Willingness to Pay of Air Passengers for Carbon-Offset

Abstract

An important source of anthropogenic greenhouse gas (GHG) emissions is the air transport sector, which accounts for approximately 2% of global GHG emissions. Therefore, reducing GHG emissions from aircrafts has become a major challenge for transportation authorities worldwide. In recent years, much research has focused on tax ideas related to the CO2 emissions produced by air transport, such as the voluntary carbon offset (VCO). This study investigates the willingness of economy class air passengers to pay to compensate for the CO2 emissions produced during their journeys from Taiwan to Hong Kong. Together with the Spike model, a framework known as the contingent valuation (CV) method offers a way to investigate how much the air passenger would be willing to pay to offset a journey’s airplane-generated CO2 emissions. The Spike model was applied to address the problem of zero willingness to pay (WTP). The results obtained in this study are consistent with the results found in previous studies and therefore can provide valuable insights into pricing strategies for airlines.

Full text

Sustainability 2015, 7, 3071-3085; doi:10.3390/su7033071
sustainability
ISSN 2071-1050
www.mdpi.com/journal/sustainability
Article
Willingness to Pay of Air Passengers for Carbon-Offset
Rong-Chang Jou * and Tzu-Ying Chen
Department of Civil Engineering, National Chi Nan University, No.1, University Rd., Puli, Nantou 54561,
Taiwan; E-Mail: tychen@ncnu.edu.tw
* Author to whom correspondence should be addressed; E-Mail: rcjou@ncnu.edu.tw;
Tel.: +886-49-291-0960 (ext. 4956); Fax: +886-49-291-8679.
Academic Editor: Giuseppe Ioppolo
Received: 29 October 2014 / Accepted: 9 March 2015 / Published: 13 March 2015
Abstract: An important source of anthropogenic greenhouse gas (GHG) emissions is the
air transport sector, which accounts for approximately 2% of global GHG emissions.
Therefore, reducing GHG emissions from aircrafts has become a major challenge for
transportation authorities worldwide. In recent years, much research has focused on tax
ideas related to the CO2 emissions produced by air transport, such as the voluntary carbon
offset (VCO). This study investigates the willingness of economy class air passengers to
pay to compensate for the CO2 emissions produced during their journeys from Taiwan to
Hong Kong. Together with the Spike model, a framework known as the contingent
valuation (CV) method offers a way to investigate how much the air passenger would be
willing to pay to offset a journey’s airplane-generated CO2 emissions. The Spike model
was applied to address the problem of zero willingness to pay (WTP). The results obtained
in this study are consistent with the results found in previous studies and therefore can
provide valuable insights into pricing strategies for airlines.
Keywords: air transport CO2 emissions; voluntary carbon offset; Spike model
1. Introduction
With the increase in global emissions of carbon dioxide, the greenhouse effect is considered to be an
increasingly serious problem with significant effects on the human living environment. Although
environmental issues have become a widespread concern in recent years, the question of how to slow
OPEN ACCESS

Sustainability 2015, 7 3072
global warming remains. To reduce the production of greenhouse gas (GHG), multiple agreements have
been signed, including the Kyoto Protocol [1] and the Copenhagen Accord [2]. In addition, the GHG
produced by the air transport industry accounts for approximately 3% of the world’s carbon emissions [3].
There are studies on voluntary carbon emissions reductions from an economic approach, such as the
empirical literature on impure public goods [4].
In the study of Simone et al. [5], aviation is responsible for 3% of global fossil fuel consumption and
12% of transportation-related carbon emissions. Although aviation is not one of the major drivers of
global warming, the significant growth suggests it could be a major factor over the next decades [6,7].
Therefore, reducing carbon emissions related to aviation is a necessary task despite its currently
relatively small contribution to total carbon emissions [8]. As such, during the annual meeting of the
International Air Transport Association (IATA) held in Kuala Lumpur in June 2009, the IATA agreed to
reach “carbon-neutral growth” in 2020, on behalf of the global aviation industry, and to reduce carbon
emissions by 50%.
To respond to the concerns over the environmental impacts of aviation, the International Air
Transport Association (IATA) in 2009 aimed at achieving stable carbon emissions from 2020 onwards
despite further growth in air traffic, by a combination of different methods such as fleet renewal,
operational and infrastructure measures, retrofits, offset mechanisms and the use of alternative fuels [9].
In recent years, several studies have found that improved aircraft technology and highly efficient
allocation can reduce an aircraft’s carbon dioxide emissions. In addition, changing the attitude of airline
passengers toward carbon emissions during travel and expanding advocacy and public education about
the understanding of carbon emissions can facilitate the implementation of low-carbon policies [10,11].
The overall global trend reveals that many European Union member states have adopted a tax policy
that raises the ticket price to force airline passengers to pay an additional sum of money to compensate
for the carbon emissions generated by air travel. This tax covers the external costs, such as air pollution,
energy security and climate change caused by GHG emissions and is expected to reduce the load on the
environment and help achieve the sustainable growth of the air transport industry.
A similar policy, known as the voluntary carbon offset (VCO), has been presented to the aviation
industry and at the airports of a number of countries in Europe and the Americas. Primarily, this policy
enables passengers to use a website to calculate the carbon emissions generated by their journey, and the
passengers can then use methods, such as personal payment or bonus point redemption, to compensate
for (“offset”) the economic costs of carbon-dioxide emissions or to donate to tree-planting programs or
other environmental projects implemented by relevant agencies [12,13]. In addition, informing passengers
regarding corporate social responsibility (CSR) by implementing this policy and the benefits of this
policy for the passenger can enhance the willingness to participate in the policy [7,14,15]. Other research
on the carbon offset includes using the theory of planned behaviour or the model of goal-directed
behaviour (MGB) to investigate the factors influencing on the intention to participate in carbon-offset
schemes [13,15–17].
As indicated above, the amount of money that airline passengers pay as a carbon offset is primarily
determined by the airlines based on the relevant flight data. That is, the amounts are often determined by
the decision-making of the supplier, and little has been done to investigate the price that the airline
passengers are willing to pay for the carbon offset. Only a few studies have applied the contingent
valuation (CV) method to investigate the price that air passengers are willing to pay for carbon-offset.

