Burning wood pellets for US electricity generation? A regime switching analysis

Abstract

Applying a regime switching model under the theoretic framework of real options, we inspect the optimal timing boundaries for coal and coal mixed wood pellets as two alternative fuels for a power plant in Georgia, United States. Results indicate that cofiring wood pellets with coal is generally not a commercially viable option. However, lower-level (with wood pellets < 15%) cofiring could have been feasible during the infancy period (2009–2011) when wood pellet price was declining. Sensitivity analysis shows that our conclusions are robust and the most important factors are relative prices of coal and mixed fuel. Therefore, we reject the null hypothesis that cofiring is economically feasible and suggest using policy vehicles to stimulate the bioenergy market and meet the greenhouse gas emission reduction target. In particular, a subsidy of $1.40/mmbtu to the 10% mixed fuel or a tax of $1.50/mmbtu on coal would prompt the conversions of coal-only power plants to cofiring ones, and a subsidy of $0.45/mmbtu to the 10% mixed fuel or a tax of $0.50/mmbtu on coal would maintain existing cofiring power plants in the status quo.

Full text

Burning wood pellets for US electricity generation? A regime
switching analysis
Bin Mei
a,
⁎, Michael Wetzstein
b
a
University of Georgia, Warnell School of Forestry and Natural Resources, Athens, GA 30602, USA
b
Purdue University, Department of Agricultural Economics, West Lafayette, IN 47907, USA
abstractarticle info
Article history:
Received 26 October 2015
Received in revised form 5 July 2016
Accepted 29 May 2017
Available online 09 June 2017
JEL classification:
C61
D81
Q23
Q42
Q58
Applyinga regime switching model under the theoretic framework of realoptions, we inspect the optimaltiming
boundaries for coal and coal mixed wood pellets as two alternative fuels for a power plant in Georgia, United
States. Results indicate that cofiring wood pellets with coal is generally not a commercially viable option.
However, lower-level (with wood pellets b15%) cofiring could have been feasible during the infancy period
(2009–2011) when wood pellet price was declining. Sensitivity analysis shows that our conclusions are robust
and the most important factors are relative pricesof coal and mixed fuel. Therefore, we reject the nullhypothesis
that cofiring is economically feasible and suggest using policy vehicles to stimulate the bioenergy market and
meet the greenhouse gas emission reduction target. In particular, a subsidy of $1.40/mmbtu to the 10% mixed
fuel or a tax of $1.50/mmbtu on coal would prompt the conversions of coal-only power plants to cofiring ones,
and a subsidy of $0.45/mmbtu to the 10% mixed fuel or a tax of $0.50/mmbtu on coal would maintain existing
cofiring power plants in the status quo.
Published by Elsevier B.V.
Keywords:
Bioenergy
Decision-making
Forestry
Greenhouse gas emission
Real options
1. Introduction
Historically, coal is the major fuel type for power plants. Electricity
generated from coal-fired power plants accounts for N40% and 39%
globally and within the United States, respectively (EIA, 2016). On a
per-unit energy basis, coal is one of the largest emitters of carbon diox-
ide among all fossil fuels, and coal-fired power plants represent a major
source of man-made carbon dioxide emissions. To reduce greenhouse
gas (GHG) emissions, most countries have set reduction targets. The
world-leader in this effort is the European Union (EU) with the United
Kingdom (UK) as an EU leader. In recent years, the EU in general and
the UK in particular have burned an increasing amount of biomass for
electricity generation. In 2015, the United States launched the Clean
Power Plan aimed to lower carbon dioxide emitted by electrical power
generation by 32% within 25 years relative to the 2005 level. The plan
is focused on reducing emissions from coal-burning power plants,
as well as increasing the use of renewable energy, and energy
conservation.
1
Given the fact that electricity produced from renewable
resources is b7% in the US (EIA, 2016), there remains a great expansion
potential in the bioenergy market.
A typical coal-fired power plant bears a huge capital investment
with a design life of 20 to 50 years. Therefore, it is usually not econom-
ical to totally abandon a coal-fired power plant and replace it with
cleaner technology prior to the end of its useful life. Nonetheless, it is
feasible to substitute some portion of the coal by biomass (cofire coal
with biomass) so as to reduce carbon emissions. In particular, wood
Energy Economics 65 (2017) 434–441
⁎Corresponding author.
E-mail address: bmei@uga.edu (B. Mei).
1
Specifically, the Environmental Protection Agencyrequires individual states to imple-
ment theirplans by focusing onthree building blocks: increasing the generation efficiency
of existingfossil fuel plants, substituting lowercarbon dioxide emitting natural gas gener-
ation for coal powered generation, and substituting generation from new zero carbon di-
oxide emitting renewable sourcesfor fossil fuel powered generation. Thisstudy focuses on
the last one.
http://dx.doi.org/10.1016/j.eneco.2017.05.025
0140-9883/Published by Elsevier B.V.
