Journal of Cleaner Production xxx (2014) 1e8

Contents lists available at ScienceDirect

Journal of Cleaner Production

journal homepage: www.elsevier.com/locate/jclepro

Production lot-sizing and carbon emissions under cap-and-trade and

carbon tax regulations
Ping He a, Wei Zhang a, Xiaoyan Xu a, Yiwen Bian b, *
a
School of Management, Zhejiang University, Hangzhou, Zhejiang 310058, PR China
b
SHU-UTS SILC (Sydney Institute of Language & Commerce) Business School, Shanghai University, Shanghai 201800, PR China

a r t i c l e i n f o a b s t r a c t

Article history: Cap-and-trade and carbon tax are two emission regulations widely used to curb the carbon emissions
Received 20 November 2013 generated from firms. Based on economic order quantity (EOQ) model, this paper examines the pro-
Received in revised form duction lot-sizing issues of a firm under these two regulations, respectively. The optimal lot-size and
28 August 2014
emissions under the two regulations are achieved. We then investigate the impacts of production and
Accepted 29 August 2014
Available online xxx
regulation parameters on the optimal lot-size and emissions. Furthermore, we compare the firm's
optimal carbon emissions under the two regulations. It is found that under the cap-and-trade regulation,
the firm's decisions of the optimal emissions as well as permits trading depend on the differentiated
Keywords:
Lot-sizing
permits trading prices. If setup incurs the same cost as holding incurs per unit of generated emissions,
Carbon emission regulation both regulations always lead to the same optimal emissions (which is also equal to that without emission
Economic order quantity (EOQ) regulation). Otherwise, neither regulation always leads to lower emissions than the other does.
© 2014 Elsevier Ltd. All rights reserved.

1. Introduction
There is an increasing consensus that the carbon emission
generated from firms' activities is one of the main causes of global
climate change. To curb the carbon emissions, many countries and
regions enact various regulations on firms' activities. “Cap-and-
trade (or emissions trading)” and carbon tax are two most popular
regulations implemented in the world. Under the cap-and-trade
regulation, firms initially receive a free amount of permits (“cap”)
over a planning horizon (e.g., one year), and are allowed to trade
the permits with other firms or government agencies through
special markets (e.g., carbon market). The European Union's
Emissions Trading System (EU ETS) is the first and biggest inter-
national scheme for permits trade. Up to 2010, the EU ETS covers
11,000 power stations and industrial plants in 30 countries
(European Commission, 2013), and involves over 50% of all emis-
sions in the European Union (Benjaafar et al., 2013). Advocated as
an alternative cost-effective instrument for reducing emissions,
carbon tax regulation is much easier to implement than cap-and-
trade regulation is. Under carbon tax regulation, firms are
charged for their carbon emissions at a constant tax rate level. A
growing number of scholars (Avi-yonah and Uhlmann, 2009),
politicians and economists (Inglis and Laffer, 2008) and business
* Corresponding author. Tel.: þ86 21 69980028; fax: þ86 21 69980017.
E-mail addresses: ywbian@shu.edu.cn, ywbian@gmail.com (Y. Bian).
leaders (Pontin, 2010) advocate carbon tax regimes rather than cap-
and-trade.
As we know, carbon emissions are generated in almost all ac-
tivities of firms, e.g., procurement, production, inventory holding,
order processing, transportation and some others (Hua et al., 2011;
Chen et al., 2013). Generally, carbon emissions from different ac-
tivities are generated in different ways. For example, emissions
from procurement are generated only when a procurement activity
is implemented, usually irrelevant to the procured quantity; while
emissions from inventory holding depend on the inventory quan-
tity and inventory time. In production process, if the production lot-
size is too small (which is advocated by Just-In-Time production
theory), lots of emissions are generated from frequent setups;
otherwise, if the production lot-size is too large, lots of emissions
are generated from inventory. In the presence of emission regula-
tions, emission-related costs arise in terms of buying additional
permits (under cap-and-trade regulation) or paying tax (under
carbon tax regulation). These emission-related costs can be sub-
stantial (Drake et al., 2010), which induces carbon-intensive firms
to take the emission-related costs into consideration when deter-
mining the production lot-size.
This paper addresses the issues of the production lot-sizing of a
firm under cap-and-trade and carbon tax regulations based on EOQ
model. Under each regulation, the optimal lot-size and emissions of
the firm are characterized, and the impacts of production param-
eters and regulation parameters on the optimal lot-size and

http://dx.doi.org/10.1016/j.jclepro.2014.08.102
0959-6526/© 2014 Elsevier Ltd. All rights reserved.

