Tourism Management 45 (2014) 16e27

Contents lists available at ScienceDirect

Tourism Management
journal homepage: www.elsevier.com/locate/tourman

A framework to account for the tourism carbon footprint at island

destinations
Ya-Yen Sun
Department of Transportation & Communication Management Science, National Cheng Kung University, No.1, University Road, Tainan City, 70101, Taiwan

h i g h l i g h t s

A framework is provided to measure the national tourism carbon footprint.

It is based on the Tourism Satellite Account and the Environmentally Extended InputeOutput model.

Islands may incur half of the tourism carbon footprint outside their territory.

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

Article history: Given concerns over greenhouse gases and the role of tourism in generating such environmental ex-
Received 19 April 2013 ternality, a consistent carbon measurement framework is needed. This paper combines principles
Accepted 31 March 2014 derived from production and consumption accounting measures to better allocate the responsibility for
carbon emissions. Utilizing a boundary that includes domestic tourism expenditure, inbound tourism
Keywords: expenditure, and local spending associated with outbound travel, this paper (a) proposes a framework to
Tourism
measure the domestic total carbon effect and foreign-sourced effect, and (b) applies the analytical
Carbon emission
framework to Taiwan. The empirical study indicates that the carbon emissions for domestic tourism
Environmental inputeoutput model
Tourism satellite account
industries, international aviation, and imports accounted for 47%, 28% and 25% of the tourism carbon
Taiwan footprint. It is suggested that an island’s dependence on both aviation and international trade leads to a
larger share of emissions outside their geographic territory with respect to tourism development.
Ó 2014 Elsevier Ltd. All rights reserved.

1. Introduction the consumption accounting principle (CAP) (Munksgaard &

Evaluation of the tourism carbon footprint has gained prominent
attention in recent years as a means to account for the environ-
mental externality of tourism growth. Empirical applications range
from the national analysis (Becken & Patterson, 2006; Dwyer,
Forsyth, Spurr, & Hoque, 2010; Patterson & McDonald, 2004), to
regional scales (Kelly & Williams, 2007; Konan & Chan, 2010;
Whittlesea & Owen, 2012), and from targeting a single industry
(Becken, 2002) to individual tourism events (Hanandeh, 2013; Jones,
2008). These empirical studies encompass different research scope,
and raise questions about which components of tourism demand
and operations should be addressed when assessing the full scale of
tourism carbon emissions (Munday, Turner, & Jones, 2013).
The most common debate for the carbon research rests on the
philosophical question of whether to designate carbon re-
sponsibility using the production accounting principle (PAP) or
Pedersen, 2001; Peters, 2008; Turner, Munday, McGregor, &
Swales, 2012). The first philosophy, PAP, argues that a region is
liable for all the carbon emissions in its local production for which
products are sold domestically or exported. This principle is also
embraced by the Kyoto Protocol (KP)1 for determining emission
reduction targets among developed nations. For the national
tourism carbon estimation, the PAP concept includes local emis-
sions that are associated with internal tourism (domestic and
inbound tourism) and transactions related to outbound tourism
within the geographic boundary of the departure country. This
approach, however, excludes all imported products consumed by
visitors or the imported intermediated goods used by the tourism
industry directly or indirectly. Furthermore, PAP does not incor-
porate carbon emissions associated with residents traveling to
and at foreign destinations.

1
The first commitment period of the Kyoto Protocol started in 2008 and ended in
E-mail address: yysun@ncku.edu.tw. 2012, and the second commitment period is from 2013 to 2020 (UNFCCC, 2013).

http://dx.doi.org/10.1016/j.tourman.2014.03.015
0261-5177/Ó 2014 Elsevier Ltd. All rights reserved.
 Y.-Y. Sun / Tourism Management 45 (2014) 16e27
On the other hand, the CAP philosophy attributes carbon
emissions to the end consumers for all the products and services
consumed, regardless of product origins. Under CAP, national
tourism carbon emissions cover domestic tourism consumption
and outbound tourism consumption by residents for all products at
tourist-generating regions as well as at tourist-hosting regions.
Inbound tourism is considered an export, so CAP excludes those
emissions. The current tourism carbon studies generally fall be-
tween these two extremes of carbon accounting principles
(Munday et al., 2013).
Besides CAP versus PAP approach, another common debate is
whether to include international aviation and marine trans-
portation in the tourism greenhouse gas (GHG) emission account-
ing framework (Dwyer et al., 2010). The standard national CO2
accounting procedures proposed by the Intergovernmental Panel
on Climate Change (IPCC) exclude energy uses from international
transportation when compiling National Emission Inventories (NEI)
(IPCC., 1997). International emissions are excluded because it is
difficult to assign jurisdictions to cross-border emissions in a
manner that is consistent with the Kyoto Protocol, which assigns
responsibility using territorial boundaries (Bureau of Energy, 2011).
Assigning international aviation emissions in the context of
tourism is complex. The territorial or KP perspective would
completely ignore emissions associated with international bunkers
to be in line with the IPCC suggestions. The PAP approach takes a
different perspective by including all the emissions produced by
national carriers for its inbound, outbound and stop-over services.
The CAP approach, on the other hand, traces the emissions of res-
idents’ outbound travel to a specific country. The CAP output
comprises air pollution emitted from domestic and foreign-
registered airlines by a share contributed by its own residents.
Treatment of these two above-mentioned issues has important
implications for measuring the tourism carbon emissions at island
destinations. This type of economy is generally located in isolated
area, with a small industrialization scale, limited natural resources
and a relatively small population on the island (McElroy & Parry,
2010). These natural characteristics dictate aviation as the pri-
mary method for cross-border movement of people and goods, and
require island-nations to engage in large-scale international trade
for critical supplies of energy and products. It is also probable that
the inbound visitor volume will be disproportionately higher than
domestic and outbound travel. The decision to include the over-
whelming inbound consumption by the PAP concept versus to ac-
count for the outbound spending in the CAP concept is itself a
challenge in tourism carbon calibration.
The purpose of this study is first to provide a calibration
framework for a national tourism CO2 account. We argue that a
country’s tourism carbon footprint (CF) should be addressed in line
with the Tourism Satellite Account (TSA) concept. In essence, we
attempt to estimate the energy use and carbon emissions that
support the tourism economic activities reported within a
geographic boundary. The framework equally addresses the
following question: Without these tourism economic activities re-
ported in the TSA, how much will the carbon footprint be reduced
domestically and globally?
The second purpose is to present a case study, Taiwan, under the
proposed framework to demonstrate the pattern of tourism carbon
emissions for an island destination. The carbon footprint of Tai-
wan’s tourism consumption is calibrated using the Tourism Satel-
lite Account (TSA) and the Environmentally Extended Inpute
Output (EEIO) model, which links to 16 different energy sources
employed in Taiwan. This rich dataset allows us to estimate a more
complete and accurate energy use and carbon emissions pattern,
instead of using a general greenhouse gas coefficient for tourism
industries. Emissions associated with imported products and air
17
transportation will be emphasized to contrast the differences when
these aspects are not addressed.
This paper is structured as follows: the literature review section
presents recent tourism carbon studies with a detailed comparison
of their analytical processes and research scopes. This information
helps to identify a general pattern that was employed by the pre-
vious tourism carbon applications. The third section presents the
recommended accounting boundary and framework for calibrating
the national tourism carbon footprint. This is followed by the case
study of Taiwan for 2007, the latest year for which detailed industry
GHG emissions and TSA data are available. The final sections
elaborate the empirical results and conclude with discussion.
2. Literature
The literature section first provides information on major arti-
cles related to GHG estimations in tourism applications. Based on
our observations, the tourism carbon study is a relatively new
research area, so the publications referenced in this review are
generally published after 2006. We limit our focus on destination-
oriented GHG studies where Table 1 presents the national tourism
research and Table 2 addresses the regional cases. Papers that
discuss individual tourism events, such as Jones (2008), Collins,
Flynn, Munday, and Roberts (2007), and Andersson & Lundberg
(2013) or single-industries [the aviation sector in Becken (2002)]
are not included in this review. It is important to note that the
carbon estimations in Tables 1and 2 are not directly comparable
across studies because their analytical approaches, assumptions,
timeframes and the research scopes are different, yielding an
inconsistent basis for comparison. Next, we address the analysis
method and the research scope that each study employed.
2.1. Analysis method
There are two main approaches for accounting for CO2 emis-
sions from tourism: a bottom-up analysis involving tourism end-
user behaviors and energy use, and a top-down analysis using
environmental accounting and the Tourism Satellite Account,
which were both first applied by Becken and Patterson (2006).2 The
bottom-up analysis computes energy use and GHG emissions based
on information related to energy end-uses for typical tourism in-
dustries and tourist behavior. In the example of New Zealand,
Becken and Patterson (2006) first sampled transportation, accom-
modation and attraction businesses to calibrate the average energy
efficiency and coefficients with respect to per dollar sales (industry
analysis). They then combined these results with tourist travel
behavior (transportation mode, accommodation type or recrea-
tional activity) and visitor volume (tourist analysis) to estimate
total energy use in the tourism sector.
Using this bottom-up approach, detailed energy information can
be gathered through business surveys to reflect the regional differ-
ences in the production function and carbon intensity by detailed
sub-sectors. For example, transportation can be differentiated by
domestic air, private air, rental car, coach, train, motorcycle, charter
bus, or ferry, to name a few, depending on the transportation modes
that are best utilized in the area. Supplemented with visitor surveys,
tourists can then be segmented into coach tourists, visiting friends
or relatives (VFR), auto tourists, backpackers, or campers, for
2
The other approaches to address the environmental impact of tourism devel-
opment are the ecological footprint analysis, which measure the area required to
support a certain type of development (Gössling, Hansson, Hörstmeier, & Saggel,
2002; Hunter & Shaw, 2007), and life cycle assessment (Filimonau, Dickinson,
Robbins, & Reddy, 2011).
18 Y.-Y. Sun / Tourism Management 45 (2014) 16e27