Contents lists available at ScienceDirect
Energy Economics
journal homepage: www.elsevier.com/locate/eneeco

pellets
2
are easily adaptable to automated combustion systems and the
cost to convert existing coal boilers to mixed fuel burning is less prohib-
itive than plant retirement (Zhang et al., 2010). The saving of GHG emis-
sions from wood pellets ranges from 72.6% to 82.4% for each kWh of
electricity (Dwivedi et al., 2011). Within the EU and specifically in the
UK, many power plants are cofiring wood pellets with coal as a transi-
tion option toward a carbon-free power sector. This has created a rapid-
ly growing international market for wood pellets. Given the high
productivity of the forest sector in the US Southeast, much of this mar-
ket is supplied by southeastern wood pellet mills (Spelter and Toth,
2009). Forisk Consulting (2015) projects that US wood pellet produc-
tion could grow from about 5 million tons in 2009 to near
18 million tons by 2018, of which, 97% would be intended for export
markets.
Corresponding to the expanded supply, real wood pellet prices have
been generally declining from 2009 to 2012 and since stabilized (Fig. 1).
In the same period, coal prices have steadily declined, primarily because
of the competition from declining natural gas prices, resulting from the
advent of commercially viable hydraulic fracturing technologies and
horizontal drilling methods. In terms of price volatility, both wood pellet
and natural gas exhibit higher variations than coal. Therefore, an in-
triguing question for coal power-plant managers is how to make the op-
timal decision on fuel selection. In the energy economics literature, a
few studies have examined this issue. Specifically, applying real options
analysis, Pederson and Zou (2009) evaluate ethanol plant investments;
Lee and Shih (2010),Lima et al. (2013),andMonjas-Barroso and
Balibrea-Iniesta (2013) study solar- and wind-energy projects; Song
et al. (2011), and Gazheli and Corato (2013) examine the conversion
option of traditional farmland for energy crops; Bednyagin and
Gnansounou (2011),Detert and Kotani (2013),andZambujal-Oliveira
(2013) investigate the investment decisions among combined-cycle,
coal-fired, wind, solar, and nuclear power plants; Cheng et al. (2011)
assess the clean-energy mix policy; and Siddiqui and Fleten (2010) an-
alyze the staged commercialization and deployment of alternative ener-
gy technologies.
Past research on wood pellets mainly focuses on decentralized
household heating systems (e.g., Claudy et al., 2011; Hyysalo et al.,
2013; Michelsen and Madlener, 2012). Studies on wood pellets for elec-
tricity generation, however, have been limited. Steininger and
Voraberger (2003) employ a computable general equilibrium model of
the Austrian economyand demonstrate that fostering the use of cofiring
could lead to a decline in both gross domestic product (GDP) and em-
ployment. Ehrig and Behrendt (2013) assert that cofiring wood pellets
with coal represents one of the most cost-attractive ways to reach the
EU-2020 carbon targets. Dwivedi et al. (2014) reveal that the use of
wood pellets for electricity generation could reduce the UK's GHG emis-
sions by 50–68% relative to fossil fuels. Xian et al. (2015) account for un-
certain energy markets and examinethe economic feasibility of cofiring
wood pellets with coal for electricity generation. In this study, we apply
a regime switching model under the framework of real options analysis
to investigate the economic boundary conditions between coal and coal
mixed with wood pellets as the fuel for power plants. We intend to con-
tribute to thecurrent literature by considering reciprocal switch options
between coal-only and cofiring for a power plant, and incorporating the
switch cost explicitly as a function of the energy prices.Considering the
shifting energy patterns in the US market (Fig. 1), we conduct analyses
on two distinct periods in addition to the whole sample period. One is
the infancy period (2009–2011), which is the early stage when coal
prices are relatively high and wood pellet prices are declining because
of initial rapid supply expansion. The other is the substitution period,
when cheap natural gas undermines coal's dominance as the fuel for
US power plants. The null hypothesis is that both coal-only and cofiring
are economically viable options for US power plants, which solely de-
pends on contemporary market situations but not government
involvement.
3
2. Method
Based upon the classic real options approach proposed by Dixit and
Pindyck (1994),Adkins and Paxson (2011) examine the reciprocal
energy-switching options and provide a quasi-analytical solution for
the case of two competing energy inputs. Extending their analysis, we
adopt a general regime switching model, which incorporates price un-
certainty of two alternative fuels to investigate a power plant's optimal
choice of the fuel type. Consider an active, perpetual operating power
plant that turns the chemical energy in coal into electricity and has an
option to exchange the incumbent fuel (coal) with a substitute fuel
(coal mixed with wood pellets). The switch is reciprocal and incurs a
known sunk cost K
ij
,i,j∈{c,m} and i≠j.
4
Gains from a switch include
the net cost saving from using cheaper fuel and the option value of
switching back.