Please cite this article in press as: He, P., et al., Production lot-sizing and carbon emissions under cap-and-trade and carbon tax regulations,
Journal of Cleaner Production (2014), http://dx.doi.org/10.1016/j.jclepro.2014.08.102
2 P. He et al. / Journal of Cleaner Production xxx (2014) 1e8
emissions are also investigated, respectively. Due to their different
mechanisms, cap-and-trade and carbon tax regulations lead to
different forms of emissions costs, and have different impacts on
firms' operational decisions. The comparison of these two regula-
tions may provide governments the guidance on determining the
cap (under cap-and-trade regulation) or the tax rate level.
In the cap-and-trade regulation considered in this paper, the
permits buying and selling prices of the firm can be different. To our
best knowledge, most of the existing studies on operational decisions
under the cap-and-trade regulation treat these two prices as the
same. The only exception is Gong and Zhou (2013), who investigate
the impact of emission trading on a manufacturer's technology choice
and production planning by using differentiated permits buying and
selling prices. Since the emission trade takes place in a carbon market
and a firm can buy permits from or sell permits to agencies, the
permits buying price and selling price of a firm could be different.
Following Gong and Zhou (2013), we differentiate these two prices
and assume that the firm's permits buying price is not smaller than
the firm's permits selling price in the cap-and-trade regulation. The
rationale for this price differentiation is well-documented in Gong
and Zhou (2013). First, the trading prices of permits actually repre-
sent the cost and the revenue of buying and selling a unit of permits,
respectively, which include transaction costs. Transaction costs in
emissions trading can be significant and have been studied both
empirically and theoretically (e.g., Stavins, 1995; Woerdman, 2001).
Second, the bideask price spreads, often seen in various trading
markets, are another cause of non-identical selling and buying prices.
For instance, the ask and bid prices for ECX EUA (European Union
Allowances: carbon credits issued under the EU ETS to CO2-emitting
installations) futures for December 2010 are V15.48 and V14.20 per
metric ton (12:00 p.m., Aug. 14, 2010, Hong Kong Time), respectively
(http://www.ecx.eu/market-data). This implies that the firm's per-
mits buying price may be higher than the selling price in practice.
Furthermore, we hold that if the buying price is smaller than the
selling price, firms might raise profit by purely buying and selling
carbon permits. This speculation in turn weakens the effect of carbon
trading regulation on reducing firms' emissions. It is noteworthy that,
the production lot-sizing issue with identical permits buying and
selling prices is a special case of what is discussed in our paper.
The rest of this paper is organized as follows. In Section 2,
related literature is reviewed. In Sections 3 and 4, the lot-sizing
decisions under cap-and-trade and carbon tax regulations are
explored, respectively. In Section 5, the two regulations are
compared with respect to the optimal emissions. Section 6 con-
cludes this paper.
2. Literature review
In recent years, the research on cap-and-trade and carbon tax
regulations has received extensive attentions both in empirical and
theoretical studies.
The first stream mainly discusses the concepts, advantages and
disadvantages of cap-and-trade and carbon tax regulations at
strategic levels based on empirical studies. Ekins and Barker (2001)
provide a detailed survey of the literature on carbon tax and
emissions trading as well as their implementations. They conclude
that there is a general agreement that market-based instruments of
carbon control will achieve a given level of emission reductions at
lower cost. As indicated by Harrison and Smith (2009), the cap-and-
trade regulation is business-friendly and can produce more jobs.
However, carbon tax regulation is simpler and easier to implement
than cap-and-trade regulation is, and the tax increases the revenue
of government which can be used as the investment of carbon
abatement (Baranzini et al., 2000). Theoretically, both cap-and-
trade and carbon tax can achieve cost-effective emission
Please cite this article in press as: He, P., et al., Production lot-sizing and carbon emissions under cap-and-trade and carbon tax regulations,
Journal of Cleaner Production (2014), http://dx.doi.org/10.1016/j.jclepro.2014.08.102
reductions (Stavins, 2008), and there is a broad equivalence be-
tween emissions trading scheme and carbon tax regulation under
some assumptions (Pezzey, 1992; Farrow, 1995).
The second stream examines the operational decisions of firms
under emission regulations. Letmathe and Balakrishnan (2005)
study the production mix and production quantities of a firm un-
der several different environmental constraints, e.g., threshold
values, penalties and taxes, and/or emissions trading. From the
perspective of carbon abatement efficiency, Mandell (2008) shows
that utilizing the two regulations (i.e., cap-and-trade and carbon
tax) can be superior to adopting only one regulation (either cap-
and-trade or a carbon tax). Benjaafar et al. (2013) introduce a se-
ries of simple and general models to illustrate how carbon footprint
could be incorporated into operational decisions, where many ob-
servations and insights are obtained. Drake et al. (2010) study a
two-stage decision problem of a firm under the two regulations
(cap-and-trade and carbon tax). In the first stage, the firm chooses
capacities under two technologies, “dirty” and “clean”. With the
given technology, the firm in the second stage chooses production
quantities to maximize its own profit. Hoen et al. (2014) examine
the effect of different emission regulations (including voluntary
targets) on transportation mode selection for a ‘carbon-aware’
company (either by choice or enforced by regulation) under sto-
chastic demand. Jaber et al. (2013) study the coordination in a two-
level supply chain in the EU ETS, where greenhouse gas emissions
are generated in the manufacturing processes. Jin et al. (2013)
investigate the impact of carbon policies on supply chain design
and logistics of a major retailer, where three carbon policies are
considered: carbon emission tax, inflexible cap and cap-and-trade.
The third stream is related to the estimation of emission costs and
carbon accounting under carbon emission regulations. Tsai et al.
(2012a) develop a mixed Activity-Based Costing (ABC) decision
model for green airline fleet planning under emissions trading
scheme. Ståhls et al. (2011) investigate the impacts of international
commodity trade on carbon flows of forest industry in Finland, using a
quantitative analysis method. The carbon flows are embodied in the
traded forest. They show that in Finland, the direct impact of the forest
industry is only a minor fraction of the total CO2 emissions related to
production, and almost all of the emissions are caused due to pro-
duction of exports. Stechemesser and Guenther (2012) systematically
review the literature related to carbon accounting. One can refer to Tsai
et al. (2011, 2012b, 2013) and Mo  zner (2013) for other similar studies.
Close to our work, Van der Veen and Venugopal (2014) incor-
porate the cost of energy usage into EOQ model, and find that the
economic and environmental performance of a firm can be synergy
or trade-off, depending on the values of specific parameters of the
emission regulations. Hua et al. (2011) investigate how firms react in
inventory management under carbon emission regulation based on
EOQ model. They derive the optimal order quantity, and examine
the impacts of regulation parameters on the optimal decisions,
carbon emissions and total costs. However, in their work, only the
cap-and-trade regulation is discussed, and the permits buying and
selling prices are assumed to be equal. Bonney and Jaber (2011)
incorporate transportation cost and waste into EOQ model, and
develop an environmental economic order quantity model, which
results in a larger optimal ordering lot-size than that under the
standard EOQ model. Arslan and Turkay (2010) revise the standard
EOQ model by incorporating sustainability constraints. Various
sustainability constraints, such as carbon tax, cap and trade, direct
cap and carbon offset, are considered. They show that in most cases,
the optimal ordering quantity with the presence of sustainability
constraints is larger than that without the constraints. It is note-
worthy that, in the above mentioned studies, the permits buying
and selling prices are all assumed to be the same. One exception is
Chen et al. (2013). They investigate the EOQ model under various
 P. He et al. / Journal of Cleaner Production xxx (2014) 1e8 3

carbon constraints including carbon tax, carbon offset, and cap and (ecD) associated with production for the fixed demand is constant.
price (i.e. cap-and-trade). For each carbon constraint, they show Thus, we focus on the remaining part of the emissions. Let
whether and how the ordering quantity can be adjusted to decrease b ¼ E  ec D ¼ eK D=Q þ e Q =2. It is easy to verify that
E h
firms' costs. Although they mention two prices (award and penalize) pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
E  2eK eh D. By solving the equation b
b E ¼ eK D=Q þ eh Q =2, we
under cap and price constraint, they do not examine the optimal
decisions of firms when these two prices are different. have two lot-sizes in terms of a function of b
E, i.e.
Our paper differentiates itself from the existing studies in that it qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

treats the permits trading prices to be two different variables and b b 2
Q1 ¼ E E  2eK eh D eh and Q 2
assumes that the permit buying price is not smaller than the permit
selling price. Based on this assumption, some novel insights for  qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2
firms' optimal decisions can be obtained. ¼ b
Eþ b
E  2eK eh D eh :