Table 1
Comparison of tourism GHG emission studies e national level.

Articles Patterson and McDonald (2004) Becken and Patterson (2006) Dwyer et al. (2010) de Bruijn, Dirven, Eijgelaar, &
Peeters (2013)

Destination New Zealand New Zealand Australia Netherlands

Reference Year
Environmental variables
Analysis approach
Scope: visitor segments
included
Scope: type of consumption
covered
Scope: level of impact included
Result
Ranking/contribution
a
The study uses produced-based estimation and consumption-based estimation where the first method is similar to the top-down approach while the second is comparable
to the bottom-up approach.
b
It includes emissions associated with expenditures in Australia by foreign tourists, domestic tourists, and outbound Australian residents within Australia prior to or
following flights.
1997/1998
Land, water, energy, water
discharges, NOx, CH4, CO2
Top-down approach
Domestic and inbound tourists
Expenditures in tourism-
characteristics, tourism-
connected industries, and
aviation
Direct and indirect impact (not
include imports)
2.7 MT of CO2 or 6.8 MT of CO2 if
including international travel
The tourism ranked 17th worst
of the 25 sectors in CO2
emission or 22nd worst if
including overseas travel.
1997/98
CO2, energy
Top-down and bottom-up
approach
Domestic and inbound tourists
Expenditures in tourism-
characteristic industries, and
domestic aviation, and
international aviation
Direct impact
47.36 PJ of energy and 2689 KT
of CO2
The tourism ranked 17th worst
of the 26 sectors in CO2
emission.
2003e2004 2002e2012
CO2e CO2
Top-down and bottom-up Bottom-up approach
approacha
Domestic, inbound tourists, and Domestic and outbound
outbound touristsb tourists
Expenditures in tourism- Transport (include
characteristics, tourism- international aviation), activity
connected industries, and and accommodation
international aviation
Direct and indirect impact Direct impact
(include imports)
54.4e61.5 MT of CO2e 15.4 Mt of CO2
The broad tourism ranked 5th NA
worst of the 29 sectors,
accounting for
3.9%w5.3% of total industry
GHG in Australia.

example, in order to cross-examine their travel behavior and
transportation choices. This detailed information enables re-
searchers to identify changes in GHG over time because visitors may
switch their transport modes, length of stay, or type of accommo-
dations, leading to a distinct energy-use coefficient by year. Some
bottom-up examples can be found for the city of Whistler, British
Columbia, Canada (Kelly & Williams, 2007), Penghu island, Taiwan
(Kuo & Chen, 2009), and Hawaii (Konan & Chan, 2010).
In general, this bottom-up approach is best suited for small re-
gions because the investment of data collection at the industry level
is manageable and because visitor surveys can be customized to
reflect greater levels of detail. The only exception to taking the
bottom-up approach to the national level is provided by NHTV
University, Netherlands (de Bruijn et al., 2013). This project sur-
veyed Dutch residents for information regarding their accommo-
dation type, transportation mode, length of stay, flight distance and
travel format from 2002 to 2012. The commitment to collect this
long-term nation-wide information is quite rare, but it provides
valuable insights for evaluating CF for changing travel volume and
behavior (Peeters & Landré, 2012).
The second approach, top-down analysis which is referred to as
Integrated Economic-Environmental Accounting, allows the
assessment of tourism as a sector within a comprehensive national
economic platform. The fundamental idea of the top-down
approach is that tourism does not occur within the framework of a
single acknowledged industrial sector, but rather is determined by a
combination of items and services that tourists purchase (Fletcher,
1989; Smith, 1994). To account for information that is embedded
in multiple industries, this approach combines the Tourism Satellite
Account (TSA) and Environmentally Extended InputeOutput model
(EEIO) to match up the supply and demand information in the
tourism production and consumption process. In the TSA, sectors
can be defined as “tourism characteristic”, “tourism connected”, and
“tourism non-related”, as determined by a tourism industry ratio
that indicates the proportion of total sales, total energy use and total
GHG emissions contributed by tourist consumption in one industry.
With the tourism expenditure data supplied by TSA on hand, the
EEIO model then steps in to quantify the direct, indirect or induced
effects of GHG emissions resulting from tourism consumption
(Becken & Patterson, 2006; Dwyer et al., 2010; Jones & Munday,
2007; Konan & Chan, 2010; Minx et al., 2009). Collins, Jones, and
Munday (2009) documented several advantages of EEIO analysis
in tourism evaluations; one of those advantages is that the model
gives detailed results by industry, useful in formulating strategies
and prioritizing policy. In addition, the EEIO methodology is
transparent, widely accepted and applicable to various tourism
scenarios, for either destinations or for event-specific applications.
The model can be easily expanded and incorporated with multiple
environmental indicators on ecological resources or waste. Minx
et al. (2009) conducted an extensive review of EEIO and its
related applications and further advocated this framework for its
comparability across regions, countries and continents, relative to
other bottom-up resource flow methodologies.3
Overall, the top-down analysis is best suited for comparing
tourism’s eco-efficiency with other sectors or for formulating
macroeconomic instruments, such as carbon taxes. Different from
the bottom-up approach, the top-down analysis is based on sec-
ondary data, including an InputeOutput table, the Tourism Satellite
Account, and ecological resources/waste emission data by industry.
This type of information is generally available at the national level
unless the regional government has a great investment in data
compliance. Empirical examples of this approach include the GHG
estimates for New Zealand (Becken & Patterson, 2006; Patterson &
McDonald, 2004), Australia (Dwyer et al., 2010) and Wales, UK
(Jones & Munday, 2007; Munday et al., 2013).
2.2. Research scope
Besides the different analysis methods, the research scope of
tourism carbon studies varies due to three factors: type of visitor
3
The EEIO model has its own disadvantages, which are related to the assump-
tions employed by the IO model, such as linearity, constant price and no capacity
constraints. See Dwyer, Forsyth, and Spurr (2006, p. 324) for a good comparison of
assumptions used in a standard IeO model and in a CGE model.
Table 2
Comparison of tourism GHG emission studies - regional level.