Price for fuel X
i
,i∈{c,m}, is assumed to follow a geometric Brownian
motion,
dXi¼αiXidt þσiXidzið1Þ
where αis the drift rate, σis the volatility rate and dz is the increment of
a standardWiener process. Correlation between the two price variables
is described by ρ(|ρ|≤1), so that cov(dX
c
,dX
m
)=ρσ
c
σ
m
dt. To state the
valuation relationship in terms of one unit of output, price for each
fuel can be adjusted by a conversion factor.
5
The function F
i
(X
c
,X
m
), i∈{c,m}, denotes the plant value from using
fuel iand the embedded switch option, which depends on prices of
both the incumbent and substitute fuels. Using the dynamic program-
ming approach, the following partial differential equation can be
2
Wood pellets are small nuggets of compressed, sawdust-sized wood fiber that have
higher energy density andlower moisture contentthan their raw input. The sustainability
of wood pellets as feedstock for energy is largely a matter of carbon cycle calculations,
which depends on the originand type of trees used for woodpellets. We believethat burn-
ing wood pellets locally for energy is more carbon efficient than burning coal, even after
accounting for the emissionsfor collecting and processing biomass.
Fig. 1. Weeklyreal energy prices ($/mmbtu) for 06/05/2009–04/25/2014. Deflator: PPI for
crude material, base time period January 2013.
3
The EU biomass market is driven by government mandates. The same has not been
mirrored in the US.
4
Letter cfor coal and mfor mixed fuel (coal mixed with wood pellets). K
c
denotes the
conversion cost from coal to mixed fuel and K
m
denotes the conversion cost from mixed
fuel to coal. For example,for a coal-burningpower plant to burn wood and meet emission
requirements, some accommodations to facility operation and physicalstructure are nec-
essary, including ash and air emission control, hard coating cleaning, wood storage, and
grinding and blowing systems.
5
1kWh=0.0034mmbtu.
435B. Mei, M. Wetzstein / Energy Economics 65 (2017) 434–441

obtained
1
2σ2
cX2
c
∂2Fi
∂X2
cþ1
2σ2
mX2
m
∂2Fi
∂X2
mþρσcσmXcXm
∂2Fi
∂XcXm
þαcXc
∂Fi
∂XcþαmXm
∂Fi
∂Xm
−μFiþY−Xi¼0ð2Þ
where μis the discount rate, and Yis the output (electricity) price net of
operating costs. The generic valuation function F
i
takes the form
FiXc;Xm
ðÞ¼AiXβi
cXηi
mþY
μ−Xi
μ−αið3Þ
where A(A≥0), βand ηare unknown parameters of the product power
function. The first term in Eq. (3) represents the option value of
switching fuel inputs, and the last two terms represent the value of op-
eration without any switch option. By applying the limiting boundary
conditions, it can be shown that
Fc¼Ac4Xβc4
cXηc4
mþY
μ−Xc
μ−αcð4Þ
Fm¼Am2Xβm2
cXηm2
mþY
μ−Xm
μ−αmð5Þ
where β
c4
N0, η
c4
≤0, β
m2
≤0, and η
m2
N0. Using the value-matching con-
ditions, smooth-pasting conditions, and the two characteristic root
equations, a system of eight equations can be established and the switch
timing boundaries can be determined numerically. The price ratios
along the two discriminatory boundaries are given by
Wcm ¼Xc
Xm
N1andWmc ¼Xm
Xc
N1ð6Þ
where W
ij
designates the price ratio when fuel icurrently in use should
be replaced by fuel j. Imposing the property of homogeneity of degree
one on the value functions (i.e., β
c4
+η
c4
=1 and β
m2
+η
m2
=1) and
the conversion cost function, the value-matching and the smooth-
pasting conditions are
Ac4Wβc4
cm −Wcm
μ−αc¼Am2Wβm2
cm −1
μ−αm
−kcWϕc
cm ð7Þ
Am2W1−βm2
mc −Wmc
μ−αm¼Ac4W1−βc4
mc −1
μ−αc
−kmWϕm
mc ð8Þ
βc4Ac4Wβc4−1
cm −1
μ−αc¼βm2Am2Wβm2−1
cm −ϕckcWϕc−1
cm ð9Þ
1−βm2
ðÞAm2W−βm2
mc −1
μ−αm¼1−βc4
ðÞAc4W−βc4
mc −ϕmkmWϕm−1
mc ð10Þ
where k
i
and ϕ
i
are parameters in the conversion cost function K
ij
=
k
i
X
i
ϕ
i
X
j
1−ϕ
i
, and the implied characteristic root equation has closed-
form solutions for β
c4
and β
m2
βc4¼1
2−αc−αm
σ2
Hþffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1
2−αc−αm
σ2
H
!
2
þ2μ−αm
ðÞ
σ2
H
v
u
u
tN0;ð11Þ
βm2¼1
2−αc−αm
σ2
H
−ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1
2−αc−αm
σ2
H
!