Let f ð b
EÞ be the minimal total cost of setups and inventory
3. Cap-and-trade regulation
holding under given bE, i.e. f ð b
EÞ ¼ min fKD=Q þ hQ =2g.
eK D=Q þeh Q =2¼b
E
This section examines the optimal production lot-sizing and the By substituting Q and Q2 into f ð b
1
EÞ, we have.
corresponding optimal carbon emissions under cap-and-trade
  . qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
regulation based on EOQ model. 2
Consider a carbon-intensive firm which produces a product to
b
f E ¼ ðK=eK þ h=eh Þ E 2  jK=eK  h=eh j b
b E  2eK eh D 2:
satisfy the market demand under cap-and-trade regulation. The
firm first receives a free quantity of permits (i.e. cap C) on its
Note that if K/eK  h/eh, f ð b EÞ achieves its minimum at Q ¼ Q1;
emissions. If necessary, the firm can buy more or sell the granted b
otherwise f ð EÞ achieves its minimum at Q ¼ Q2. K/eK (h/eh) can be
permits through an outside market with unit buying price b and
interpreted as the average setup (holding) cost for each unit of
unit selling price s (s  b), respectively. Similar to Hua et al. (2011)
generated emissions. For given emissions b E, K/eK  h/eh indicates
and Chen et al. (2013), the variability of buying/selling price over
that setup incurs a lower cost than holding does for each unit of
time is not considered in this paper. The annual market demand is
generated emissions, which leads to more setups with smaller lot-
fixed at D. Once the firm starts a production run, a setup cost K and
size. In contrast, K/eK  h/eh leads to fewer setups with larger lot-
related emissions eK are incurred. The unit production cost and the
size. Particularly, if K/eK < h/eh or K/eK > h/eh, it can be easily veri-
corresponding emissions are denoted by c and ec, respectively. 2
fied that v2 f ð b
EÞ=v b
E > 0, and thus f ð b EÞ is strictly convex in bE. If K/
Assume that the production is instantaneous. Each unit of product 2
eK ¼ h/eh, it is clear that vf ð EÞ=v E > 0 and v2 f ð b
b b EÞ=v b
E ¼ 0, thus f ð b
EÞ
kept in inventory incurs an annual holding cost h and annual
is linearly increasing in b E.
emissions eh. The three types of emissions eK, ec and eh, also called b ¼ C  ec D. Based on the above process, Model (1) can be
Let C
emission intensities (Drake et al., 2010), relate the production lot-
transformed as:
size decision to the emissions cost. The objective of the firm is to
determine the optimal lot-size Q to minimize the sum of produc- n  
tion cost, setup cost, holding cost and emissions cost. TCc ¼min minb pffiffiffiffiffiffiffiffiffiffiffiffi b b TCc1 b
E ;
E 2eK eh D; E C
Denote by E the total annual emissions of the firm. For given lot-  o (2)
size Q, the annual setup times is D/Q, thus the emissions generated minb pffiffiffiffiffiffiffiffiffiffiffiffi b b TCc2 b
E ;
E 2eK eh D; E C
during setups is eKD/Q. The average inventory during a year is Q/2,
thus the emissions generated by holding inventory is ehQ/2. Besides
where TCc1 ð b
EÞ ¼ cD þ f ð b
EÞ þ b b b
E  bC and
these two parts, the total annual emissions of the firm also include
the emissions generated during the production process. Therefore, TC ð EÞ ¼ cD þ f ð b
c b
EÞ þ s b
E  sCb are the costs for the case with buying
2
the total annual emissions of the firm is E ¼ ecD þ eKD/Q þ ehQ/2. permits ð b
EC b Þ and selling permits ð b b Þ, respectively. Note that
EC
Hereafter, the corresponding emissions under the optimal lot-size b
the overall emissions are E þ ec D.
are termed as “the optimal emissions”. We next develop the optimal lot-size and the optimal emissions
Let TCc be the minimal total annual cost under cap-and-trade based on Model (2). Define the following thresholds:
regulation, which can be solved by the following model:
n o n   o n   o
TCc ¼ minQ > 0 cD þ KD=Q þ hQ =2 þ bðE  CÞþ  sðC  EÞþ ; Cdargminb f b
E þ bb
E and Cdargminb f b
E þ sb
E :
E >0 E >0

(1)
Since b  s, it is clear that C  C based on the convexity of f ð b
EÞ. Note
where cD þ KD/Q þ hQ/2 is the total cost caused by production, b c the optimal solution to Model (2).
that C ¼ C if b ¼ s. Denote by E
setups and inventory holding, and b(E  C)þ  s(C  E)þ
(uþ ¼ max(u,0)) is the emissions cost (revenue) resulted from Based on the convexity of f ð b
EÞ, we have the following result.
buying or selling permits. Lemma 1. The thresholds C and C satisfy:
The following monotonic properties of the optimal total cost can
be directly obtained from Model (1) (throughout this paper, we use pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ðK=eK þ h=eh þ 2bÞ 2eK eh D
“increasing” and “decreasing” in the non-strict sense to mean “non- C¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ; and C
decreasing” and “non-increasing”, respectively): (i) TCc is 2 ðK=eK þ bÞðh=eh þ bÞ
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
increasing in c, K, h, ec, eK, eh, respectively; and (ii) TCc is increasing ðK=eK þ h=eh þ 2sÞ 2eK eh D
in b but decreasing in s. Therefore, in the subsequent sections, we ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ;
2 ðK=eK þ sÞðh=eh þ sÞ
do not discuss the above properties, but focus on the properties of
the optimal decisions, i.e., the optimal lot-size and emissions. pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi c c c
with C  C  2eK eh D, and b E satisfies that b b < C, b
E ¼ C when C E ¼
For ease of analysis, we first solve the optimal emissions and
c
then the optimal lot-size. It is clear that the amount of emissions b when C  C
C b  C, and b E ¼ C when Cb > C.