Articles Jones and Munday Kelly and Williams Kuo and Chen (2009) Konan and Chan (2010) Hoque et al. (2010) Farreny et al. (2011) Whittlesea and Owen Li, Zhang, Chen,
(2007) (2007) (2012) Zhou, and Wang
(2012)

Destination Wales, UKa Whistler, Canada Penghu island, Taiwanb Hawaii, USAc Queensland, Australia Antarctic South West England HuangShan
National Park,
China
Reference Year 2000 2000 2006 1997 2003e04 2008e09 2006 2006e2009
Environmental CO2, waste outputs CO2e for energy and the 4 GHG gases, energy, Seven fuel types and CO2e CO2 CO2e CO2
variables disposal of solid waste solid waste, water three GHG gases (CO2,
usage, and wastewater Methane and NOx)

Y.-Y. Sun / Tourism Management 45 (2014) 16e27

discharge
Analysis approach Top-down approach Bottom-up approach Bottom-up approach Top-down approach Top-down approach Bottom-up approach Top-down approach Top-down
and bottom-up approach
apporachd
Scope: visitor Visitors Visitors Visitors Visitors Visitors Visitors Residents and visitors Visitors
segments
included
Scope: type of Expenditure in Residential, Transportation, Residents and visitors Expenditures in Flight and vessel fuel Accommodation, food, NA
consumption Tourism-characteristic transportation, accommodation and consumption tourism-characteristics consumption travel, shopping,
covered industries (include commercial and recreation activity and tourism-connected activities, attractions,
aviation) municipal buildings (include aviation) industries events, services
(include imports and
aviation)
Scope: level of Direct and indirect Direct and indirect Direct impact Direct and indirect Direct and indirect Direct impact Direct and indirect Direct impact
impact included impact impact impact impact impact
Result 1462.5 of CO2 and 2.9 million GJ of energy 0.737 GJ of energy and 5.17 MT of GHG or 12.01e13.86 MT of 198,843 ton of CO2 12.3 MT of CO2e 3456.76 hm5
912.5 KT of waste. and 131 KT of CO2. 0.05 MT of CO2. 2.42 MT without air CO2e
transportation.
Ranking/ The tourism ranked Tourism accounted NA Tourism accounted for Queensland’s carbon NA NA NA
contribution 12th worst of the 28 more than 65% of over 22% of total GHG footprint is 22% of total
sectors based on per energy and GHG, or 90% emissions in Hawaii. direct and indirect GHG
million gross value of total energy and GHG emissions by Australia
added. if external travel was tourism.
included.
a
Similar study for Wales UK can be found in Munday et al. (2013).
b
Similar study approach for PengHu can be found in Kuo, Lin, Chen, and Chen (2012).
c
An older study of Hawaii energy and tourism can be found in Tabatchnaia-Tamirisa, Loke, Leung, and Tucker (1997).
d
The study uses production-based estimation and consumption-based estimation where the first method is similar to the top-down approach while the second is comparable to the bottom-up approach.

19
20 Y.-Y. Sun / Tourism Management 45 (2014) 16e27
segments that are evaluated, type of tourism consumption included
in GHG emissions, and type of economic effects that are considered.
2.2.1. Type of visitor segment evaluated
The first factor, visitor segment, debates whether to include the
consumption of domestic tourism, inbound tourism or outbound
tourism within the CF assessment. For national CF studies, domestic
and inbound visitors are typically included. This approach is more
in consistent with PAP because domestic and inbound visitors
consume tourism products that are generally provided by local
entrepreneurs.
On the other hand, quite rarely does a study adopt the principle
of CAP to calibrate the energy uses associated with its residences in
domestic and outbound travel. The Netherlands “Travel Large in
2012” is the only application that concerned the direct carbon
emissions of transportation, accommodation and recreational ac-
tivities for their own residents (de Bruijn et al., 2013). The statistics
indicated that 18% of all Dutch holiday emissions were produced by
domestic travel and 82% by outbound travel in 2012. Meanwhile its
eco-efficiency was worsening over time due to the strong increase
in long-haul trips. By applying the CAP approach, a country accepts
the philosophy of taking responsibility for their residents travel
behavior and offers the possibility to design policies on the sus-
tainable options of outbound travel.
2.2.2. Type of tourist consumption included in GHG emissions
Each carbon study covers various components of tourism con-
sumption/activities and their emissions during the travel process.
The first world-wide tourism GHG emissions report estimated that
tourism consumption accounted for around 5% of the global CO2
emissions in 2005 (WTO-UNEP-WMO, 2008). The scope of that
estimate covers only five tourism services: air transport, car, other
transport, accommodation and recreational activities. In Becken
and Patterson’s (2006) and Jones & Munday’s (2007) studies, only
spending in tourism-characteristic industries in the TSA are
included. These industries are 1) accommodation and catering; 2)
road, rail and water passenger transport; 3) air transport, other
transport and transport services; 4) equipment hiring (e.g., car
rentals and recreational equipment rentals); and 5) cultural and
recreational services. Their scope leaves out “tourism-connected
industries.” Dwyer et al. (2010), on the other hand, included
spending in six tourism-characteristic industries (travel agencies,
taxi transport, air and water transport, motor vehicle rental, ac-
commodations and cafes/restaurants), and 13 tourism-related in-
dustries (such as rail transport, retail trade, casinos, museums) in
their analysis, providing a more extensive accounting of tourism
consumption across all sectors.
In general, the energy uses associated with regional trans-
portation (without border-crossing), accommodation and recrea-
tional activities are the most common items included in tourism CF
research. In addition to these basic categories, international avia-
tion, imported materials, and shopping items have begun to receive
consideration in selected cases. Due to high energy use and strong
radiative forcing, air transportation arriving to and departing from
destinations is the major driver of total tourism carbon emissions
(Peeters, Gossling, & Becken, 2006). Cross-board aviation accounts
for as much as 25% of direct tourism carbon emission and, in some
instances, is projected to increase to 86% of total emissions for
certain countries by 2050 (Dubois & Ceron, 2006). Although the
importance of international flights is greatly recognized, aviation
emissions are handled differently case by case. Some may only take
into account energy use by national carriers, while others may look
into all flights by inbound visitors to the country (Dwyer et al.,
2010). No consensus has been reached for dealing with the
tourism aviation. Even so, it is stressed that this critical component
should be included in the national accounting framework so that
adequate tourism carbon efficiency can be established and
compared across industries (Becken & Patterson, 2006; de Bruijn
et al., 2013; Dwyer et al., 2010).
Besides aviation, Dwyer et al. (2010) further addressed the en-
ergy consumption associated with imports required in domestic
production processes, and indicated that imports accounted for 18%
of total tourism CF in the base years of 2003e2004 for Australia. In
Whittlesea and Owen (2012), the CF for tourist shopping was
addressed explicitly, making a good comparison when souvenirs
are either produced locally or imported in Southwest UK. The
“shopping” component has received increasing attention in recent
studies for two reasons. First, shopping expenses account for a large
share of final demand injection for certain destinations and visitor
segments. This pattern has been especially dominant with the
influx of Chinese visitors globally, who are well-known for their
shopping sprees (Sun & Pratt, 2014; Wang & Davidson, 2010).
Secondly, shopping is a general term referring to all tangible
products that are purchased by tourists in a region. The CF of this
specific type of consumption various by areas, and is dependent on
the products that are available in the local retail stores. For
example, a t-shirt would yield a quite different CF per dollar than an
electronic product. Borrowing the general energy intensity figures
from other studies, such as 6.9 MJ/visitor (from Becken and
Patterson, 2006), to substitute the carbon emission pattern locally
would admit a large error margin, thus requiring region-specific
study to provide a better picture.
2.2.3. Type of economic effects considered
The third component of differences in tourism carbon studies is
the types of economic activities that are considered: direct, indirect
or induced effects. Case studies equipped with the inputeoutput
analysis framework are able to calibrate the carbon emissions
associated with indirect or induced effects simultaneously. From
the empirical results shown in Tables 1and 2, it can be seen that the
scale of indirect CF effect was between 40% and 110% of the direct
effect, depending on the geographic boundary. The larger the
geographic area that is estimated, the greater the impact of indirect
economic activities. Thus, leaving out the indirect CO2 emission
would substantially underestimate the total carbon footprint
generated by the tourism consumption. Most studies delimited
their efforts on the direct and indirect effects and few included the
induced effects (type II multipliers) in accounting for tourism CF.
The inclusion of household income from employment bears the risk
of overestimation on CF. The linearity of the IO model which as-
sumes employee income will be totally re-spent and re-circulate
throughout the local economy is challenged in empirical cases
due to the savings factor, wage re-spending outside the region, and
the non-linear income to sales ratio with respect to additional
sales4 (Sun, 2007).
3. National tourism CO2 accounting framework
The first objective of this study is to provide a comprehensive
framework for GHG emissions associated with tourism activities
and consumption for a nation. We support the top-down approach
based on the Tourism Satellite Account for its consumption scope
and the calculation approach of the Environmentally Extended
4
Some authors may argue that human consumption is the driving force of the
environmental problems and the inclusion of included effect (adopting type II
multipliers) would help to reveal the scale of human consumption and energy use.
See Turner, Munday, McIntyre, and Jensen (2011) for pointing out this perspective.
 Y.-Y. Sun / Tourism Management 45 (2014) 16e27 21

Fig. 1. Research scope and the GHG output.