2
þ2μ−αm
ðÞ
σ2
H
v
u
u
tb0;ð12Þ
where σ
H
2
=σ
c
2
−2ρσ
c
σ
m
+σ
m
2
.Eqs.(7)–(10) can be solved numerically.
The conversion cost is an increasing function of ϕ
i
, which indicates
the relative importance of the two price levels in determining the
converting cost ϕ
i
. When ϕ
i
approaches one, the conversion cost almost
only depends on the price of the incumbent but not the substitute fuel.
That is, when ϕ
i
=1 the conversion cost is proportional to the price of
the incumbent, prevailing fuel but not the potential substitute because
of the lack of production using the latter during the transition period.
The optimal switch decisions can be illustrated in Fig. 2. The locus OA
denotes the optimal switching boundary from coal as the current fuel to
mixed fuel as the substitute, whereas the locus OB denotes that from
mixed fuel as the current fuel to coal as the substitute. If a price pair
falls into the region OAX and the incumbent fuel is coal, it is optimal
to switch to mixed fuel. Instead, if a price pair falls into the region OBY
and the incumbent fuel is mixed fuel, it is optimal to switch to coal.
Therefore, the continuance region is OAY if the incumbent is coal and
OBX if the incumbent is mixed fuel.
3. Data and variable description
All energy prices, expressed as of $/mmbtu, are of weekly frequency
and range from June 5, 2009 to April 25, 2014. Coal prices of US Central
Appalachian are used because N33% of total coal burned by power
plants in the Southeast is supplied by this region. Natural gas prices of
the Henry Hub are used because of its importance to the North
American natural gas market. Both coal and natural gas prices are ob-
tained from US Energy Information Administration (EIA, 2016). Wood
pellet prices (energy density 17 GJ/ton and free on board USsoutheast)
are from Argusmedia (2015). All prices are deflated by the Producer
Price Index (PPI) for crude material and stated in January 2013 dollars.
A transportation cost of $1.15/mmbtu, which is the average railway
cost from Central Appalachian to Atlanta, Georgia in 2013 (EIA, 2016),
Fig. 2. Switching boundaries between two fuel types for a power plant.
Table 1
Summary statistics of real energy prices in $/mmbtu.
Fuel type Whole period Infancy period Substitution period
2009–2014 2009–2011 2012–2014
Mean SD Mean SD Mean SD
Coal 4.01 0.29 4.22 0.21 3.77 0.15
Natural gas 3.54 0.88 3.61 0.76 3.45 0.99
Wood pellet 9.89 1.09 10.47 1.18 9.23 0.40
WP10 4.67 0.31 4.92 0.17 4.38 0.14
WP15 5.01 0.34 5.28 0.19 4.70 0.15
WP25 5.70 0.41 6.01 0.28 5.34 0.16
Note: Price deflator is PPI for crude material with January 2013 as the base. WP10, WP15,
and WP25 represent 10%, 15%, and 25% wood pellet cofiring with coal, respectively.
436 B. Mei, M. Wetzstein / Energy Economics 65 (2017) 434–441

is added to the real price of coal to make it comparable to wood pellet
price.
Mixed fuels are definedas 10%, 15%, and 25% of wood pellets cofiring
with coal. Their price series (X
mi
) are weighted averages of wood pellet
(X
w
) and coal (X
c
) prices, and adjusted for fuel efficiency
Xmi ¼λiwwiXwþ1−wwi
ðÞXc
½ ð13Þ
where w
wi
is the share of wood pellets in the cofiring and λ
i
is the effi-
ciency multiple defined as the ratio of coal-to-electricity efficiency
over mixed-fuel-to-electricity efficiency. The net efficiency of coal-to-
electricity is 32.67% based on the average heat rate of 10,444 btu/kWh
of a coal power plant (EIA, 2016). The efficiency loss for low level
cofiring is about 0.5% per each 10% of wood pellet input (Robinson
et al., 2003). Therefore, the efficiency multiples for 10%, 15%, and 25%
wood pellet cofiring are 1.016, 1.024, and 1.040, respectively.
Summary statistics of energy prices are reported in Table 1.Overthe
whole sample period, wood pellet hasthe highest average price and vol-
atility, and natural gas has the lowest average price but a relative mod-
erate volatility. Blending more wood pellets with coal increases the
mixed fuel cost and volatility. Note that the impact on volatility is less
than proportional, given wood pellet prices are not perfectly correlated
with coal prices. Considering the overall evolvement of the energy mar-
ket, two sub-sample periods are investigated. In the infancy period,
2009–2011, US wood pellet production was primarily consumed do-
mestically for home heating (EIA, 2016). In contrast, during the substi-
tution period, 2012–2014, wood pellet exports from the United States
to the EU increased dramatically and relatively cheap natural gas
began to substitute coal in US power plants. Energy pricesin the substi-
tution period are lower than those in the infancy period resulting from a
more intense competition of alternative fuels in the energy market. In
addition, the volatility of mixed fuel in the two sub-samples is compara-
ble or even lower than that of coal due to the low correlations between
these two price series.