Please cite this article in press as: He, P., et al., Production lot-sizing and carbon emissions under cap-and-trade and carbon tax regulations,
Journal of Cleaner Production (2014), http://dx.doi.org/10.1016/j.jclepro.2014.08.102
4 P. He et al. / Journal of Cleaner Production xxx (2014) 1e8

The Proof of Lemma 1 (and all subsequent results) can be found Proposition 1(i) indicates that if setup incurs the same cost as
c
in the Appendix part. Lemma 1 shows the values of b E under holding does for each unit of generated emissions, then the cap-
b
different C . In particular, the buying and selling prices (of the and-trade regulation has no direct impact on the firm's optimal
c
permits) determine the lower bound C and the upper bound C of b E , amount of emissions. The reason for this is that, the optimal lot-
respectively. A larger difference between b and s implies a wider size also minimizes the carbon emissions, so the firm should
span between C and C. keep the lot-size without engaging in carbon trade. However, if
Now, we characterize the optimal solution to Model (1). Denote setup incurs a lower or higher cost than holding does for each
c c
by Q and E the optimal lot-size and emissions under the cap-and- unit of generated emissions, in the presence of smaller cap,
trade regulation, respectively. Based on Lemma 1, we have the higher buying or selling prices, the firm will have more in-
following theorem: centives to decrease its emissions, as indicated in Proposition
1(ii).
Theorem 1. Under the cap-and-trade regulation,
The following proposition characterizes the impacts of regu-
c *
(i) when C < ec D þ C, the optimal emissions are E ¼ ec D þ C, and
c lation parameters on the optimal lot-size ðQ Þ. Denote by Q the
c pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi optimal lot-size in the p
classic EOQ model without the emissions
the optimal lot-size is Q ¼ 2ðK þ beK ÞD=ðh þ beh Þ; * ffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi
regulation, then Q ¼ 2KD=h. We have the following results
(ii) when ec D þ C  C  ec D þ C, the optimal emissions are equal
c
about the comparison of the optimal lot-sizes.
to the cap, i.e., E ¼ C, and the optimal lot-size satisfies: c *
(a) if K/eK  h/eh, then
c
Q ¼ ½ðC  ec DÞ Proposition 2. By comparing Q with Q , we find:
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ðC  ec DÞ2  2eK eh D=eh , c
(i) if K/eK ¼ h/eh, then Q is independent of C, s and b, and
(b) if K/eK  h/eh, then
c
Q ¼ ½ðC  ec DÞþ c pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi *
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Q ≡ 2eK D=eh ¼ 2KD=h ¼ Q ;
2 c * c
ðC  ec DÞ  2eK eh D=eh ; (ii) if K/eK < h/eh, then Q > Q , and Q is decreasing in C but
c increasing in s and b, respectively;
(iii) when C > ec D þ C, the optimal emissions are E ¼ ec D þ C and *
c pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi c c
(iii) if K/eK > h/eh, then Q < Q , and Q is increasing in C but
the optimal lot-size is Q ¼ 2ðK þ seK ÞD=ðh þ seh Þ.
decreasing in s and b, respectively.
c
Based on Theorem 1, we know that the emissions E are
Proposition 2(i) indicates that if setup incurs the same cost as
increasing (from ec D þ C to ec D þ C) in the cap C. Recall in Lemma 1
holding does for each unit of generated emissions, the cap-and-
that if b ¼ s, then C ¼ C, which implies that the optimal lot-size and
trade regulation has no impact on the firm's optimal lot-size,
emissions are independent of cap C. The rationale is that the im-
which is equal to the optimal solution to the classic EOQ model
pacts of emissions E under different levels of cap C on the total cost
without the emission regulation. Proposition 2(ii) and (iii) indi-
are the same when b ¼ s (the emissions trading cost can be
cate that if setup incurs a lower or higher cost than holding does
expressed as the sum of two separate linear functions of E and C,
for each unit of generated emissions, the optimal lot-size is
respectively). Note that the optimal emissions are increasing in ec,
different from that in the classic EOQ model without emission
but independent of the unit production cost c.
regulation. Note that if the cap approaches infinitely large, or the
Theorem 1 also shows the optimal decision for emissions
permits buying and selling prices approach zero, the optimal lot-
trading. It is optimal to buy C þ ec D  C unit permits when
size under the cap-and-trade regulation approaches that without
C < ec D þ C, to buy and sell nothing when ec D þ C  C  ec D þ C,
emission regulation.
and to sell C  ec D  C unit emissions when C > ec D þ C. The
We next derive the impacts of production parameters of the firm
resulted permits range (ec D þ C to ec D þ C) is ultimately the range c c
(i.e., K, h, eK and eh) on Q and E , as shown in Proposition 3. Note
of the actual amount of emissions. The determination of the two
that c has no effect on the optimal decisions.
thresholds on the firm's emissions depends on the permits
buying price and selling price, respectively. The lower threshold Proposition 3. Under the cap-and-trade regulation,
decreases in the permit buying price, and the upper threshold
decreases in the permit selling price. The larger the difference (i) the optimal lot-size is increasing in K but is decreasing in h;
between permits buying price and selling price is, the larger the (ii) if K/eK  h/eh, then the optimal emissions are increasing in K but
range between these two thresholds is. It is noteworthy that the decreasing in h;
difference between the permit buying price and selling price in (iii) if K/eK  h/eh, then the optimal emissions are decreasing in K
this paper can be regarded as the transaction cost. but increasing in h.
Particularly, if these two prices are the same, as assumed in many
Proposition 3(i) shows that under the cap-and-trade regulation,
existing studies, the firm will always buy or sell some permits unless
the setup cost has the same impact on the optimal lot-size as that in
the initial cap equals a special value, i.e. C ¼ ec D þ C ¼ ec D þ C. This
the classic EOQ model, so does the holding cost. Higher setup
means, the optimal emissions of the firm will always be a fixed
(holding) cost leads to larger (smaller) optimal lot-size. Proposition
amount, which is determined by the permits trading prices but not
3(ii, iii) imply that the optimal emissions can attain its minimum,
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
the firm's received free permits.
i.e., ec D þ 2eK eh D, when K/eK ¼ h/eh.
Based on Theorem 1, we derive the impacts of regulation pa-
c
rameters (i.e., C, s and b) on the optimal emissions E , as shown in
Proposition 1. 4. Carbon tax regulation
Proposition 1. Under the cap-and-trade regulation,
In this section, we assume that the firm operates under the
(i) if K/eK ¼ h/eh, then the optimal emissions are independent of C, s carbon tax regulation. The firm is charged for each unit emissions
c pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi with a constant tax rate level t. Based on this assumption, we derive
and b, and E ≡ec D þ 2eK eh D;
the optimal lot-size and emissions for the firm.
(ii) if K/eK s h/eh, then the optimal emissions are increasing in C,
Let TCt be the minimal total annual cost of the firm under the
but decreasing in s and b, respectively.
carbon tax regulation, which can be solved by the following model:

Please cite this article in press as: He, P., et al., Production lot-sizing and carbon emissions under cap-and-trade and carbon tax regulations,
Journal of Cleaner Production (2014), http://dx.doi.org/10.1016/j.jclepro.2014.08.102
 P. He et al. / Journal of Cleaner Production xxx (2014) 1e8
TCt ¼ minQ > 0 fcD þ KD=Q þ hQ =2 þ tEg;
where tE is emissions cost (i.e. the tax).
The following monotonic properties of the optimal total cost
can also be directly obtained from Model (3): (i) TCt is increasing
in c, K, h, ec, eK, eh, respectively; and (ii) TCt is increasing in t.
Comparing the emissions cost under Model (3) with that under
Model (1), it is clear that carbon tax regulation is equivalent to cap-
and-trade regulation with C ¼ 0 and b ¼ t, while s can be any value
smaller than b because there is no permits to be sold in carbon tax
regulation (or equivalently C ¼ 0). Thus in this section we directly give
the optimal decisions of the firm under the carbon tax regulation.
t t
Denote by Q and E the optimal lot-size and the optimal
emissions under the carbon tax regulation, respectively. By setting
C ¼ 0 and b ¼ t, the values and properties of the optimal lot-size
and emissions under the carbon tax regulation are given in Theo-
rem 2. Theorem 2 can be directly obtained based on Theorem 1 and
Propositions 1–3, and thus its proof is omitted for brevity.
Theorem 2. Under the carbon tax regulation, the optimal lot-size
and emissions satisfy:
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
t t
Q ¼ 2ðK þ teK ÞD=ðh þ teh Þ; E lower bound of E . If the tax rate level is lower than the selling
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ðK=eK þ h=eh þ 2tÞ 2eK eh D
¼ ec D þ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ; and
2 ðK=eK þ tÞðh=eh þ tÞ
(i) if K/eK ¼ h/eh, both the optimal lot-size and emissions are in-
t pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi c
dependent of t, with Q ≡ 2eK D=eh ¼ 2KD=h ¼ Q ,
t c pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
E ¼ E ≡ec D þ 2eK eh D;
(ii) if K/eK s h/eh, the optimal emissions are decreasing in t.
Particularly, if K/eK < h/eh, the optimal lot-size is increasing in t,
while if K/eK > h/eh, the optimal lot-size is decreasing in t;
(iii) the optimal lot-size is increasing in K but decreasing in h.
Particularly, if K/eK  h/eh, the optimal emissions are increasing
in K but decreasing in h; if K/eK  h/eh, the optimal emissions
are decreasing in K but increasing in h.
Theorem 2 shows the optimal lot-size and emissions under
the carbon tax regulation, and summarizes the impacts of pro-
duction and regulation parameters on the optimal decisions. By
comparing Theorem 2(iii) with Proposition 3, we find that the
impacts of production parameters (i.e. K and h) on the optimal
lot-size and emissions are the same under the both regulations.
Similar to the discussion of the cap-and-trade regulation, if setup
incurs the same cost as holding does for each unit of generated
emissions, then the firm's optimal lot-size under the carbon tax
regulation is also equal to the solution to the classic EOQ model
without the emission regulations.
Based on Lemma 1, Theorems 1 and 2, we can derive a condi-
tion under which the optimal decisions under the two regulations
are the same. Based on Lemma 1 and Theorem 2, we find that if
t t
t ¼ b, E ¼ C þ ec D; if t ¼ s, E ¼ C þ ec D; if t ¼ b ¼ s,
t
E ¼ C þ ec D ¼ C þ ec D. Hence, by further comparing Theorem 1
with Theorem 2, we know that when t ¼ b ¼ s, both the
optimal lot-size and emissions under the carbon tax regulation
are the same as those under the cap-and-trade regulation,
respectively, which are irrelevant to any cap C.
5. Comparison of the optimal emissions
Due to their different mechanisms, cap-and-trade and carbon
tax regulations may have different impacts on firms' performance,
Please cite this article in press as: He, P., et al., Production lot-sizing and carbon emissions under cap-and-trade and carbon tax regulations,
Journal of Cleaner Production (2014), http://dx.doi.org/10.1016/j.jclepro.2014.08.102
5
(3) especially on the optimal emissions. Based on Theorems 1 and 2, we
have the following results about the optimal emissions of the firm
under cap-and-trade and carbon tax regulations.
Theorem 3. Under cap-and-trade and carbon tax regulations,
(i) if K/eK ¼ h/eh, both the two regulations have no impact on the
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
t c
optimal emissions, and E ≡E ≡ec D þ 2eK eh D;
t
(ii) if K/eK s h/eh, the optimal emissions under two regulations E
c
and E satisfy:
t c
(a) if t > b, E < E for all C;
t c t t c t
(b) if s  t  b, E > E when C < E , E ¼ E when C ¼ E , and
t c
E < E when C > E ;
t
t c
(c) if t < s, E > E for all C.
Theorem 3 indicates that, the difference between the optimal
emissions of the firm under two regulations depends on not only
the regulation parameters (i.e., C, s, b, t) but also the intrinsic
production parameters of the firm (i.e., K, h, eh, eK).
Note that the buying price of permits under cap-and-trade
c
regulation determines the lower bound of b
c
E and thus of E
(see Lemma 1). If the tax rate level is higher than the buying
price of permits, the emissions should be even lower than the
c
price of permits, the emissions should be even higher than the
c
upper bound of E . If the tax rate level is between the range of
the buying and selling prices of permits, the relative magnitude
of the optimal emissions under these two regulations depends on
the initial cap C. Therefore, none of these two regulations leads to
lower emissions than the other does all the time.
To further elaborate Theorem 3, a numerical example is con-
ducted. To this end, let D ¼ 90,000, c ¼ 20, K ¼ 200, h ¼ 2, ec ¼ 2,
eK ¼ 25, eh ¼ 0.5, s ¼ 3 and b ¼ 7. The optimal emissions of the
firm with different caps (under cap-and-trade regulation) and
with different tax rate levels (under carbon tax regulation) are
depicted in Fig. 1.
The straight lines in Fig. 1 denote the optimal emissions of the
t
firm ðE Þ under the carbon tax regulation with different tax rate
levels. The polygonal line denotes the optimal emissions of the
c
firm ðE Þ under the cap-and-trade regulation. The caps corre-
sponding to the two flex points on the polygonal line are C þ ec D
and C þ ec D, respectively. Note that in this example, K/eK ¼ 200/
25 ¼ 8, and h/eh ¼ 2/0.5 ¼ 4, thus K/eK s h/eh. As shown in Fig. 1, if
t c
t ¼ 2, i.e. t  s, E is greater than E over all levels of cap C. If t ¼ 8,
t c
i.e. t  b, E is less than E over all levels of cap C. If t ¼ 5, i.e.
t c t t
s < t < b, E is greater than E when the cap C is less than E , and E
c t
is less than E when the cap C is greater than E . These results are
consistent with the conclusions in Theorem 3(ii).
Fig. 1. The optimal emissions under the two regulations.
6 P. He et al. / Journal of Cleaner Production xxx (2014) 1e8