InputeOutput Model, arguing that it provides the best platform to
address the tourism carbon footprint.
The system boundary of tourism consumption. Different from CPA
or APA, we argue that the type of expenditures addressed in
tourism carbon footprint analysis should be in line with the tourism
consumption figure reported in a standard Tourism Satellite Ac-
count. TSA is a systematic and internationally accepted approach to
document the scale and importance of tourism consumption at a
given region (Frechtling, 2010). It bridges the type of visitor ex-
penditures with the supply of tourism products, providing an es-
timate that is consistent under the Systems of National Account
(United Nations, 2010). A standard TSA incorporates domestic
tourism consumption, inbound tourism consumption at the host
nation, including airfares paid to the host-nation airlines, and do-
mestic consumption associated with outbound travel and airfare
paid to the national carriers (Fig. 1). TSA however does not capture
economic activities at foreign destinations from outbound travel
nor airfares paid to foreign-based airlines.
The system boundary of the supply chains. The adoption of the
Environmentally Extended InputeOutput model then fulfills the
purpose to quantify the direct and secondary business transactions
across the supply chains. It traces the movement of products and
services across regions and industries. Pairing the top-down TSA
method with the EEIO model results in a framework that provides a
carbon footprint picture that corresponds to the amount of GHG
emissions, domestically and globally, that will be reduced if tourism
demand or supply ceases to operate within the geographic territory.
It should be noted that, under this framework, energy use by
national carriers of aviation and marine transportation is counted
for by a proportion, which is computed as the tourism sales of in-
bound and outbound travel to the total annual transportation sales.
Table 3
GHG Components by the production accounting principle (PAP), consumption ac-
counting principle (CAP), and the suggested approach (TSA þ EEIO).
Visitor type GHG emissions for PAP CAP
Domestic Domestic product V V
tourist Imported product V V
Domestic aviation V V
Inbound Domestic product V V
tourist Imported product
Aviation by national carriers V V
Aviation by foreign carriers
Outbound Domestic product V V
tourist Imported product to the country V V
Products at foreign country V
Aviation by national carriers V V
Aviation by foreign carriers V
Operation of foreign carriers is not included in the TSA. They are
assumed to leave no economic impacts at this destination thus this
country holds no responsibility for their operation.
This framework is referred as “TSA þ EEIO” where the TSA
stands for the system boundary in tourism consumption while the
EEIO points out the scope to measure direct and indirect carbon
emissions for the tourism development. “TSA þ EEIO” falls between
the two extremes of the production accounting principle (PAP) and
the consumption accounting principle (CAP) proposed by
Munksgaard and Pedersen (2001). Including GHG emissions for
domestic tourism industries and the national carriers (output 1 and
output 2) reflects the philosophy of PAP, and emphasizes the re-
sponsibility of domestic producers to sustain the given volume of
tourism demand. Specifically, emission of international aviation is
allocated to the destinations for its share at national carriers. The
underlying logic for such a treatment is that the economic growth
benefits of those transactions accrue to the destinations so that the
host nation bears the responsibility to mitigate and control the
national carriers’ emissions.
Indirect environmental pressure from imported products
serving tourism (output 3) however is the exercise of CAP, attrib-
uting the emissions of imports to the end users (domestic, inbound
and outbound tourists) at the evaluated destinations. The argument
to include this component is that imported products are essential
for tourism industry operation. Some of these imports, such as
energy and natural resources, cannot be replaced by domestic
supplies. Necessity of these imported resources makes emissions
associated with these products an integral part of the total tourism
CF. The comparison of PAP, CAP, and the proposed approach
(TSA þ EEIO) is documented in Table 3.
3.1. Operation of the Environmentally Extended InputeOutput
model
In this study, the carbon footprint is differentiated by emission
TSA þ EEIO source: domestic carbon emissions, emissions associated with in-
V ternational transportation by the national carriers, and emissions
embedded in imports. For the first two sources, the direct and in-
V
direct domestic requirement of energy and the associated GHG
V emissions given one dollar change in final demand are calculated as
follows (Miller & Blair, 2009):
V R ¼ R1 ðI  Ad Þ1 Y ¼ MðXÞ1 ðI  Ad Þ1 Y (1)
V Q ¼ Q1 ðI  Ad Þ1 Y ¼ NðXÞ1 ðI  Ad Þ1 Y (2)

V: GHG emissions associated with this type of products are included. where
22 Y.-Y. Sun / Tourism Management 45 (2014) 16e27
Y ¼ Visitor consumption of products and services specified in
the TSA report
X ¼ Total output
Ad ¼ Domestic technical input coefficients
M ¼ Flow-in resource matrix
N ¼ Flow-out commodity matrix
R1 ¼ M (X)1 ¼ diagonalized matrix of the amount of resource
required per dollar’s worth of output by industry; the energy
requirement intensity
Q1 ¼ N (X)1 ¼ diagonalized matrix of the amount of commodity
emitted per dollar’s worth of output by industry; the GHG
emission intensity
R ¼ Total energy required, directly and indirectly, in domestic
production
Q ¼ Total amount of GHG emitted, directly and indirectly, in
domestic production
The general approach to account for the carbon emissions
related to imports is to trace the production function using the
global inter-national inputeoutput models or the multi-regional
inputeoutput model (Wiedmann, 2009; Wiedmann, Lenzen,
Turner, & Barrett, 2007). A detailed calculation formula for multi-
regional IO analysis can be found at Peters (2008). If such an
inter-country transaction table is not available, an approximation
can be thought of by using the “domestic technology assumption
(DTA),” which assumes that imported goods are produced under
the same technology procedures as domestic goods5 (Druckman,
Bradley, Papathanasopoulou, & Jackson, 2008; Wood & Dey,
2009). The computation of the embedded carbon footprint of im-
ports is as follows:
Qm ¼ Q1 ðI  Ad Þ1 Am ð1  Ad Þ1 Yd þ Q1 ðI  Ad Þ1 Ym ; (3)
where
Yd ¼ Visitor consumption of domestic products
Ym ¼ Visitor consumption of imported goods
Am ¼ Imports technical input coefficients
Qm ¼ The amount of GHG emitted for goods produced in foreign
industries to meet the final demand and intermediate input
demanddomestically
4. An island economy e Taiwan
Taiwan, similar to other worldwide destinations, has eagerly
embarked on tourism development over the past decade. As a
result of extensive investment and promotional activities, the total
number of annual inbound tourists to Taiwan increased from 3.4
million in 2004 to 8.0 million in 2013; this is an annual growth rate
of 11.7% (Taiwan Tourism Bureau, 2014). Based on the World
Tourism Organization’s (UNWTO) Tourism Highlight report (2011),
among the 25 Asian countries that are covered in these statistics
Taiwan ranked number one in terms of the growth rate of inbound
arrivals between 2008 and 2010, averaging around 20.5% annually.6
At the same time, domestic tourists, measured by person trips,
reached a total of 152.3 million in 2011 with spending of US$10.5
billion.
5
The other argument for adopting the DTA assumption is that the evaluated
county has no jurisdiction over foreign products. Adopting the DTA provides an
assessment for demonstrating the scale of emissions produced if these imported
products are produced domestically (Turner et al., 2011).
6
Based on Tourism Highlight 2012 edition, the annual growth rate of Taiwan
inbound visitors between 2010 and 2011 was 9.3%, higher than the averages of Asia
and Pacific region (World Tourism Organization, 2012).
Total tourism consumption in Taiwan was estimated to be
US$17.0 billion in 2007, which included inbound tourism expen-
ditures of US$6.3 billion (37%), domestic tourism of US$6.6 billion
(39%), and domestic spending associated with outbound travel of
US$4.1 billion (24%) (Taiwan Tourism Bureau, 2009). This, in total,
contributed 2.16% of the national GDP in 2007. Among the total
tourism expenditures, shopping and air transportation each
accounted for 23% of total spending, followed by food (19%) and
land transportation (13%). In the TSA report, shopping expenditures
were not itemized by major categories. An allocation percentage for
distributing the US$4.0 billion shopping expense to 10
manufacturing categories was obtained from the inbound visitor
survey which asked respondents to report their purchased items
and expenditures in details (Taiwan Tourism Bureau, 2007e2012).
It is assumed here that domestic tourists would incur the same
shopping propensity as inbound visitors with respect to these 10
items so that an estimate for the individual shopping category can
be calibrated.
In this study, the GHG emissions are reported as CO2e (carbon
dioxide equivalent), which measure the amount of CO2, CH4 and
N2O emissions, weighted according to their global warming po-
tentials (GWP). The GHG efficiency ratio is calculated for 50 Inpute
Output industry sectors of Taiwan by first summing the energy uses
of 16 different energy sources such as coal, petroleum products,
natural gases, and electricity, multiplying by the corresponding
GHG emissions rate, and then dividing by the total sales output for
each industry. Parameters of energy use, their corresponding CO2e
emissions rate, and InputeOutput data are obtained from the
Taiwan government statistics in the base year of 2006 (Bureau of
Energy, 2012; Directorate-General of Budget Accounting and Sta-
tistics, 2009). The carbon footprint is defined as the direct and in-
direct greenhouse gas emissions-measured in tonnes of carbon
dioxide multiplied with the global-warming potential (equivalent
100-year time horizon) for satisfying one-year consumption.
Without data on multi-regional IO models, this study adopts the
single-country IO model with the DTA assumption. This approach
may lead to overestimation in tourism CF because the world-wide
production technology and CO2 emission intensity is lower than the
Taiwan averages used in the model (Andrew, Peters, & Lennox,
2009). However, at the same time, the overestimation may be
offset by the fact that multidirectional trade, which evaluates the
additional feedback loops in the home country and trade patterns,
is not captured currently.
5. Results
5.1. Direct effect of energy consumption and CO2 emission
To support the tourism economic activities in 2007, it is esti-
mated that Taiwan needed 11.6 thousand kiloliters of oil equivalent
(KLOE) of coal, 4.9 million KLOE of petroleum products, 82.4
thousand tons of natural gas, and 1884 million kilowatt-hours
(kWh) of electricity (Table 4). This translates to the emission of
8.11 Mt of CO2e in support of direct tourism activities. Considering
that the 2007 national CO2 estimation for Taiwan was 262.81 Mt
(Bureau of Energy, 2011), tourism consumption was estimated to
contribute to 3.08% of the national CO2e. From a sector perspective,
aviation and land transportation accounted for two-thirds of the
direct tourism carbon emissions, followed by food and beverage
services (11%), processed food (8%), and accommodation services
(5%). Tourism development overall places a high demand on elec-
tricity (10.8%) and natural gas (4.7%) as compared to Taiwan’s na-
tional energy consumption.
Air transportation by Taiwan-national airlines emitted 3.62 Mt
of CO2e in 2007. Because domestic air transportation only accounts
 Y.-Y. Sun / Tourism Management 45 (2014) 16e27 23