Parameters in Eq. (1) are estimated and calibrated as follows.
6
Let
r
t
=dln(X
t
)=ln(X
t
)−ln(X
t−1
) be the continuously compounded re-
turn in the tth time interval, then ^
α¼r=Δþs2=2Δand ^
σ1¼s=ffiffiffiffi
Δ
p,
where rand sare the sample mean and standard deviation of the series
r
t
and Δis the equally spaced time interval measured in years (i.e., Δ=
1/52 year for weekly data). As indicated by the magnitudes of the drift
parameters (Table 2), all energy price series show a declining trend dur-
ing the whole sample and the substitution periods. In contrast, for the
infancy period, all energy price series except for wood pellets exhibit a
rising trend. Regarding the variation, coal prices have a fairly constant
volatility, whereas wood pellet prices stabilize over time and become
less volatile than coal prices duringthe substitution period. The correla-
tion coefficient estimates remain low at 0.327–0.348 and therefore
volatilities of mixed fuel prices are generally lower than those of wood
pellets and coal. Because of the portfolio effect, the correlation coeffi-
cient decreases as the percentage of wood pellets in the mixed
fuels increases. Finally, the values of parameters k's and ϕ's in the
conversion cost function are based on Adkins and Paxson (2011),
where k
c
=k
m
=0.5 means switch options are equally reciprocal and
ϕ
c
=ϕ
m
=1 means the conversion cost merely depends on the price
of the incumbent fuel. Other key variables and their adopted values
are presented in Table 3, including a discount rate of 8% for US power
plants, an average retail electricity price of $0.050/kWh net of operating
costs, an average coal price of $0.014/kWh, and average mixed fuel
prices of $0.016/kWh, $0.017/kWh, and $0.019/kWh with 10%, 15%,
and 25% wood pellets, respectively.
4. Empirical results and discussion
4.1. Base case results
Numerical solutions based on the system of four nonlinear equa-
tions, Eqs. (7)–(10), are presented in Table 4. Parameters β's and A's
are of expected signs and all price ratios are greater than one. The
total value per kWh of a coal power plant and a mixed fuel power
plant can be calculated according to Eqs. (4) and (5) for a given coal
and mixed fuel price pairs. For example, the total value of a coal
(mixed fuel with 10% wood pellets) power plant using the whole sam-
ple parameter values and average energy prices is $0.7837/kWh
($0.8052/kWh) and the option value to switch to mixed fuel with 10%
wood pellets (coal) is $0.0014/kWh ($0.0081/kWh).
The threshold price ratios W
cm
and W
mc
define the optimal switching
boundaries between coal and mixed fuel as fuel options for a power
plant for nine different scenarios. The boundaries together with histor-
ical mixed fuel and coal price ratios are plotted in Fig. 3.Inmostcases,
the price pairs fall into the switch-to-coal region, meaning that it is
noteconomicaltocofire wood pellets with coal because the mixed
fuel cost increases with the share of wood pellets. Exceptions are 10%
and 15% cofiring during the infancy period, where some portion of the
price pairs fall into the continuation region. In these two exceptions, a
power plant should continue to use whichever is its incumbent fuel.
Consequently, a cofiring power plant could have operated efficiently
during that time period. In general, Fig. 3 suggests that cofiring wood
pellets with coal is not economically feasible in the United States over
the 2009–2014 period with high wood pellet prices.
Next, we consider the impact of potential government interventions
on the renewable resource energy market (Fig. 4). First, we include a di-
rect subsidy of $1.40/mmbtu to the mixed fuel with 10% wood pellets,
which essentially cuts the input cost of a cofiring power plant. As
6
For completeness, we conducted unit root tests on the energy price series and find
mixed results for or against the null hypothesis of unit roots. An alternative stochastic
model for price series is the geometric Ornstein-Uhlenbeck process. We also estimate
the parameters for the geometric Ornstein-Uhlenbeck processand find low rates of mean
reversion and similar vo latility estimates. Therefore, we conclude that the geometric
Brownian motion well captures the short-run stochastic nature of the energy price series.
Table 2
Parameterestimates of the geometric Brownianmotion for real energy pricesin $/mmbtu.
Fuel
type
Whole period Infancy period Substitution period
2009–2014 2009–2011 2012–2014
ασρασρασρ
Coal −0.009 0.103 1.000 0.050 0.100 1.000 −0.075 0.105 1.000
Wood
pellet
−0.011 0.123 0.327 −0.001 0.149 0.348 −0.019 0.085 0.329
WP10 −0.013 0.093 0.959 0.035 0.095 0.942 −0.065 0.090 0.981
WP15 −0.014 0.091 0.915 0.029 0.096 0.886 −0.061 0.085 0.957
WP25 −0.015 0.092 0.808 0.021 0.102 0.765 −0.053 0.079 0.886
Note: WP10,WP15, and WP25 represent 10%,15%, and 25% wood pellet cofiring withcoal,
respectively.
