.5
South
Annual net power flows
into California, Jan-Nov 2015
(million MWh)
Data source: California ISO
CISO
NEVP
BPAT IPCO
WACMPACE
WAUW
NWMT
PNM
PSCO
AZPS
WALC
BANC
Western
Electricity
Coordinating
Council (WECC)
EPETEPC
CFE
IID
LDWP
TIDC
DEAA
GRMA
HGMA
GRIF
PACW
PGE
SCL
TPWR
GRID
CHPD
GCPD
AVA
PSEI DOPD
WWA
GWA
SRP