Table 4
Direct tourism expenditure, energy consumption and CO2e emission by IO sectors in Taiwan, 2007.

IO sectors Final demand Coal products (KLOE) Petroleum products Natural gas Electricity Total CO2e emission Pct of CO2e
(NT$ billion’s) (thousand KLOE) (Mt) (million kWh) (thousand ton)

Primary sectors 0.0 0 0 0 0 0 0%

Manufacturing sectors 87.6 11,609 113 2692 615 620 8%
Process foods 44.9 0 30 449 270 256 3%
Apparel and clothing 12.6 1010 24 758 179 188 2%
Cosmetics 4.4 7077 36 262 65 80 1%
Medicines 2.2 3522 18 130 32 40 0%
Other products 10.4 0 1 518 27 22 0%
Entertainment articles 9.1 0 1 454 24 19 0%
Tobacco 2.1 0 1 21 13 12 0%
Electrical products 2.0 0 0 100 5 4 0%
Service sectors 480.8 0 4789 79,715 1270 7485 92%
Air transportation 128.6 0 1343 0 0 3616 45%
Land transportation 72.8 0 3240 0 0 2325 29%
Food and beverage 106.4 0 140 54,256 574 866 11%
Accommodation 49.9 0 66 25,459 270 406 5%
Arts, entertainment 51.9 0 0 0 338 215 3%
Wholesale trade 27.8 0 0 0 49 31 0%
Retail trade 15.4 0 0 0 27 17 0%
Travel agency services 18.1 0 0 0 7 5 0%
Support services 9.9 0 0 0 4 3 0%
Other services 0 0 0 0 0 0 0%
Total 562.4 11,609 4901 82,407 1884 8106 100%
National total 8,018,281 45,537 1,752,806 235,698 262,811
Tourism contribution 0.14% 10.76% 4.70% 0.80% 3.08%

for 3.1% of total passenger kilometers (pkm), the 3.62 Mt of carbon
emissions is mainly contributed by international flights operated by
Taiwan-based airlines. Due to this consideration, if we then exclude
international aviation, Taiwan’s tourism carbon emissions are
estimated to be 4.5 Mt, or 1.71% of the national CO2 emission.
Calculation in this manner produces a CO2 emission value that is
45% lower than the baseline point.
5.2. Domestic indirect effect & imported products
The Taiwan sector-specific direct, indirect and imported effects
are first standardized7 to compare the composition of CO2e by
different spending category (Fig. 2). For land and air transportation,
which are energy intensive, the CO2e is mainly emitted in the direct
effect stage (>65%). Also, air transport generates a higher propor-
tion of carbon emission through imports (mainly petroleum
products) as compared to land transportation. On the other hand,
the core tourism consumption, including food & beverages, enter-
tainment, and accommodation emit half their carbon emissions
during the direct effect stage; 30% through the domestic inter-
industry supply chains, and around 20% from imports. Travel ser-
vice and car rental services are energy efficient, supporting fewer
carbon emissions per dollar sales at the front end. Thus, in com-
parison, their indirect and imported products lead to a larger pro-
portion of GHG (>90%). This provides a good example as to why
tourism consumption is not as clean as the general public has
perceived it to be.
Shopping accounts for the largest share in total tourism con-
sumption in Taiwan, and its average carbon emission intensity
depends greatly on the proportion of retail margins as well as the
composition of products that visitors purchase. Based on the in-
bound visitor survey, 40% of purchased items are food-related
souvenirs, such as pastry and Chinese tea, followed by clothes
(15%) and handicrafts (10%), while around one-third of the tag price
goes to the retail margin. This current shopping pattern produces
27% of its carbon emissions at the direct effect stage, 35% at the
domestic supply-chain, and 39% for imported materials.
Total carbon emission, after taking direct, indirect and imported
effects, for tourism activities in 2007 was estimated to be 14.69 Mt
tons of CO2e (Table 5). Sectors that produce the most carbon di-
oxide were aviation (38%) and land transportation (21%), followed
by shopping (17%) and food (13%). Shopping and food consumption
were responsible for the largest indirect GHG emissions while
aviation generated the most carbon emission through imports.
The comparative analysis of the environmental performance of
the tourism sector with other sectors placed tourism 9th among 35
sectors and households in regard to direct carbon contribution (1st
place is the worst sector). If not considering international aviation,
tourism’s direct emissions would be reduced to 4.45 Mt, placing it
15th in the economy. The carbon-efficiency analysis of tourism,
measured by the direct CO2e emissions per million dollar output,
was 14.26, ranking it 14th (1st place is the worst sector) among the
35 sectors in Taiwan. In general, due to the energy-intensive nature
of land and air transportation, the carbon-efficiency of average
tourism consumption is around 80% worse than the national carbon
emission average per dollar output.
To assign carbon responsibility, we distinguished and estimated
three levels of tourism GHG.