Table 3
Description of the variables used in the regime switch analysis.
Symbol Definition Value
μDiscount rate for US power plants 0.08
k
c
Parameter of the conversion cost function from coal to mixed fuel 0.5
k
m
Parameter of the conversion cost function from mixed fuel to coal 0.5
ϕ
c
Parameter of the conversion cost function from coal to mixed fuel 1
ϕ
m
Parameter of the conversion cost function from mixed fuel to coal 1
YElectricity price net of operating costs 0.050
X
c
Coal price 0.014
X
m
Mixed fuel price (10% wood pellets) 0.016
Mixed fuel price (15% wood pellets) 0.017
Mixed fuel price (25% wood pellets) 0.019
Note: Real prices are in $/kWh.
437B. Mei, M. Wetzstein / Energy Economics 65 (2017) 434–441

such, all the energy price pairs move just under the coal-to-mixed-fuel
switch boundary. Second, we impose a tax of $1.50/mmbtu on a coal-
only power plant, which is equivalent to increasing the cost of coal.
Accordingly, all the energy price pairs fall just below the coal-to-
mixed-fuel switch boundary. In both scenarios, a coal power plant
should convert to wood pellets and coal cofiring. Therefore, a minimum
tax of $1.50/mmbtu on coal has a similar effect as a minimum subsidy of
$1.40/mmbtu on mixed fuel in triggering the investment in cofiring
power plants. Given an average cost of $0.12/kWh ($35/mmbtu) in
the United States, the mixed fuel subsidy or coal tax represents
about 4% of the electricity rate. Alternatively, a minimum subsidy of
$0.45/mmbtu on the mixed fuel or a minimum tax of $0.50/mmbtu on
the coal could maintain existing cofiring power plants in the status
quo, which represents about 1.3% of the electricity rate.
Table 4
Results of the regime switching model.
Symbol WP10 WP15 WP25
Whole period Infancy period Substitution period Whole period Infancy period Substitution period Whole period Infancy period Substitution period
2009–2014 2009–2011 2012–2014 2009–2014 2009–2011 2012–2014 2009–2014 2009–2011 2012–2014
β
c4
11.022 2.809 46.028 8.293 2.276 32.430 6.097 1.894 22.703
β
m2
−19.057 −28.428 −10.787 −13.048 −20.325 −7.448 −8.289 −12.983 −4.564
A
c4
0.372 10.684 0.0001 0.550 13.310 0.004 0.804 16.007 0.0013
W
cm
1.100 1.051 1.178 1.121 1.064 1.211 1.157 1.090 1.268
A
m2
0.040 0.002 0.190 0.088 0.005 0.359 0.205 0.018 0.750
W
mc
1.138 1.199 1.102 1.164 1.244 1.110 1.201 1.302 1.121
Note: WP10, WP15, and WP25 represent10%, 15%, and 25% wood pellet cofiring with coal.
Fig. 3. Optimalswitching boundaries for nine different wood pellet and coal cofiring scenarios for electricity generation. All prices are in $/mmbtu.
438 B. Mei, M. Wetzstein / Energy Economics 65 (2017) 434–441

Fig. 4. Impactof the subsidy on mixed fueland the tax on coal on optimal switch decisions.A subsidy of $1.40/mmbtu to the 10% mixedfuel or a coal tax of $1.50/mmbtuwould trigger the
conversions of coal-onlypower plants to cofiringones, and a subsidyof $0.45/mmbtu tothe 10% mixed fuel ora tax of $0.50/mmbtuon coal would maintainexisting cofiringpower plants
in the status quo.
Fig. 5. Sensitivity analysis on key variables in the regime switch model.
439B. Mei, M. Wetzstein / Energy Economics 65 (2017) 434–441

4.2. Sensitivity analysis
Sensitivity analysis is conducted on the discount rate, drift and vola-
tility parameters of coal price, correlation coefficient, and wood pellet
price for the whole sample period and the case of 10% wood pellet
cofiring (Fig. 5). When the discount rate is reduced from 8% to 6%, the
major impact is on the switch boundary from coal to mixed fuels. A
lower discount rate yields higher cost mixed fuels more affordable and
thus shifts the switch boundary up. When the drift parameter of coal
price is reduced from −0.009 to −0.020, both coal price and mixed
fuel price tend to fall over time, so that the wait region narrows and
both switches are more likely to occur. When the volatility parameter
of coal is lowered from 0.103 to 0.050, the volatility of mixed fuel de-
creases as well. With less uncertainty in the mixed fuel cost, a coal
power plant is more willing to switch to mixed fuel. When the correla-
tion coefficient is reduced from 0.959 to 0.800, the portfolio effect be-
comes more significant. Hence, the impact is quite similar to that of a
reduction in the volatility of mixed fuel. In summary, when values of
the key variables in the switch regime model change by a significant
amount, the optimal decision for a power plant does not change appre-
ciably. Specifically, cofiringwood pellets with coal is not an economical-
ly viable option.