Sustainability 2015, 7 3073
For examples, MacKerron [7] and Brouwer et al. [18] applied a binary probabilistic model to estimate
the price that air passengers are willing to pay for the carbon offset. Lu and Shon [19], on the other hand,
used interval regression model to only estimate the determinants of willingness to pay (WTP) but not the
amount of WTP.
Stated preference methods are a useful way to understand the preferences of individuals by observing
their choices in hypothetical scenarios presented in a survey (i.e., the observed choices are contingent on
the scenarios provided). In particular, CV is used to represent the process of using stated preference data
for valuation purposes. There are many ways in which to elicit preference information in a CV study,
with discrete choice experiments (DCEs) the most common (Readers interested in more materials on
CV and DCE can refer to Carson and Czajkowski [20]). Despite DCE having been widely applied in
the transportation field, yet, CV can be justified in certain circumstances that have to do with the
study population.
Because carbon-offset is not supported by all airline passengers, there will be situations where the
price that many respondents are willing to pay is zero (the previously mentioned studies did not consider
this issue). If the WTP of many respondents is zero, bias will result in the WTP estimated using econometric
models (standard discrete choice model). Various approaches to dealing with zero bids have been used
in the CV literature. One option is to simply remove these observations from the sample [21,22];
however, such a complete removal may not be appropriate when mean bids are calculated [23].
Vaughan et al. [24] also noted that the statistical estimation would result in a negative estimator in the
field of zero bids, while Calia and Strazzera [25] observed that the exclusion of zero bids might induce
sample selection bias. In addition, the first paper to have criticized the papers for their failure to examine
zero willingness to pay responses and to conduct an appropriate control exercise is Diederich and
Goeschl [26].
Another approach explicitly allows for a point mass at zero, which is the truncation at zero of the
WTP distribution. In this stream, Kriström [27] proposed a framework based on the Spike model, which
takes into account that the WTP of many respondents is zero. Thus, the WTP derived after model
estimation more realistically reflects the WTP when the data contain many zeros, e.g., above 10% of the
total sample [28–33]. Therefore, the Spike model provides a more realistic depiction than other models.
Given the dearth of the utilisation of this approach in the literature, this study used the CV method to
assess non-market value to survey the WTP of economy class airline passengers and prepared an
econometric model suitable for the price measured in this study (The WTP of air passengers for other
classes and flight routes can be estimated by using the same approach adopted in the paper. It is only a
matter of data collection. Further research can be done by comparing WTPs among different groups).
In addition, only the flight route Taiwan to Hong Kong was considered in this study (The study looks
at a single flight connection, which limits the generalizability of its data, readers interpret the results
extended beyond the specific route shall be cautious). Spike model, which can better reflect the actual
circumstances, was applied to process the WTP of airline passengers for the carbon offset. This study
will help the airline industry obtain a better understanding of the potential populations that will accept
this policy, develop the necessary marketing strategy to target these populations and achieve a
competitive edge in operation. The study also provides references for the air transport industry in
promoting the carbon-offset pricing mechanism.

Sustainability 2015, 7 3074
2. Model Framework
When a high percentage of the sample observational values of consumer spending is zero, the sample
is obviously not conforming to a normal distribution. In this case, if we continue to use the regression
model to analyse the sample data, it will inevitably lead to bias of standard errors in the estimated results.
In addition, estimated WTP will be negative in an unbounded linear WTP model, when a large
proportion of respondents say no to the lowest bid amount or state a zero WTP. Therefore, focusing on a
processing method for samples with many zero values, Kriström [27] proposed the Spike model to solve
the problems caused by zero WTP in the sample data.
According to the random utility theory and assuming that the utility U includes the observable utility
V and the unobservable random error