7
Direct, indirect and imported CO2 is divided by the total CO2, respectively, to Fig. 2. The proportion of direct CO2, indirect CO2 and imported CO2 by tourism
derive the percentage of each component. consumption.
24 Y.-Y. Sun / Tourism Management 45 (2014) 16e27
Table 5
CO2e (direct, indirect and embedded in imports) by tourism consumption in Taiwan, 2007.
unit: million ton CO2e for domestic products (direct effect)
Air transport 3.62
Land transport 2.33
Shopping 0.67
Food 0.87
Lodging 0.41
Entertainment 0.22
Travel service 0.00
Car-rental service 0.00
Total 8.11
Pct 55%
Level 1: Domestic direct and indirect carbon emissions
This scope of GHG is consistent with the Kyoto Protocol that
specifies GHG responsibility by the national territory and excludes
emissions from imported material and international aviation. Un-
der this scope, to support tourism development of Taiwan in 2007,
the direct and domestic indirect effect of land transport, shopping,
food, lodging, entertainment, travel service, and car-rental, yields a
total of 9.38 Mt of CO2e.
Level 2: Domestic carbon emissions þ international aviation
emissions
This scope includes level 1 GHG emissions plus the direct and
domestic indirect emissions associated with international aviation
for Taiwan-based airlines.8 The level II emissions are a total of
11.00 Mt of CO2e.
Level 3: Domestic carbon emissions þ international aviation
emissions þ imported emission
This scope includes all GHG emitted during both the direct and
indirect stages as well as the foreign-sourced emissions for imports
required by Taiwan tourism consumption in 2007. The Level 3
emissions estimate for the baseline year is 14.69 Mt of CO2e.
The Level 3 emissions portray the most complete picture of the
emissions required to sustain the tourism activities within Taiwan
for the base year, without differentiating the size of the carbon
footprint by geographical area or by international accounting
agreements. If tourism activities ceased to operate in 2007, the
Level 3 carbon footprint would reduce to zero, among which 47% is
emitted in Taiwan, 28% in international airspace, and 25% in foreign
countries. The Level 3 carbon emission not only benchmarks the
tourism carbon emission of the base year but also serves as a re-
sponsibility of this island country.
6. Discussion
The scope of tourism consumption referenced in a carbon
assessment study determines the scale that we perceive and for
which we are responsible. The standard production accounting
principle and consumption accounting principle typically adopted
in calibrating the national carbon emissions are somewhat inade-
quate in addressing the tourism carbon footprint. The PAP approach
ignores the GHG impact embedded with imports which are utilized
directly and indirectly by local business operations. The CAP
approach, on the other hand, leaves out the segment of inbound
8
Radiative forcing (RF) of aviation is not included in this calculation as IPCC did
not incorporate RF when computing aviation carbon footprint (IPCC., 1997).
CO2e for domestic products (indirect effect) CO2e embedded in imports Grand total Pct
0.47 1.54 5.62 38%
0.47 0.35 3.15 21%
0.87 0.97 2.51 17%
0.56 0.42 1.85 13%
0.33 0.21 0.94 6%
0.14 0.11 0.47 3%
0.04 0.04 0.08 1%
0.03 0.04 0.07 0%
2.90 3.68 14.69 100%
20% 25% 100%
tourists completely. The proposed approach in this research for
calibrating the national tourism carbon footprint is based on the
scope of Tourism Satellite Account and the computation of the
Environmentally Extended InputeOutput model to trace direct and
secondary effects. This approach bridges both PAP and CAP con-
cepts and places emphasis on the energy use for supporting the
economic significance documented in the TSA scope.
The “TSA þ EEIO” approach has several advantages in its
applicability. The TSA conceptual framework and data compliance
is comprehensive and consistent across nations, which has been
well promoted by the World Tourism Organization (United Nations,
2010). The standard procedure of TSA and EEIO provides each
country an analytical tool to benchmark its tourism economic sig-
nificance, attribute tourism energy use & GHG emission, and
compare its tourism eco-efficiency against other sectors in the
economy. For countries that do not compile the TSA statistics,
World Travel & Tourism Council produces country-specific TSA
estimation annually (World Travel & Tourism Council, 2012),
providing a convenient platform to calibrate and compare the
country to country tourism CO2 emission.
The second advantage of the “TSA þ EEIO” approach is the
straightforward treatment of the international aviation issue. In
this approach, aviation emissions are only attributable to the
tourism sector of a country when the transaction of the air trans-
portation creates an economic significance at the geographic ter-
ritory. National carriers are included as their sales are documented
in the Systems of National Account, and their energy use is allo-
cated for the proportion contributed by inbound and outbound
tourists arriving to and departing from the country. Internationally-
registered carriers are assumed to leave no economic contribution
and their emissions are excluded.
This approach corresponds more closely to the PAP concept: let
local business at the evaluated country take responsibility for its
GHG emissions as those firms are the beneficiaries of such eco-
nomic transaction. This approach does not allocate emissions to the
departing countries of incoming visitors nor does it assign re-
sponsibility by airspace where the jurisdiction may involve a third-
party beyond the countries of origins and final destinations. One
advantage of this approach in treating air emissions is that national
carriers are closely supervised and monitored by the government
for it energy use under operation. The ‘carrot and stick’ policies,
such as public discourses, education, carbon-justified levies or the
cap and trade system, can then be tailored to national carriers to
gain their cooperation in improving their carbon efficiency level
over time (Gössling & Peeters, 2007; Meltzer, 2012; Peeters &
Landré, 2012; Scott, Gössling, & Hall, 2012).
The above-recommended system boundary is determined by
the scope reported in the national TSA. It is noted that two com-
ponents are currently left out. First the externality associated with
outbound travel for their consumption at foreign countries (such as
Taiwan visitors in Japan) is not included in the estimates. The
 Y.-Y. Sun / Tourism Management 45 (2014) 16e27
carbon footprint of this aspect can be compiled once the multi-
regional environmental IO table and detailed expenditure pattern
abroad is made available. The major concern to tackle this
component is the lack of control over the business operation at
foreign countries and, currently, no systematic analysis has been
presented. This hint around the issue that the emissions associated
with outbound travel are a responsibility at the hands of other
destinations.
The second component that needed to be discussed is whether
to include the resources and pollutants embodied in capital terms
used by the tourism sectors. The standard TSA suggests to include
the analysis of stocks and flows of physical investment e the fixed
capital formation (United Nations, 2010). However, most tourism
applications are currently concerned with the energy use that is
associated with daily industry operations in sustaining tourists’
collective consumption. Rarely has any macro assessment on the
tourism-specific fixed assets (hotel, theme park) and tourism-
related infrastructure (roads, airports) been presented. The CF of
fixed capital formation deserves further research as its scale of
influences is currently unclear.
6.1. Empirical observation
Comparing the case study of Taiwan with other tourism carbon
research, a general pattern has occurred as 1) the eco-efficiency of
tourism is typically worse than the national average of carbon
emissions per dollar output, 2) tourism inefficiency in GHG emis-
sions results primarily from the high usage of air and land trans-
portation, and 3) emissions embodied in imports contributes
around 20% of total tourism CF. Two aspects of empirical differences
also emerged: “Shopping” is the third largest CO2 contributor, after
aviation and land transportation (in Taiwan); The non-Kyoto
emissions associated with international aviation and imports ac-
counts for a higher percentage of the total tourism CF at an island
economy. In the following, each point of observation is elaborated.
The major contributor to tourism CO2 and its inefficiency orig-
inates from the need for mobility between regions. In the case of
Taiwan, transportation-related expenditures accounted for 35% of
total tourism consumption in 2007 but contributed more than 70%
of the direct carbon emissions and 60% of the total carbon impact.
Based on Taiwan Civil Aeronautics Administration (2013), total in-
bound and outbound visitors increased from 12.3 million to 17.6
million (þ42%) during the period 2008 through 2012 while the
corresponding international passenger kilometers (pkm) on
national-based airlines increased by 54%. This figure supports that
not only do people travel more frequently by air, but they also travel