At the plant level, heat rates and delivered fuel prices can deviate
from the industry averages. A lower heat rate implies higher
operation efficiency and thus reduces fuel input cost, all else equal.
This means that the price pairs in Fig. 3 fall more into the “switch
to coal”region. Similarly, the same would occur if a plant can
negotiate a lower coal delivered price. In other words, plants with
lower heat rates or better control of delivered coal prices are less
willing to switch to wood pellets cofiring and larger incentives are
needed to induce the conversion.
5. Conclusions
Using the regime switching model under the theoretic framework of
real options, we examine the optimal timing boundaries for coal and
mixed fuel astwo alternative fuels for a US power plant. Results indicate
that cofiring wood pellets with coal is not a commercially viable option
in most cases. However, lower-level (with wood pellets b15%) cofiring
could have been feasible during the infancy period when wood pellet
price is declining. Sensitivity analysis indicates that our conclusions
are robust and the most important factors are relative prices of coal
and mixed fuel. Therefore, we reject the null hypothesis that cofiring
is economically feasible and suggest using policy vehicles to stimulate
the bioenergy market and meet the GHG emission reduction target.
Specifically, a subsidy of $1.40/mmbtu to the 10% mixed fuel
7
or a
coal tax of $1.50/mmbtu would trigger the conversions of coal-only
power plants to cofiring ones, and a subsidy of $0.45/mmbtu to the
10% mixed fuel or a tax of $0.50/mmbtu on coal would maintain
existing cofiring power plants in the status quo. Given the total
electricity of 1596 billion kWh generated from coal (EIA, 2016), the
estimated government spending is about $8 billion to prompt the
conversion to mixed fuel and $2.7 billion to retain current cofiring
power plants. This spending will increase as the share of wood
pellets in the mixed fuel enlarges. These numbers are roughly
comparable to the subsidy levels of $0.03–0.20/kWh or $8.82–58.82/
mmbtu to solar energy (Fthenakis et al., 2009), and the production tax
credit of $0.019/kWh or $5.59/mmbtu toward wind energy
(Greenblatt et al., 2007). Therefore, renewable energy policies should
give equal priorities towood pellets cofiring as to solar and wind energy
in the US.
Unlike most countries in Europe, where domestic cost of
manufacturing wood pellets is not competitive with the import price,
the US Southeast has a productive forest industry and well-established
infrastructure. Hence, producing wood pellets from intensively man-
aged timberland in this region is likely to increase local employment
as well as GDP. For example, a 500-megawatt coal power plant takes
about 2540 jobs, whereas a same-sized power plant with mixed fuel
is estimated to take 3480 jobs, or a 37% increase in employment
(Strauss, 2014). In addition, a long-term demand for wood pellets can
help preserve existing working forests and attract more investments
in commercial forests, which in turn increases carbon sequestration.
Thus, the government expenditure on boosting wood pellet usage by
power plants does not simply represent a cost but has some benefits
and can potentially improve the rural economy.
Acknowledgement
The authors thank Dr. Hui Xian for her help on the data collection.
McIntire-Stennis GEOZ-MS-0173 funded part of this research.
References
Adkins, R.,Paxson, D., 2011. Reciprocal energy-switching options. J. Energy Markets 4 (1),
91–120.
Argusmedia, 2015. Argus Biomass Markets (Various Issues).
Bednyagin, D., Gnansounou, E., 2011. Real options valuation of fusion energy R&D pro-
gramme. Energy Policy 39 (1), 116–130.
Cheng, C.T., Lo, S.L., Lin, T.T., 2011. Applying real options analysis to assess cleaner energy
development strategies. Energy Policy 39 (10), 5929–5938.
Claudy, M.C., Michelsen, C., O'Driscoll, A., 2011. The diffusion of microgeneration technol-
ogies –assessing the influence of perceived product characteristics on home owners'
willingness to pay. Energy Policy 39 (3), 1459–1469.
Detert, N., Kotani, K., 2013. Real options approach to renewable energy investments in
Mongolia. Energy Policy 5 6, 136–150.
Dixit, A.K., Pindyck, R.S., 1994. Investment under Uncertainty. Princeton University Press,
Princeton, NJ.
Dwivedi, P., Bailis, R., Bush, T., Marinescu, M., 2011. Quantifying GWI of wood pellet pro-
duction in the Southern United States and its subsequent utilization for electricity
production in the Netherlands/Florida. Bioenergy Res. 4 (3), 180–192.