, the utility function is represented by Equation (1):
(, ; ) (, ; )UyXQ VyXQ


 (1)
where V represents the indirect utility function of the amount of money for the carbon offset paid by
airline passengers, y represents the income level of airline passengers,
X
represents the
socioeconomic variables of airline passengers, Qrepresents the related variables that affect the
participation of economy class air passengers in the carbon-offset policy, and ε is the random item of
the utility function and has an independent and identical distribution (iid). When airline passengers are
willing to pay the bid amount ( A) to offset the carbon emissions, the new utility 1
V is the utility
function when the airline passengers pay the offset for carbon emissions and is higher than the original
utility ( 0
V). Therefore, the utility function can be rewritten as Equation (2):
111000
(,;) (,;)Vy AXQ VyXQ


  (2)
0

and 1

are independently and identically distributed with a zero mean. Thus, the probability
function for the economy class air passengers’ willingness to pay the scenario price (
A
) can be
derived from Equation (2) to Equation (3).
Pr( ) Pr( ) 1 ( ) ( ( ))
WTP
Yes WTP A F A F V

 (3)
Because WTP represents the maximum amount that the economy class air passengers pay, when the
scenario provides a price lower than the WTP (WTP A), the economy class air passengers always
pay the price. ( )
WTP
FA represents the cumulative distribution function and is shown in Equation (4).
, 0
() (), 0
WTP
WTP
pA
FA GAA



 (4)
p belongs to (0, 1), and ( )
WTP
FAis a continuous increasing function if (0)
WTP
Gp

and
lim ( ) 1
WTP
AGA
 .
The expected amount of the WTP for carbon offsets from airline passengers can be
expressed as follows:
0
()(1 ()
WTP
E
WTP F A dA


 (5)
Therefore, the parameter estimation of the Spike model is calculated using maximum likelihood
estimation (MLE), as shown in Equation (6).
()EWTP

Sustainability 2015, 7 3075


ln 1 ( )
(1 ) ln ( ) (0)
(1 )ln (0)
N
ii WTP
i
N
iiWTP WTP
i
N
iWTP
i
LMW FA
MW FAF
MF

 




(6)
where
M
indicates whether there is a value range of the airline passenger’s (WTP>0). W
indicates the response of whether the airline passenger pays the last bid price A, that is, whether WTP
is greater than A.
M
and W are defined in Equations (7) and (8), respectively.
1, 0
0,
WTP
Mothers



 (7)
1, >
0,
WTP A
Wothers


 (8)
Without a loss of generality, we assume that the utility function is linear and consider only the effect
of income y. For simplicity, we derive the WTP with the income variable only. It is straightforward to
derive the WTP with other variables included, such as Q and X. However, these two types of variable
must be specified as alternative specific variables instead of generic variables for estimation purposes.
Thus, the difference between the utilities of the new state and the current state can therefore be
expressed as follows:
10
(*) =
=
VA
A





 (9)
We then assume that ( )
WTP
GA has the form of a logistical function, which means that (*)FV


can
be shown as follows:
1
(*) 1 exp( )
FV A




 (10)
Equation (4) can be further expressed as follows:
1
1
[1 exp( )] , 0
() [1 exp( )] , 0
WTP
A
FA AA












 (11)
where α is the marginal utility of improving environmental conditions after adopting the VCO policy
and β is the marginal utility of paying the amount of money for the VCO. We can derive the expected
WTP as follows:

1
( ) ln 1 exp( )E willingnesstopay


 (12)
The Spike value can be defined as in the following equation by setting 0A

.

1
1exp
Spike

 (13)
WTP

Sustainability 2015, 7 3076
3. Data Analysis
3.1. Description of Survey Data
This study primarily focused on the WTP of economy class airline passengers for the airline
carbon-offset policy. A total of 505 questionnaires was distributed at Taiwan’s international airport from
1 July to 31 August 2011. All of the questionnaires were recovered, and after screening, there was a total
of 477 valid questionnaires. The study subjects primarily consisted of airline passengers boarding
economy class.
The first part of the questionnaire included socioeconomic variables and included 11 items, such as
gender, age, educational level, monthly personal income, average annual number of flights, the main
reason for air travel, the name of the airline company, and whether the passengers are members of an
airline frequent flyer program and the flight class. Because the carbon-offset policy has not been
formally implemented in Taiwan, few airline passengers are well informed regarding this policy. For the
respondents to fully understand this policy—to investigate the level of understanding of the passengers
regarding the carbon-offset policy and their willingness to participate in it and to enhance the accuracy
rate in answering the subsequent questions—we provided in the second part of the questionnaire an
example using the Hong Kong–Taiwan flight described on the Cathay Pacific website. This example
informed the respondents of the distance and the duration of the trip and explained how much an
individual needs to donate to offset the carbon emissions generated by the one-way journey.
The questionnaire’s third part addressed the WTP for the airline carbon-offset policy. In our study, this
part investigated the validity of the WTP of the respondents and used the triple-bound inquiry to sharpen
the interval boundaries received in the single bounded format and, hence, increase statistical efficiency
(Possible strategic incentive effects in stated preferences when using donation public goods, while
important, are not investigated in this study. Interested readers can refer to Carson [34]). The
respondents were asked to respond in accordance with the triple-bound price scenarios, and in the end,
the questionnaire included the price for seven scenarios, and two prices provided by the respondents
elicited by an open format after three subsequent YES or NO responses (Figure 1). The starting value of
NT$ 20 (US $1 = NT $30) was set according to the amount of money that passengers of economy class
travelling from Taiwan to Hong Kong (the total distance was 829 kilometres, and the flight time was
90 min) would have to donate to offset the journey’s carbon emissions, and 50% of this value was used
as the increment or decrement of the set WTP.
Figure 1. WTP of airline passengers for carbon offset under triple-bound price scenarios.
Note: NTD = New Taiwan dollar.