to more remote and distant locations. The increasing demand for
aviation and its emission trend is observed for islands and for
continental destinations such as the Netherlands (de Bruijn et al.,
2013). With a substantial volume of international travel observed
and predicted across the continents, air emissions will continue to
be the main driver for the rising tourism energy consumption over
time.
Emissions embodied in imports is another component that
cannot be overlooked (Andrew et al., 2009). No country is exempt
from the globalization of the commerce system, and imports are
broadly adopted in the local production structure, especially for the
manufacturing sectors. For example, foreign-sourced emissions for
tourism consumption in Australia is estimated to be 18% (Dwyer
et al., 2010) and a larger ratio (25%) is estimated for Taiwan. The
main imports for Taiwan tourism services are petroleum and coal
products for aviation; chemical materials, and processed food for
shopping items. This pattern reflects the dependence on foreign
energy supplies as well as the composition of typical souvenirs that
visitor purchased in a region. Excluding this component will
25
generally result in tourism CF being underestimated by a range of
15e25%. This magnitude of error is expected to be larger for
regional destinations that rely more heavily on outside interme-
diate inputs.
In this case study, we observed that shopping expenditures turn
out to be the third largest carbon contributor in Taiwan tourism
development (next to air and land transportation), and are
responsible for more direct, indirect and imported GHG than ac-
commodation, food & beverage or entertainment. In 2007, shop-
ping expenses accounted for 15% of total tourism expenditure but
generated 17% of total tourism CF. This finding is different from
previous literature among which accommodations always ranked
second in importance (Dwyer et al., 2010; WTO-UNEP-WMO,
2008). The observation in Taiwan can be explained by the signifi-
cant share of shopping in relation to total tourism spending and by
amount of energy per unit output required in the local production
structures. Among the 10 shopping categories, “cosmetics,”
“clothing” and “Chinese medicine” generate higher CF than “pro-
cessed food” and “handicraft” on the domestic supply chains and
foreign providers. Because the relationship between GHG emission
and shopping are generally neglected from the carbon studies, no
comparable results at the national level are currently available to
reference Taiwan’s situation. It is suspected that destinations with
larger economic systems will provide more selection and oppor-
tunities for the shopping consumption, thus leading to a larger
share of tourism CF through indirect and imported effects.
The last but not least issue for tourism carbon emissions is the
responsibility of accounting, monitoring and mitigating emissions
outside the national territory. Based on the Kyoto Protocol frame-
work, international aviation and carbon embodied in imports are
exempt from the national carbon production, and are referred to as
non-Kyoto emissions. In the Australia tourism carbon study which
adopted a scope similar to that of the current research, the ratio
between Kyoto and non-Kyoto emissions was found to be around
2:1 (Dwyer et al., 2010). For Taiwan, the Kyoto versus non-Kyoto
emissions is about 1:1. This pattern supports a general observa-
tion that the reliance on international aviation and imports is far
more substantial for a small island economy. Under the KP frame-
work, the island-responsible carbon is largely reduced: tourism CF
is cut back by 53% from the baseline point, in which 28% is from
international aviation and 25% is from imports. This finding allows
us to reiterate the importance of addressing (and evaluating) all
aspects of tourism GHG without excluding the amount of carbon
footprint by geographic areas or by international accounting
agreements, a philosophy strongly promoted in the post-Kyoto
framework.
7. Conclusions
The first step by which a nation can fully understand and bal-
ance the economic benefits and environmental costs of further
tourism development is to analyze the baseline picture of carbon
emissions and its eco-efficiency, in comparison to other sectors. The
purpose of this research is to provide a comprehensive framework
for measuring the national tourism CO2e based on the TSA and the
EEIO models, which bridges both PAP and CAP concepts. The rec-
ommended scope includes expenditure of inbound tourism, do-
mestic tourism, and domestic spending associated with outbound
travel.9 This scope also includes airfares paid to national carriers of
the evaluated country for the cross-border movement from both
inbound and outbound tourists. The tourism carbon footprint is
calibrated for the following categories: domestic direct effect,
9
Single purpose durable goods are not included in the scope of this study.
26 Y.-Y. Sun / Tourism Management 45 (2014) 16e27
domestic indirect effect, and foreign-sourced effect to demonstrate
the attribution of responsibility on the supply chains.
Under this analytical framework and research scope, in 2007,
tourism consumption contributed 2.16% of Taiwan’s GDP, while
direct carbon emissions accounted for 3.08% of the nationwide
GHG, and ranked as the 14th worst (least efficient) of the 35 sectors
using the eco-efficiency analysis. The observations of Taiwan, New
Zealand (Patterson & McDonald, 2004) and Australia (Dwyer et al.,
2010) reveal commonalities that the eco-efficiency of tourism is
typically worse than the national average of carbon emission per
dollar output.
Based on the empirical study of Taiwan, the finding supports
that an island economy which relies heavily on a high volume of air
transportation and imported materials will lead to a larger pro-
portion of non-Kyoto emission (emissions outside the country
territory) in tourism development. The difference between “Kyoto
CO2 emission” and “CO2 responsibility” can be substantial
depending on the research scope covered and the destination
production characteristics. This further supports the importance of
taking into account the energy use of national carriers for inter-
national transportation and the energy use embedded with imports
with respect to tourism development.
Additional empirical research is called for to address the char-
acteristics of tourism carbon emissions from both spatial and
temporal perspectives. Cross-country comparisons can be
employed to identify the tourism CF pattern between island and
continental destinations, to trace the differences based on devel-
oping and developed countries, and to monitor tourism and energy
patterns during each stage of the destination life cycle or by the
Tourism Penetration Index. Such comparisons help to reveal the
competitive advantages across regions. It should be kept in mind
that such assessments need to be based on similar research scope
and analytical frameworks so that the fallacy of comparing apples
and oranges can be avoided. From a temporal perspective, current
literature mainly provides a snapshot of tourism carbon structure
but to obtain a deeper understanding, a consistent process that
addresses time-series changes in final demand, technical struc-
tures, and eco-efficiency is encouraged. Insights from broader
spatial and temporal perspectives, we believe, will facilitate better
public discussion and decision making in balancing tourism
development and carbon mitigation.
Acknowledgments
Financial support from the Taiwan National Science Council
under NSC 101-2410-H-390-030 is gratefully acknowledged. We
also thank three anonymous reviewers for their helpful comments
to the previous version of this paper.
References
Andrew, R., Peters, G. P., & Lennox, J. (2009). Approximation and regional aggre-
gation in multi-regional inputeoutput analysis for national carbon footprint
accounting. Economic Systems Research, 21(3), 311e335.
Becken, S. (2002). Analysing international tourist flows to estimate energy use
associated with air travel. Journal of Sustainable Tourism, 10(2), 114e131.
Becken, S., & Patterson, M. (2006). Measuring national carbon dioxide emissions
from tourism as a key step towards achieving sustainable tourism. Journal of
Sustainable Tourism, 14(4), 323e338.
de Bruijn, K., Dirven, R., Eijgelaar, E., & Peeters, P. (2013). Travelling large in 2012: The
carbon footprint of Dutch holidaymakers in 2012 and the development since 2002.
Breda, Netherlands: NHTV Breda University of Applied Sciences.
Bureau of Energy. (2011). Carbon emissions in Taiwan. Taipei, Taiwan: Buruea of
Energy, Ministry of Economic Affairs.
Bureau of Energy. (2012). The GHG Reduction Act. Retrieved Sep 1, 2012, from http://
unfccc.epa.gov.tw/unfccc/english/04_our_efforts/061_policy.html.
Civil Aeronautics Administration. (2013). Aeronautic statistics. Retrieved April 5,
2013, from http://www.caa.gov.tw/big5/content/index.asp?sno¼186.
Collins, A., Flynn, A., Munday, M., & Roberts, A. (2007). Assessing the environmental
consequences of major sporting events: the 2003/04 FA Cup Final. Urban
Studies, 44(3), 457e476.
Collins, A., Jones, C., & Munday, M. (2009). Assessing the environmental impacts of
mega sporting events: two options? Tourism Management, 30(6), 828e837.