Dwivedi, P., Madhu, K., Robert, B., Adrian, G., 2014. Potential greenhouse gas benefits of
transatlantic wood pellet trade. Environ. Res. Lett. 9 (2), 024007.
Ehrig, R., Behrendt, F. , 2013. Co-firing of imported wood pellets –an option to efficiently
save CO
2
emissions in Europe? Energy Policy 59, 283–300.
EIA, 2016. U.S. Energy Information Administration. http://www.eia.doe.gov (last accessed
June 28, 2016).
Forisk Consulting, 2015.Wood bioenergy in the US: pelletvolumes grow; biofuel projects
stall. Forisk Res. Q. 1.
Fthenakis, V., Mason, J.E., Zweibel, K., 2009. The technical, geographical, and economic
feasibility for solar energy to supply the energy needs of the US. Energy Policy 37
(2), 387–399.
Gazheli, A., Corato, L.D., 2013. Land-use change and solarenergy production: a real option
approach. Agric. Financ. Rev. 73 (3), 507–525.
Greenblatt, J.B., Succar, S., Denkenberger, D.C., Williams, R.H., Socolow, R.H., 2007.
Baseload wind energy: modeling the competition between gas turbines and com-
pressed air energy storage for supplemental generation. Energy Policy 35 (3),
1474–1492.
Hyysalo, S., Juntunen, J.K., Freeman, S., 2013. User innovation in sustainable home energy
technologies. Energy Policy 55, 490–500.
Lee, S.C., Shih, L.H., 2010. Renewable energy policy evaluation using real option model -
the case of Taiwan. Energy Econ . 32, S67–S78.
Lima, D.,Colson, G., Karali, B., Guerrero,B., Amosson, S., Wetzstein, M., 2013. A new look at
the economic evaluation of wind energy as an alternative to electric and natural gas-
powered irrigation. J. Agric. Appl. Econ. 45 (04), 739–751.
Michelsen, C.C., Madlener, R., 2012. Homeowners' preferences for adopting innovative
residential heating systems: a discrete choice analysis for Germany. Energy Econ.
34 (5), 1271–1283.
Monjas-Barroso, M., Balibrea-Iniesta, J., 2013. Valuation of projects for power generation
with renewable energy: a comparative study based on real regulatory options. Ener-
gy Policy 55, 335–352.
Pederson, G., Zou, T., 2009. Using real options to evaluate ethanol plant expansion deci-
sions. Agric. Financ. Rev. 69 (1), 23–35.
Robinson, A.L., Rhodes, J.S., Keith, D.W., 2003. Assessment of potential carbon dioxide re-
ductions due to biomass-coal cofiring in the United States. Environ. Sci. Technol. 37
(22), 5081–5089.
Siddiqui, A., Fleten, S.E., 2010. How to proceed with competing alternative energy tech-
nologies: a real options analysis. Energy Econ. 32 (4), 817–830.
Song, F., Zhao, J., Swinton, S.M., 2011. Switching to perennial energy crops under uncer-
tainty and costly reversibility. Am. J. Agric. Econ. 93 (3), 768–783.
Spelter, H., Toth, D., 2009. North America's Wood Pellet Sector. USDA Forest Service,
Forest Products Laboratory, Madison, WI, p. 21.
7
On a hypothetical 100% carbonreduction basis, a multipleof 10 should be applied. So,
roughly speaking, the corresponding subsidy should be $14/mmbtu to trigger the switch
from coal to 100% wood pellets burning.
440 B. Mei, M. Wetzstein / Energy Economics 65 (2017) 434–441

Steininger, K., Voraberger, H., 2003. Exploiting the medium term biomass energy poten-
tials in Austria: a comparison of costs and macroeconomic impact. Environ. Resour.
Econ. 24 (4), 359–377.
Strauss, W., 2014. The many benefits of replacing coal with wood pellet fuel. Biomass
Mag. (http://biomassmagazine.com/articles/10750/the-many-benefits-of-replacing-
coal-with-wood-pellet-fuel (last accessed October 9, 2015)).
Xian, H., Colson, G., Mei, B.,Wetzstein, M.E., 2015. Co-firing coal with wood pellets forU.S.
electricity generation: a real options analysis. Energy Policy 81 (0), 106–116.
Zambujal-Oliveira, J., 2013. Investments in combined cycle natural gas-fired systems: a
real options analysis. Int. J. Electr. Power Energy Syst. 49, 1–7.
Zhang, Y., McKechnie, J., Cormier, D., Lyng, R., Mabee, W., Ogino, A., MacLean, H.L., 2010.
Life cycle emissions and cost of producing electricity from coal, natural gas, and
wood pellets in Ontario, Canada. Environ. Sci. Technol. 44 (1), 538–544.
441B. Mei, M. Wetzstein / Energy Economics 65 (2017) 434–441