Sustainability 2015, 7 3077
3.2. Statistical Analysis
The results of the data analysis of the WTP of economy class air passengers for carbon offsets are
shown in Table 1. Because the carbon-offset policy has not been widely implemented globally and many
passengers have not actually participated in the policy, we predicted that there would be many zero
WTPs in the survey. Therefore, the table divided the samples into two categories, one including a WTP
equal to 0 and another excluding a WTP equal to 0. To reflect the carbon-offset more acceptable to all
passengers, the WTP including 0 is more preferred to the one excluding 0. That is the reason that we
adopted the Spike model which can capture the effects of zero WTP. The p-value in the left column is
used for testing whether there are significant differences in WTP between WTP including 0 and WTP
excluding 0.
Table 1. Analysis of the WTP of economy class air passengers for carbon offsets.
Va ri ab l es
WTP including 0
No. of observations = 477
WTP excluding 0
No. of observations = 380
Sample size % WTP Sample size % WTP
Gender
P value = 0.556
DF = 1
Female 310 65 42.5 260 68.4 50.7
Male 167 35 35.8 120 31.6 49.8
Age *
P value = 0.037
DF = 3
30 or younger 163 34.2 39.6 144 37.9 44.9
31–40 92 19.3 26.7 65 17.1 37.8
41–60 202 42.3 47.2 152 40.0 62.7
61 or more 20 4.2 35.3 19 5.0 37.1
Educational level *
P value = 0.003
DF = 2
High school or below 116 24.3 34.7 81 21.3 49.7
College degree 270 56.6 36.4 229 60.3 42.9
Master’s degree or above 91 19.1 58.2 70 18.4 75.7
Average monthly
personal income *
P value = 0.000
DF = 4
Less than NT$ 30K 172 36.1 42.9 150 39.5 49.2
NT$30K to NT$50K 137 28.7 29.5 108 28.4 37.5
NT$50K to NT$70K 108 22.6 37.3 72 18.9 55.9
NT$70K to NT$90K 38 8.0 78.7 35 9.2 85.4
NT$90K to NT$110K 22 4.6 32.5 15 3.9 47.7
Average annual number
of flights *
P value = 0.000
DF = 2
1–3 385 80.7 33.4 296 77.9 43.4
3–6 42 8.8 59.2 38 10 65.4
7 or more 50 10.5 77.6 46 12.1 83.2
Purpose of flight *
P value = 0.000
DF = 3
Business 42 8.8 32.2 29 7.6 46.6
Tourism 345 72.3 33.9 266 70 44.0
Visiting friends and family 61 12.8 83.8 59 15.5 86.6
Other 29 6.1 34.1 26 6.8 38.1
Member of airline
frequent flyer program
P value = 0.625
DF = 1
Yes 38 8 33.2 30 7.9 42
No 439 92 40.8 350 92.1 51.1
Notes: * significant at P value = 0.10; DF: degree of freedom.
It can be observed from the table that the overall samples were mostly female (65.0%), who had a
relatively high WTP for carbon offsets. Further analysis found that there were no significant differences