6. Conclusions pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ðK=eK þ h=eh þ 2bÞ 2eK eh D
C¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
This paper addresses the production lot-sizing issues of a firm
2 ðK=eK þ bÞðh=eh þ bÞ
under cap-and-trade and carbon tax regulations based on EOQ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ðK=eK þ bÞ þ ðh=eh þ bÞ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
¼ 2ek eh D pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi  2eK eh D:
model. We characterize the optimal lot-size and corresponding 2 ðK=eK þ bÞðh=eh þ bÞ
(optimal) emissions under each regulation, and compare the
performance of the firm with respect to the optimal carbon Similarly, we have
emissions under the two regulations.
Our results show that, under the cap-and-trade regulation,
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
the firm may buy some permits for production, or sell some ðK=eK þ h=eh þ 2sÞ 2eK eh D
surplus permits, or buy and sell no permits at all, depending on C¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2 ðK=eK þ sÞðh=eh þ sÞ
the value of initial cap. If setup incurs the same cost as holding
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ðK=eK þ sÞ þ ðh=eh þ sÞ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
incurs per unit of generated emissions, both regulations always ¼ 2ek eh D pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi  2eK eh D:
lead to the same optimal emissions (which is also equal to that 2 ðK=eK þ sÞðh=eh þ sÞ
without emissions regulation). Otherwise, neither regulation al-
ways leads to lower emissions than the other does. Recall that C  C from the convexity of f(z). Consequently, we have
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
In this work, we only consider the lot-size and emissions C  C  2eK eh D.
decisions in one particular period, and thus assume that the Define
permits buying and selling prices do not change over time. This
modelling and analysis is appropriate for those carbon-intensive
c n   o c
firms with small or modest carbon emissions, especially for those b
E 1 dargminb pffiffiffiffiffiffiffiffiffiffiffiffi b b f b E þ bb
E and b
E2
producing single-period products such as food and electricity, E 2eK eh D; E C
n   o
which are more likely to buy or sell permits on the spot market. ¼ argminb pffiffiffiffiffiffiffiffiffiffiffiffi b b f bE þ sb
E :
Firms with large carbon emissions are likely to make permits E 2eK eh D; E C
trading actions by taking actual production into account, multi-
ple period analysis may be more appropriate. On the other hand, Since Cdargminb ff ð EÞ b þ bb
Eg and Cdargminb ff ð bEÞ þ s b
Eg,
E >0 E >0
we focus our research on a single firm's decision making and do b c
b b c
b
we have E 1 ¼ maxðC; C Þ and E 2 ¼ minð C ; CÞ based on the con-
not consider multiple competitive firms. These aspects are the
main limitations of our work, which can be further examined in vexity of f ð b
EÞ. Consequently, we have
future research. Furthermore, more research can also be done in c
b < C, b
(i) when C E 1 ¼ C and E bc ¼ C b . Since C
b is a feasible solution
the following ways. First, the decision problem can be investi- 2
c
gated based on other models, e.g. newsvendor model. Stochastic to minb p ffiffiffiffiffiffiffiffiffiffiffiffi ff ð EÞ þ b Eg, we have b
b b E ¼ C and
E 2eK eh D;b Eb
C
demand can be more realistic and may generate some other in-
TCc ¼ cD þ f ðCÞ þ bC  b Cb ¼ cD þ f ðCÞ þ bC  bC þ bec D.
sights. Second, the decision-making of the government can be c c c
incorporated. The firm and the government take part in a deci- (ii) when C  C b  C, b E1 ¼ b
E2 ¼ Cb . Hence b E ¼C b and
c
sion game, with the objective of minimizing the total cost of the TC ¼ cD þ f ðC  ec DÞ.
firm and maximizing the social welfare, respectively. b > C, b c
b and b c
b is a feasible solution
(iii) when C E1 ¼ C E 2 ¼ C. Since C
to b b
minb pffiffiffiffiffiffiffiffiffiffiffiffi b b ff ð EÞ þ s Eg, we have
E 2eK eh D; E C

Acknowledgements b ¼ cD þ f ðCÞ  sðC  ec D  CÞ.

TCc ¼ cD þ f ðCÞ þ sC  s C ▫

This research was supported by the National Natural Science

Foundation of China (nos. 71001094, 71101085, 71201153,
71371176), NSFC Major Program (nos. 71090401/71090400), and
Innovation Program of Shanghai Municipal Education Commis-
Proof of Theorem 1
sion (no. 12ZS099). The authors thank the editor and three ref-
erees for helpful comments and suggestions on the earlier
(i) and (iii) can be easily verified from Lemma 1. Details are
manuscript.
omitted for brevity. We next prove (ii). When
ec D þ C  C  ec D þ C, we have CCb  C, and thus
c
E ¼ ec D þ Eb c ¼ ec D þ C b ¼ C, and the feasible lot-sizes are
Appendices qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ððC  ec DÞ± ðC  ec DÞ2  2eK eh DÞ=eh . Substituting these lot-sizes
Proof of Lemma 1 in Model (1), we have the total costs are.

It is clear that Cdargminb ff ð b EÞ þ b b

Eg can be found by
E >0 cD þ ðK=eK þ h=eh ÞðC  ec DÞ=2±ðh=eh
computing the corresponding emissions when the lot-size is the qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
solution to the following problem.  K=eK Þ ðC  ec DÞ2  2eK eh D 2:

minQ > 0 fKD=Q þ hQ =2 þ bðeK D=Q þ eh Q =2Þg;

Consequently, if K/ek  h/eh,
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
c
where KD/Q þ hQ/2 is the sum of setup and inventory holding cost, 2
Q ¼ ððC  ec DÞ  ðC  ec DÞ  2eK eh DÞ=eh , and if K/ek  h/eh,
and bE ¼ eK D=Q þ ep
hQ =2. It is clear that theffi optimal solution for the
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi c
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
above problem is 2ðK þ beK ÞD=ðh þ beh Þ. Hence the correspond- Q ¼ ððC  ec DÞ þ ðC  ec DÞ2  2eK eh DÞ=eh .
ing emissions satisfy It can be verified that

Please cite this article in press as: He, P., et al., Production lot-sizing and carbon emissions under cap-and-trade and carbon tax regulations,
Journal of Cleaner Production (2014), http://dx.doi.org/10.1016/j.jclepro.2014.08.102
 P. He et al. / Journal of Cleaner Production xxx (2014) 1e8 7