Directorate-General of Budget Accounting and Statistics. (2009). 2006 National
input-output table. Retrieved Jan 1, 2012, from http://www.dgbas.gov.tw/.
Druckman, A., Bradley, P., Papathanasopoulou, E., & Jackson, T. (2008). Measuring
progress towards carbon reduction in the UK. Ecological Economics, 66(4), 594e
604.
Dubois, G., & Ceron, J. P. (2006). Tourism/leisure greenhouse gas emissions forecasts
for 2050: factors for change in France. Journal of Sustainable Tourism, 14(2), 172e
191.
Dwyer, L., Forsyth, P., & Spurr, R. (2006). Economic evaluation of special events. In
L. Dwyer, & P. Forsyth (Eds.), International handbook on the economics of tourism.
Glos, UK: Edward Elgar Publishing Limited.
Dwyer, L., Forsyth, P., Spurr, R., & Hoque, S. (2010). Estimating the carbon footprint
of Australian tourism. Journal of Sustainable Tourism, 18(3), 355e376.
Farreny, R., Oliver-Solà, J., Lamers, M., Amelung, B., Gabarrell, X., Rieradevall, J., et al.
(2011). Carbon dioxide emissions of Antarctic tourism. Antarctic Science, 23(06),
556e566.
Filimonau, V., Dickinson, J. E., Robbins, D., & Reddy, M. V. (2011). A critical review of
methods for tourism climate change appraisal: life cycle assessment as a new
approach. Journal of Sustainable Tourism, 19(3), 301e324.
Fletcher, J. E. (1989). Inputeoutput analysis and tourism impact studies. Annals of
Tourism Research, 16(4), 514e529.
Frechtling, D. C. (2010). The tourism satellite account: a primer. Annals of Tourism
Research, 37(1), 136e153.
Gössling, S., Hansson, C. B., Hörstmeier, O., & Saggel, S. (2002). Ecological footprint
analysis as a tool to assess tourism sustainability. Ecological Economics, 43(2),
199e211.
Gössling, S., & Peeters, P. (2007). ‘It does not harm the environment!’ An analysis of
industry discourses on tourism, air travel and the environment. Journal of
Sustainable Tourism, 15(4), 402e417.
Hanandeh, A. E. (2013). Quantifying the carbon footprint of religious tourism: the
case of Hajj. Journal of Cleaner Production, 52(1), 53e60.
Hoque, S., Forsyth, P., Dwyer, L., Spurr, R., Van Ho, T., & Pambudi, D. (2010). The
carbon footprint of Queensland tourism: CRC for sustainable tourism Pty Ltd.
Hunter, C., & Shaw, J. (2007). The ecological footprint as a key indicator of sus-
tainable tourism. Tourism Management, 28(1), 46e57.
IPCC. (1997). Revised 1996 IPCC Guidelines for national greenhouse gas inventories refer-
ence manual (Vol. 3). Paris: Organization for Economic Cooperative Development.
Jones, C. (2008). Assessing the impact of a major sporting event: the role of envi-
ronmental accounting. Tourism Economics, 14(2), 343e360.
Jones, C., & Munday, M. (2007). Exploring the environmental consequences of
tourism: a satellite account approach. Journal of Travel Research, 46(2), 164e172.
Kelly, J., & Williams, P. W. (2007). Modelling tourism destination energy con-
sumption and greenhouse gas emissions: Whistler, British Columbia, Canada.
Journal of Sustainable Tourism, 15(1), 67e90.
Konan, D. E., & Chan, H. L. (2010). Greenhouse gas emissions in Hawaii: household
and visitor expenditure analysis. Energy Economics, 32(1), 210e219.
Kuo, N.-W., & Chen, P.-H. (2009). Quantifying energy use, carbon dioxide emission,
and other environmental loads from island tourism based on a life cycle
assessment approach. Journal of Cleaner Production, 17(15), 1324e1330.
Kuo, N.-W., Lin, C.-Y., Chen, P.-H., & Chen, Y.-W. (2012). An inventory of the energy use
and carbon dioxide emissions from island tourism based on a life cycle assess-
ment approach. Environmental Progress & Sustainable Energy, 31(3), 459e465.
Li, M., Zhang, J., Chen, J., Zhou, J., & Wang, N. (2012). Estimating the energy carbon
footprint of Huangshan National Park. Advanced Materials Research, 535-537,
2214e2219.
McElroy, J. L., & Parry, C. E. (2010). The characteristics of small island tourist
economies. Tourism and Hospitality Research, 10(4), 315e328.
Meltzer, J. (2012). Climate change and trade-The EU aviation directive and the WTO.
Journal of International Economic Law, 15(1), 111e156.
Miller, R. E., & Blair, P. D. (2009). Inputeoutput analysis: Foundations and extensions.
Cambridge, UK: Cambridge University Press.
Minx, J., Wiedmann, T., Wood, R., Peters, G., Lenzen, M., Owen, A., et al. (2009).
Inputeoutput analysis and carbon footprinting: an overview of applications.
Economic Systems Research, 21(3), 187e216.
Munday, M., Turner, K., & Jones, C. (2013). Accounting for the carbon associated
with regional tourism consumption. Tourism Management, 36, 35e44.
Munksgaard, J., & Pedersen, K. A. (2001). CO2 accounts for open economies: pro-
ducer or consumer responsibility? Energy policy, 29(4), 327e334.
Patterson, M., & McDonald, G. (2004). How clean and green is New Zealand tourism?
Lifecycle and future environmental impacts. Lincoln, NZ: Landcare Research.
Peeters, P., Gossling, S., & Becken, S. (2006). Innovation towards tourism sustain-
ability: climate change and aviation. International Journal of Innovation and
Sustainable Development, 1(3), 184e200.
Peeters, P., & Landré, M. (2012). The emerging global tourism geography e an
environmental sustainability perspective. Sustainability, 4(1), 42e71.
Peters, G. P. (2008). From production-based to consumption-based national emis-
sion inventories. Ecological Economics, 65(1), 13e23.
Scott, D., Gössling, S., & Hall, C. M. (2012). International tourism and climate change.
Wiley Interdisciplinary Reviews: Climate Change, 3(3), 213e232.
Smith, S. L. J. (1994). The tourism product. Annals of Tourism Research, 21(3), 582e595.
 Y.-Y. Sun / Tourism Management 45 (2014) 16e27
Sun, Y.-Y. (2007). Adjusting InputeOutput models for capacity utilization in service
industries. Tourism Management, 28(6), 1507e1517.
Sun, Y.-Y., & Pratt, S. (2014). The economic, carbon emission, and water impacts of
Chinese Visitors to Taiwan: eco-efficiency and impact evaluation. Journal of
Travel Research. http://dx.doi.org/10.1177/0047287513517420.
Tabatchnaia-Tamirisa, N., Loke, M. K., Leung, P., & Tucker, K. A. (1997). Energy and
tourism in Hawaii. Annals of Tourism Research, 24(2), 390e401.
Taiwan Tourism Bureau. (2007e2012). Annual survey report on visitors expenditure
and trends in Taiwan. Retrieved August 20, 2012, from http://admin.taiwan.net.
tw/english/statistics/File/200912/98citizen.pdf.
Taiwan Tourism Bureau. (2009). Taiwan tourism satellite account 2007. Taipei:
Ministry of Transportation and Communications, Republic of China.
Taiwan Tourism Bureau. (2014). Visitor arrivals. Retrieved March 3, 2014, from
http://admin.taiwan.net.tw/statistics/File/200912/table01_2009.pdf.
Turner, K., Munday, M., McGregor, P., & Swales, K. (2012). How responsible is a
region for its carbon emissions? an empirical general equilibrium analysis.
Ecological Economics, 76, 70e78.
Turner, K., Munday, M., McIntyre, S., & Jensen, C. (2011). Incorporating jurisdiction
issues into regional carbon accounts under production and consumption ac-
counting principles. Environment and Planning A, 43(3), 722e741.
UNFCCC. (2013). Kyoto protocol. Retrieved Sep 23, 2013, from http://unfccc.int/
kyoto_protocol/items/2830.php.
United Nations. (2010). Tourism satellite account: recommended Methodological
framework 2008. New York: United Nations.
Wang, Y., & Davidson, M. C. (2010). Chinese holiday makers’ expenditure: Impli-
cations for marketing and management. Journal of Hospitality Marketing &
Management, 19(4), 373e396.
Whittlesea, E. R., & Owen, A. (2012). Towards a low carbon futureethe development
and application of REAP Tourism, a destination footprint and scenario tool.
Journal of Sustainable Tourism, 20(6), 845e865.
Wiedmann, T. (2009). A review of recent multi-region inputeoutput models used
for consumption-based emission and resource accounting. Ecological Economics,
69(2), 211e222.
27
Wiedmann, T., Lenzen, M., Turner, K., & Barrett, J. (2007). Examining the global
environmental impact of regional consumption activitiesdPart 2: review of
inputeoutput models for the assessment of environmental impacts embodied
in trade. Ecological Economics, 61(1), 15e26.
Wood, R., & Dey, C. J. (2009). Australia’s carbon footprint. Economic Systems
Research, 21(3), 243e266.
World Tourism Organization. (2011). UNWTO tourism highlights. Madrid, Spain:
UNWTO, 2011 Edition.
World Tourism Organization. (2012). UNWTO tourism highlights. Madrid, Spain:
UNWTO, 2012 Edition.
World Travel & Tourism Council. (2012). Economic impact research e country
report. Retrieved Sep 13, 2013, from http://www.wttc.org/research/economic-
impact-research/country-reports/.
WTO-UNEP-WMO. (2008). Climate change and tourism: responding to global chal-
lenges. Spain: World Tourism Organization.
Dr. Ya-Yen Sun is an assistant professor at the National
Cheng Kung University, Taiwan. Her reach area is on the
economic impact evaluation associated with tourism
development using the InputeOutput Model. The empir-
ical cases that she has worked with included economic
impact studies of US. National Park Service visitor con-
sumption, Taiwan National Tourism Policy and 2009 World
Games In addition to empirical works, her research has
involved with theoretical perspectives of the standard
InputeOutput model, the characteristics of tourism in-
dustries, and the carbon emission evaluation framework.