Sustainability 2015, 7 3078
between the two sexes in average WTP. Most of the passengers were within the 41–50 year age group,
followed by 21–30 years, accounting for 37.1% and 28.7% of all passengers, respectively. Our analysis
revealed that passengers of different age groups differed significantly in average WTP, with more than
51 years old the highest (WTP = NT $56.0). Overall, the WTP of the samples including 0 was between
NT $26.7 and NT $56.0, and that of the samples excluding 0 was between NT $37.8 and NT $61.5.
However, we did not observe a trend for the WTP to increase with age, and we only determined that the
WTP of passengers older than 41 years was higher than that of passengers less than 40 years. The
majority of the passengers had college-level educations (56.6% of all passengers). The analysis results
indicated that there were significant differences in different educational levels of air passengers paid
average price of carbon offsets, the average WTP increasing with a higher education level.
For passengers with a master’s degree or higher, samples that included a WTP of 0 had an average WTP
of NT $58.2, whereas samples that excluded 0 had a WTP as high as NT $75.7. The largest share of the
passengers had an average income of NT$40,001–NT $60,000 (31.4% of all passengers). There were
significant differences in the average WTP between different income levels, and passengers with
incomes over NT $60,001 had the highest WTP of NT $73.9.
Most of the passengers had an average annual number of flights of one to two flights, and this group
accounted for 80.7% of all passengers. This group had the lowest WTP regardless of whether the WTP
contained 0. However, according to the results showed that different average number of flights, there are
significant differences in WTP. As the annual flight number increased, the WTP also increased. When
the annual number of flights was seven or more, the WTP including 0 was NT $77.6 and NT $83.2 when
excluding 0. The purpose of the journey was primarily tourism, accounting for 72.3%. Passengers with
different trip purposes also had significantly different WTP. Those visiting friends and relatives were
willing to pay the maximum amount of the carbon offset (NT $83.8). It was not surprising that the
business travelers had the lowest WTP (including 0). This could due to the fact that most of business
travelers cannot claim carbon contributions. The result may suggest a possible way to raise the WTP of
this group by reimbursing the expense by their companies. In addition, 92.0% of the surveyed airline
passengers were not airline frequent flyers, and most of the members of airline frequent flyer programs
were China Airlines (CI) members. The variables with statistically significant results in this part of the
analysis were further verified in the subsequent model.
Table 2 shows the proportions of responses to bids or stated open ended WTPs for the carbon offset
for different airline passengers. According to this study’s design, the starting amount of money was
NT $20. Figure 1 shows that based on the study’s WTP scenarios, the final WTP could be the amounts
set in the scenarios, such as NT $5, NT $10, NT $15, NT $20, NT $25, NT $30 and NT $35, and the rest
of the amounts could be the WTPs provided by the respondents from an open ended format.
The table shows that 29.6% of airline passengers were unwilling to pay more than NT $20 for the
carbon offset. Of these passengers, 20.5% were unwilling to pay for any carbon emissions offset
(WTP = 0), accounting for the highest proportion of all WTPs for the carbon offset. In addition, 70.4% of
the passengers were willing to pay more than NT $20 for the carbon offset, and the passengers who were
willing to pay NT $100 accounted for the highest percentage of this group of passengers (17.2%),
followed by passengers willing to pay NT $50 (14.7%) and passengers willing to pay NT $30 (10.5%).
This result indicates that for airline passengers who initially refused to pay the starting amount, most
were unwilling to pay for any carbon offset, and for passengers who paid the starting amount, the largest
share was willing to pay more than NT $35 (40.6%).

Sustainability 2015, 7 3079
Table 2. WTPs for carbon offset paid by different airline passengers.
Carbon-offset WTP Frequency % Carbon-offset WTP Frequency %
0 * 97 20.5 45 * 2 0.4
5 19 4.0 50 * 70 14.7
10 7 1.5 55 * 1 0.2
15 17 3.6 70 * 1 0.2
20 47 9.9 80 * 1 0.2
25 45 9.4 100 * 82 17.2
30 50 10.5 150 * 1 0.2
35 32 6.7 300 * 2 0.4
40 * 2 0.4
*: stated open ended WTPs.
4. Model Estimation Results
This study constructed a Spike model to analyse the WTP of economy class airline passengers for the
carbon-offset, and we primarily incorporated into the model the statistically significant variables from
the previously mentioned test. The definitions of the variables are shown in Table 3. The variables can be
divided into two categories: continuous and categorical. The continuous variables are divided into five
types. One type was the airline passenger bid price (bid). This variable was the bid price at the third layer
of the questionnaire scenarios, and it could be an open amount higher or lower than the bid price, which
was then incorporated into the model. The categorical variables included “willingness to participate in
the airline carbon-offset policy”, “The airline carbon-offset policy helps to reduce carbon emissions”
and “It is risky to pay for the carbon-offset online using a credit card”, the educational level and the
average annual number of flights. In addition, based on prior knowledge, we hypothesised the positive
and negative signs of the expected directions of variables. For example, the expected coefficients of the
“Airline passenger bid price” and “It is risky to pay for the carbon-offset online using a credit card” were
negative, and the expected coefficients of the remaining variables were positive. The expected direction
of the variables reflected that travellers that did not trust the security of online credit card payment were
unwilling to pay extra for the carbon-offset, and we could use the variables in the table to perform the
model estimation for this study.
Table 3. Variable definitions.
Variable Variable range Value Expected direction
Airline passenger bid price (bid) Actual value 0–1000 −
Educational level College degree or above 1 (otherwise, 0) +
Average annual number of flights taken 7 times or more 1 (otherwise, 0) +
Willing to participate in the carbon-offset
policy after obtaining the information Yes 1 (otherwise, 0) +
High willingness to participate in the airline
carbon-offset policy Agree, Strongly agree 1 (otherwise, 0) +
The airline carbon-offset policy helps to
reduce carbon emissions Agree, Strongly agree 1 (otherwise, 0)+
It is risky to pay for the carbon-offset online
using a credit card Agree, Strongly agree 1 (otherwise, 0)−