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
c c c
ððE  ec DÞ  ðE  ec DÞ2  2eK eh DÞ=eh is decreasing in E  ec D,
C± C 2  2eK eh D ðK=eK þ h=eh þ 2bÞ±jK=eK  h=eh j qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
¼ c c
eh 2ðK=eK þ bÞ while ððE  ec DÞ þ ðE  ec DÞ2  2eK eh DÞ=eh is increasing in
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi c c
2ðK þ beK ÞD E  ec D. Proposition 1(ii) shows that E is increasing in C while it is
; decreasing in s and b, respectively. So we have that if K/eK < h/eh,
h þ beh
c
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi then Q is decreasing in C, but is increasing in s and b, respectively,
2 c
C± C  2eK eh D ðK=eK þ h=eh þ 2sÞ±jK=eK  h=eh j and if K/eK > h/eh, then Q is increasing in C, but is decreasing in s
¼
eh 2ðK=eK þ sÞ and b, respectively. ▫
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2ðK þ seK ÞD
: Proof of Proposition 3
h þ seh
We first show the monotonic properties of the optimal emis-
Hence, we have that if K/ek  h/eh,
sions because we have shown the impact of the optimal emissions
0 1
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi , qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
on the optimal lot-size in the Proof of Proposition 2.
@C  C  2eK eh DA eh ¼ 2ðK þ beK ÞD=ðh þ beh Þ
2 Denote
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ðK=eK þ h=eh þ 2vÞ 2eK eh D
RðvÞ ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2 ðK=eK þ vÞðh=eh þ vÞ
and pffiffiffiffiffiffiffi
 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2D Keh þ heK þ 2veK eh
2 ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi :
C C  2eK eh D eh ¼ 2ðK þ seK ÞD=ðh þ seh Þ; 2 ðK þ veK Þðh þ veh Þ

and if K/ek  h/eh, ffiffiffiffiffiffiffi C ¼ RðbÞ and C ¼ RðsÞ. Since

It is clearpthat
0 1 vRðvÞ 2D Kð2h þ 3veh Þ þ veK ðh þ 2veh Þ
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi , qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi veK
¼
2
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi >0
@C þ C 2  2eK eh DA eh ¼ 2ðK þ beK ÞD=ðh þ beh Þ; 2 ðK þ veK Þ3 ðh þ veh Þ

and
and
pffiffiffiffiffiffiffi
 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi vRðvÞ 2D hð2K þ 3veK Þ þ veh ðK þ 2veK Þ
2 ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi > 0;
C þ C  2eK eh D eh ¼ 2ðK þ seK ÞD=ðh þ seh Þ: veh 2
2 ðK þ veK Þðh þ veh Þ3

So wehave that if K/eK  h/eh, we have both C and C are increasing in eK and eh, respectively. Note
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
c c c 2 that both C and C are clearly independent of. ec.
Q ¼ ððE  ec DÞ  ðE  ec DÞ  2eK eh DÞ=eh , otherwise
 c qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi The impacts of K and h on C and C can be verified as follows. R(v)
c c
(K/eK  h/eh), Q ¼ ðE  ec DÞ þ ðE  ec DÞ2  2eK eh DÞ=eh . ▫ can be rewritten as.

pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi !

Proof of Proposition 1 2eK eh D K=eK þ v h=eh þ v
RðvÞ ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi þ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2 h=eh þ v K=eK þ v
If K/eK ¼ h/eh, then Lemma 1 indicates that C ¼ C ¼ 2eK eh D. pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi!
Hence, p the optimal emissions are a constant, i.e., 2eK eh D K=eK þ v
c ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ r pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ;
E ≡ec D þ 2eK eh D. So (i) holds. 2 h=eh þ v
(ii) can be proved from the following process. Theorem 1
c c
directly indicates that E is increasing in C. Specifically, E is where r(t) ¼ t þ 1/t. It is clear that r(t) is decreasing when t  1, but
increasing from ec D þ C to ec D þ C. Recall that is increasing when t  1. Consequently, we have that: if K/eK  h/eh,
b b b b R(v) is increasing in K but is decreasing in h for all v, and if K/eK  h/
Cdargminb ff ð EÞ þ b Eg and Cdargminb ff ð EÞ þ s Eg. Conse-
E >0 E >0 eh, R(v) is decreasing in decreasing in K but is increasing in h for all v.
quently, the convexity of f(z) indicates that C(C) is decreasing in C ¼ RðbÞ and C ¼ RðsÞ indicate that the cost parameters K and h
c
b(s). So we have that E is decreasing in s and b; respectively. ▫ have the same impacts on C and C.
Now, we have that (i) both C and C are increasing in eK and eh,
respectively, but are independent of ec; and (ii) if K/eK  h/eh,
Proof of Proposition 2
both C and C are increasing in K but is decreasing in h, and if K/
c pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi eK  h/eh, is decreasing in K but is increasing in h.
Recall E ≡ec D þ 2eK eh D if K/eK ¼ h/eh. Since
that
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi From Theorem 1, we have
c c c
Q ¼ ððE  ec DÞ± ðE  ec DÞ2  2eK eh DÞ=eh (from the Proof of
c pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi *
Theorem 1), we have that Q ≡ 2eK D=eh ¼ 2KD=h ¼ Q if 8
>
< ec D þ C; C < ec D þ C
K/eK ¼ h/eh. c
E ¼ C; ec D þ C  C  ec D þ C
We next prove (ii) and (iii). Define >
: ec D þ C; C > ec D þ C
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
g1 ðzÞdz  z2  2eK eh D and g2 ðzÞdz þ z2  2eK eh D: c
Hence, E is increasing in ec, C and C, respectively, for all C. The
It is clear that g1(z) d 2eKehD/g2(z). Since g2(z) is increasing in z, above impacts of parameters on C and C indicate that: (i) the
we have g1(z) is decreasing in z. This indicates that optimal emissions are increasing in ec, eK and eh, respectively,

Please cite this article in press as: He, P., et al., Production lot-sizing and carbon emissions under cap-and-trade and carbon tax regulations,
Journal of Cleaner Production (2014), http://dx.doi.org/10.1016/j.jclepro.2014.08.102
8 P. He et al. / Journal of Cleaner Production xxx (2014) 1e8