Sustainability 2015, 7 3080
Spike Model Estimation Results
Based on the basic statistical analysis, the statistically significant variables of the average WTP of
airline passengers for the carbon-offset were incorporated into the model. The Spike model was used to
construct the related WTP model, and the results are shown in Table 4. The model estimation can be
described in two parts. One part is the univariate model that only considers the scenario price. It can
investigate the WTP considering only the scenario price. The other part integrated the effects of other
variables. It can investigate other influences. The resulting WTP is primarily used to analyse the WTP of
paying for the carbon-offset of different airline passengers.
Table 4. Spike model results.
Variable Univariate Multiple
Constants 1.08 (10.29) −1.13 (−4.49)
Scenario price (bid) −0.13 (−16.38) −0.42 (−9.77)
College degree or above 0.52 (2.16)
Annual number of flights was more than 7 3.12 (6.55)
Willing to participate in the carbon-offset policy
after understanding its purpose and content 1.56 (6.55)
High level of willingness to participate in the airline carbon-offset policy 1.25 (5.53)
Believe that the airline carbon-offset policy helps to reduce carbon emissions 0.49 (2.33)
Believe that it is risky to pay for carbon-offset online using a credit card −0.73 (−2.71)
Mean WTP (NT$)
Median WTP (NT$)
WTP 95% CI
44.80 (7.48)
42.84 (5.16)
32.82–56.78
39.05 (4.86)
37.98 (6.21)
22.98–55.11
Spike value 0.253 (12.74) 0.272 (3.99)
Wald statistic (p-value) 806.322 (0.00) 629.687 (0.00)
Log likelihood function −565.278 −447.454
Number of observations 477
Note: t-values in parentheses.
This study used a Spike model that includes the 0 bid to analyse the WTP of airline passengers.
The results of the model indicated that the average univariate WTP was NT $44.80 and that the average
WTP was NT $39.05 for the multiple model. Therefore, when the model only considered the scenario
prices, the socioeconomic conditions of the airline passengers and their behavioural intention with
respect to the carbon-offset policy were not considered, which could result in overestimating the WTP.
When the multiple model was used, the resulting WTP could better represent the WTP for the
carbon-offset of airline passengers.
The coefficients of scenario price were negative, meaning that assuming when airline passengers
participate in the carbon-offset policy, they must pay an additional fee on top of the ticket price for the
offset of carbon emissions produced during the flight. Therefore, the higher is the fee, the more reluctant
are the airline passengers to pay for it, which agrees with the expected direction of this study.
The passengers with an average annual number of flights higher than seven were also more inclined to
pay for the carbon-offset. We reason that these passengers are more capable of paying for the flight costs
and, moreover, that paying for the additional carbon-offset does not affect their finances. Therefore,

Sustainability 2015, 7 3081
these passengers are more willing to pay for the carbon-offset. The finding is consistent with that of
Lu and Shon [19].
By observing the coefficients of behaviour intention of airline passengers in the model, we find that
when airline passengers are “willing to participate in the carbon-offset policy after understanding its
purpose and content” and have “a high level of willingness to participate in the airline carbon-offset
policy”, there is a positive impact on the model, suggesting that although the carbon-offset policy is not
yet a mandatory policy in Taiwan, the willingness of the respondents to pay extra for the carbon-offset
can be enhanced through explanation of the purpose and content of the policy. When the passengers
realised that “the airline carbon-offset policy helps to reduce carbon emissions”, they were more willing
to pay for the carbon-offset. Therefore, it is not only necessary to promote the carbon-offset policy, but
more importantly, it is necessary to increase the environmental awareness of passengers, which could
help to offset carbon emissions generated during the flight. The above results are consistent with the
findings of Lu and Shon [19], and Carrus et al. [35].
However, because only online donation was offered as the payment method in the sample offset plan,
when airline travellers “believe that it is risky to pay for the carbon-offset online using a credit card”,
passengers were less willing to pay for the carbon-offset. This outcome led us to infer that the use of an
Internet platform as a way to process donation remains the key that influences the participation of airline
passengers in the carbon-offset policy.
5. Conclusions and Recommendations
Through the donation method used in the actual implementation of the carbon-offset policy, this
study informed the economy class airline passengers regarding the carbon emissions generated by their
trip and investigated the WTP model of airline passengers for the carbon-offset based on the amount of
money that the respondents were willing to donate for the carbon-offset. The data analysis showed that
economy class airline passengers with a WTP of 0 accounted for 20.5% of all passengers. Therefore, this
study first used the Spike model to estimate the WTP of economy class airline passengers for the
carbon-offset policy to avoid the bias caused by excessive numbers of 0. In addition, when the donation
amount of the Taiwan-Hong Kong flight from the airline company was used as the starting bid in the
contextual assumptions, 70.4% of the airline passengers were willing to pay more than NT $20 to offset
the carbon emissions generated during the journey.
The model estimation results suggest that compared with the model that considers scenario prices
alone, taking into account the socioeconomic conditions and behavioural intention of airline passengers
can result in a more representative WTP for the carbon-offset. The average WTP was NT $39.05, which
is higher than the standard carbon-offset donation amount for economy class passengers calculated by
airlines (The value higher than that which the airlines have reported may be related to the fact that the
questionnaire only asked people how much they would be willing to pay whereas the airlines were
asking for the donation). The main variables of the socioeconomic conditions that affect the WTP of
airline passengers are the average monthly personal income and the number of flights.
The main variable of behavioural intention is “willing to participate in the carbon-offset policy after
understanding its purpose and content”, which suggests that airline companies should make efforts to
explain and promote the carbon-offset policy. The results support the findings of Lu and Shon [19],