and (ii) if K/eK  h/eh, the optimal emissions are increasing in K Bonney, M., Jaber, M.Y., 2011. Environmentally responsible inventory models: non-
classical models for a non-classical era. Int. J. Prod. Econ. 133, 43e53.
but is decreasing in h, and if K/eK  h/eh, the optimal emissions
Chen, X., Benjaafar, S., Elomri, A., 2013. The carbon-constrained EOQ. Oper. Res. Lett.
are decreasing in decreasing in K but is increasing in h. Conse- 41 (2), 172e179.
quently, Proposition 3(ii, iii) hold. Drake, D., Kleindorfer, P.R., Van Wassenhove, L.N., 2010. Technology Choice and
From the Proof of Theorem 1 and Proof of Proposition 2, we Capacity Investment under Emissions Regulation (Working paper). Technology
and Operations Management, INSEAD.
have that if K/eK  h/eh, Ekins, P., Barker, T., 2001. Carbon taxes and carbon emissions trading. J. Econ. Surv.
 c qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
c c 2 15, 325e376.
Q ¼ ðE  ec DÞ  ðE  ec DÞ  2eK eh DÞ=eh is decreasing in European Commission, 2013. http://ec.europa.eu/clima/policies/ets/index_en.htm.
c Farrow, S., 1995. The dual political economy of taxes and tradable permits. Econ.
E  ec D, and if K/e  h/eh,
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi k Lett. 49, 217e220.
c c c Gong, X., Zhou, S.X., 2013. Optimal production planning with emissions trading.
Q ¼ ððE  ec DÞ  ðE  ec DÞ2  2eK eh DÞ=eh is increasing in Oper. Res. 61 (4), 908e924.
c Harrison, T., Smith, G., 2009. Cap and Trade Versus a Carbon Tax (Working paper).
E  ec D. Consequently, the impacts of K and h on the optimal
c Citizens Action Coalition of Indiana, Indianapolis, IN 46204.
emissions E indicate that the optimal lot-size is increasing in K Hoen, K.M.R., Tan, T., Fransoo, J.C., van Houtum, G.J., 2014. Effect of carbon emission
but is decreasing in. h. ▫ regulations on transport mode selection under stochastic demand. Flex. Serv.
Manuf. J. 26, 170e195.
Hua, G.W., Cheng, T.C.E., Wang, S., 2011. Managing carbon footprints in inventory
management. Int. J. Prod. Econ. 132, 178e185.
Proof of Theorem 3
Inglis, B., Laffer, A., 2008. An Emissions Plan Conservatives Could Warm To. The New
York Times (online edition).
Part (i) can be directly verified from Proposition 1(i) and Jaber, M.Y., Glock, C.H., El Saadany, A.M.A., 2013. Supply chain coordination with
Theorem 2(i). We next discuss the optimal emissions under the emissions reduction incentives. Int. J. Prod. Res. 51 (1), 69e82.
Jin, M., Granda-Marulanda, N.A., Down, I., 2013. The impact of carbon policies on
two regulations when K/ek s h/eh. Recall that the emissions under supply chain design and logistics of a major retailer. J. Clean. Prod. (available
the two regulations are online 12 September 2013).
8 Letmathe, P., Balakrishnan, N., 2005. Environmental considerations on the optimal
>
< ec D þ C; C < ec D þ C product mix. Eur. J. Oper. Res. 167, 398e412.
c Mandell, S., 2008. Optimal mix of emissions taxes and cap-and-trade. J. Environ.
E ¼ C; ec D þ C  C  ec D þ C
>
: ec D þ C; Econ. Manag. 56, 131e140.
C > ec D þ C Mo zner, Z.V., 2013. A consumption-based approach to carbon emission accounting
e sectoral differences and environmental benefits. J. Clean. Prod. 42, 83e95.
Pezzey, J., 1992. The symmetry between controlling pollution by price and con-
and trolling it by quantity. Can. J. Econ. 25, 983e991.
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Pontin, J., 2010. Q&A with Bill Gates. MIT Technology Review.
t ðK=eK þ h=eh þ 2tÞ 2eK eh D Ståhls, M., Saikku, L., Mattila, T., 2011. Impacts of international trade on carbon flows
E ¼ ec D þ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi : (A.1) of forest industry in Finland. J. Clean. Prod. 19 (16), 1842e1848.
2 ðK=eK þ tÞðh=eh þ tÞ Stavins, R.N., 1995. Transaction costs and tradeable permits. J. Environ. Econ. Manag.
29, 133e148.
Stavins, R.N., 2008. Cap-and-trade or a carbon tax? Environ. Forum 25, 16.
ff ð b
EÞ þ t b
t
It can be verified that E  ec D ¼ argminb Eg. Recall that Stechemesser, K., Guenther, E., 2012. Carbon accounting: a systematic literature
E >0
review. J. Clean. Prod. 36, 17e38.
f ðb
EÞ is strictly convex if K/eK s h/eh. Consequently, (1) if t > b, then Tsai, W.-H., Chen, H.-C., Leu, J.-D., Chang, Y.-C., Lin, T.W., 2013. A product-mix de-
t t cision model using green manufacturing technologies under activity-based
E  ec D < C, (2) if s  t  b, then C  E  ec D  C, and (3) if t < s,
costing. J. Clean. Prod. 57, 178e187.
t
E  ec D > C. Based on the above three points, part (ii) holds from Tsai, W.-H., Lee, K.-C., Lin, H.-L., Liu, J.-Y., Chou, Y.-W., Lin, S.-J., 2012a. A mixed
activity-based costing decision model for green airline fleet planning under the
Equation (A.1). ▫ constraints of the European Union emissions trading scheme. Energy 39 (1),
218e226.
Tsai, W.-H., Lin, S.-J., Liu, J.-Y., Lin, W.-R., Lee, K.-C., 2011. Incorporating life cycle
References assessments into building project decision making: an energy consumption and
CO2 emission perspective. Energy 36 (5), 3022e3029.
Arslan, M.C., Turkay, M., 2010. EOQ Revisited with Sustainability Considerations Tsai, W.-H., Shen, Y.-S., Lee, P.-L., Chen, H.-C., Kuo, L., Huang, C.-C., 2012b. Integrating
(Working paper). Koç University, Istanbul, Turkey. information about the cost of carbon through activity-based costing. J. Clean.
Avi-yonah, R.S., Uhlmann, D.M., 2009. Combating Global Climate Change: Why a Prod. 36, 102e111.
Carbon Tax is Better Response to Global Warming than Cap and Trade (Working Van der Veen, J.A.A., Venugopal, V., 2014. Economic and environmental perfor-
paper). University of Michigan Law School. mance of the firm: synergy or trade-off? Insights from the EOQ model. In:
Baranzini, A., Goldemberg, J., Speck, S., 2000. A future for carbon taxes. Ecol. Econ. Choi, T.-M. (Ed.), Handbook of EOQ Inventory Problems. Int. Ser. Oper. Res.
32, 395e412. Manag. Sci. 197, 121e137.
Benjaafar, S., Li, Y., Daskin, M., 2013. Carbon footprint and the management of Woerdman, E., 2001. Emissions trading and transaction cost: analyzing the flaws in
supply chains: insights from simple models. IEEE Trans. Autom. Sci. Eng. 10 (1), the discussion. Ecol. Econ. 38, 293e304.
99e116.

Please cite this article in press as: He, P., et al., Production lot-sizing and carbon emissions under cap-and-trade and carbon tax regulations,
Journal of Cleaner Production (2014), http://dx.doi.org/10.1016/j.jclepro.2014.08.102