Sustainability 2015, 7 3082
Mair [36] and Carrus et al. [35] in the following aspects. First, airlines companies can highlight that
protecting the environment is everyone’s obligation through media and promotion tools. Second, airlines
can enhance the passenger’s willingness to participate in carbon-offset policy by conveying positive
messages that individuals can improve the environment. Third, the passengers’ perceptions of the
effectiveness of the carbon-offset policy and the awareness of their duty towards the environment
determine their WTP.
In addition, Mair [36] also mentioned the pros and cons of recognising the carbon-offset policy as a
viable tool for offsetting carbon emissions and mitigating climate change. Suggestions for how this may
be achieved include making the purchase of carbon-offset mandatory, building in the price of the offset
in the price of the ticket, or making the airlines themselves responsible, rather than allowing them to pass
the costs of emission reduction on to their consumers.
Because of factors such as time and personnel, this study did not conduct a questionnaire survey
among foreign airline travellers. We recommend that future studies could expand the study’s scope by
conducting a questionnaire survey among airline passengers on flights outside Asia. This approach
could provide airline companies that implement the carbon-offset more information on the attitude of
foreign travellers towards this policy. In addition, future studies can design scenarios based on the
carbon-offset policies for different flight classes and different routes, making the study more in line with
the actual level of willingness to pay of the carbon-offset policy among airline passengers in each flight
class and to assess the impact of price changes and payment methods on the carbon-offset policy. It shall
also be noted that the spike model used in this study implicitly assumes that the truncation at zero does
not miss instances of negative WTP among the population. However, some airline passengers may have
a negative WTP for emission offsets. Finally, the administration of the survey-based data collection may
encounter one of the following issues: selection bias (or non-response bias), social desirability bias, and
common methods bias. An experimental setting might be an alternative way to survey-based data
collection [26,37].
Acknowledgments
This study was supported by the Ministry of Science and Technology (Project No. NSC
101-2410-H-260-053-MY28).
Author Contributions
Rong-Chang Jou designed the study and responded to the comments raised by the reviewers.
Tzu-Ying Chen analyzed the data and conducted the research. Both the co-authors drafted and revised
the manuscript together. Both authors read and approved the final manuscript.
Conflicts of Interest
The authors declare no conflict of interest.

Sustainability 2015, 7 3083
References
1. UNFCCC. Kyoto Protocol. Available online: http://unfccc.int/kyoto_protocol/items/2830.php
(accessed on 9 March 2015).
2. UNFCCC. Copenhagen Accord. Available online: http://unfccc.int/meetings/copenhagen_dec_
2009/items/5262.php (accessed on 15 June 2010).
3. Commission of the European Communities. Annex to the Communication from the Commission:
Reducing the Climate Change Impact of Aviation—Impact Assessment. September 2005.
Available online: http://ec.europa.eu/clima/policies/transport/aviation/docs/ia_aviation_en.pdf
(accessed on 12 March 2015).
4. Kotchen, M.; Moore, M. Private provision of environmental public goods: Household
participation in green-electricity programs. J. Environ. Econ. Manag. 2007, 53, 1–16.
5. Simone, N.; Stettler, M.; Barrett, S. Rapid estimation of global civil aviation emissions with
uncertainty quantification. Transp. Res. Part D 2013, 25, 33–41.
6. ICAO. ICAO Environmental Report 2010. Aviation and Climate Change. Available online:
http://www.icao.int/environmental-protection/Documents/Publications/ENV_Report_2010.pdf
(accessed on 20 March 2011).
7. MacKerron, G.J.; Egerton, C.; Gaskell, C.; Parpia, A.; Mourato, S. Willingness to pay for carbon
offset certification and co-benefits among (high-)flying young adults in the UK. Energy Policy
2009, 37, 1372–1381.
8. Anger, A. Including aviation in the European emissions trading scheme: Impacts on the industry,
CO2 emissions and macroeconomic activity in the EU. J. Air Transp. Manag. 2010, 16, 100–105.
9. Krammer, P.; Dray, L.; Köhler, M. Climate-neutrality versus carbon-neutrality for aviation
biofuel policy. Transp. Res. Part D: Transp. Environ. 2013, 23, 64–72.
10. Hofer, C.; Dresner, M.; Windle, R. The environmental effects of airline carbon emission taxation
in the US. Transp. Res. Part D: Transp. Environ. 2010, 15, 37–45.
11. Werner, R. Climate change and contribution of transport: Basic facts and the role of aviation.
Transp. Res. Part D: Transp. Environ. 2010, 15, 5–13.
12. Jardine, C.N. Calculating the Environmental Impact of Aviation Emissions. 2005. Available online:
http://climatecare.org/wordpress/wp-content/uploads/2013/07/Calculating-the-Environmental-Imp
act-of-Aviation-Emissions.pdf (accessed on 31 March 2011).
13. Gössling, S.; Haglund, L.; Kallgren, H.; Revahl, M.; Hultman, J. Swedish air travellers and
voluntary carbon offsets: Towards the co-creation of environmental value? Curr. Issues Tour.
2009, 12, 1–19.
14. Hooper, P.; Daley, B.; Preston, H.; Thomas, C. An Assessment of the Potential of Carbon Offset
Schemes to Mitigate the Climate Change Implications of Future Growth of UK Aviation. 2012.
Available online: http://www.cate.mmu.ac.uk/wp-content/uploads/2012/06/5-Final-Report-
Potential-Carbon-Offsetting-to-Mitigate-Clima.pdf (accessed on 22 March 2011).
15. Van Birgelen, M.; Semeijn, J.; Behrens, P. Explaining pro-environment consumer behaviour in air
travel. J. Air Transp. Manag. 2011, 17, 125–128.

Sustainability 2015, 7 3084
16. Casaló, L.V.; Flavián, C.; Guinalíu, M. Determinants of the intention to participate in firm-hosted
online travel communities and effects on consumer behavioral intentions. Tour. Manag. 2010, 31,
898–911.
17. Chen F.Y. The intention and determining factors for airline passengers’ participation in carbon
offset schemes. J. Air Transp. Manag. 2013, 29, 17–22.
18. Brouwer, R.; Brander, L.; van Beukering, P. A convenient truth: Air travel passengers’
willingness to pay to offset their CO2 emissions. Clim. Chang. 2008, 90, 299–313.
19. Lu, J.L.; Shon, Z.Y. Exploring airline passengers’ willingness to pay for carbon offsets. Transp.
Res. Part D: Transp. Environ. 2012, 17, 124–128.
20. Carson, R.T.; Czajkowski, M. The discrete choice experiment approach to environmental
contingent valuation. In Handbook of Choice Modelling, 1st ed.; Hess, S., Daly, A., Eds.; Edward
Elgar: Northampton, MA, USA, 2014; pp. 202–235.
21. Halstead, J.M.; Luloff, A.E.; Stevens, T.H. Protest bidders in contingent valuation. Environ.
Resour. Econ. 1992, 21, 131–150.
22. Whitehead, J.C.; Groothuis, P.A.; Blomquist, G.C. Testing for non-response and sample selection
bias in contingent valuation: Analysis of a combination phone/mail survey. Econ. Lett. 1993, 41,
215–220.
23. Freeman, A.M. On assessing the state of the art of the contingent valuation method of valuing
environmental changes. In Valuing Environmental Goods: An Assessment of the Contingent
Valuation Method, 2nd ed.; Cummings, R.G., Brookshire, D.S., Schulze, W.D., Eds.; Rowman &
Allanheld: Totowa, NJ, USA, 1986; pp. 148–161.
24. Vaughan, W.J.; Russell, C.S.; Darling, A.H.; Rodriguez, D.J. Willingness to Pay: Referendum
Contingent Valuation and Uncertain Project Benefits; Inter-American Development Bank,
Sustainable Development Department Technical Papers Series; Inter-American Development
Bank: Washington, DC, USA, June 1999.
25. Calia, P.; Strazzera, E. A sample selection model for protest responses in contingent valuation
analyses. Statistica 2001, 61, 473–485.
26. Diederich, J.; Goeschl, T. Willingness to pay for voluntary climate action and its determinants:
Field-experimental evidence. Environ. Resour. Econ. 2014, 57, 405–429.
27. Kristroöm, B. Spike models in contingent valuation. Am. J. Agric. Econ. 1997, 79, 1013–1023.
28. Bengochea-Morancho, A.; Fuertes-Eugenio, A.M.; Del Saz-Salazar, S. A comparison of empirical
models used to infer the willingness to pay in contingent valuation. Empir. Econ. 2005, 30, 235–244.
29. Hu, W. Use of spike models in measuring consumers’ willingness to pay for non-GM oil.
J. Agric. Appl. Econ. 2006, 38, 525–538.
30. Jou, R.C.; Chiou, Y.C.; Chen, K.H.; Tan, H.I. Freeway drivers’ willingness-to-pay for a
distance-based toll rate. Transp. Res. A-Pol. 2012, 46, 549–559.
31. Jou, R.C.; Wang, P.L. The intention and willingness to pay moving violation citations among
Taiwan motorcyclists. Accid. Anal. Prev. 2012, 49, 177–185.
32. Saz-Salazar, S.D.; Garcia-Menendez, L. Willingness to pay for environmental improvements in a
large city. Environ. Resour. Econ. 2001, 20, 103–112.
33. Yoo, S.H.; Kwak, S.J. Using a spike model to deal with zero response data from double bounded
dichotomous choice contingent valuation surveys. Appl. Econ. Lett. 2002, 9, 929–932.

Sustainability 2015, 7 3085
34. Carson, R.T.; Groves, T. Incentive and Informational Properties of Preference Questions. Environ.
Resour. Econ. 2007, 37, 181–210.
35. Carrus, G.; Passafaro, P.; Bonnes, M. Emotions, habits and rational choices in ecological behaviours:
The case of recycling and use of public transportation. J. Environ. Psychol. 2008, 28, 51–62.
36. Mair, J. Exploring air travellers’ voluntary carbon offsetting behaviour. J. Sustain. Tour. 2011, 19,
215–230.
37. Löschel, A.; Sturm, B.; Vogt, C. The demand for climate protection—Empirical evidence from
Germany. Econ. Lett. 2013, 118, 415–418.
© 2015 by the authors; licensee MDPI, Basel, Switzerland. This article is an open access article
distributed under the terms and conditions of the Creative Commons Attribution license
(http://creativecommons.org/licenses/by/4.0/).