Estimating Willingness to Pay for Improving Municipal Solid
Waste Management Using a Choice Experiment: A Case
Study of Central China
Z J Zhang
*
School of Economics, Zhejiang Gongshang University, Hangzhou, China
Abstract. To minimize the mismatch between the inadequate capacity of municipal solid
waste management (MSWM) services caused by substantially increasing municipal solid
waste and the rising demand for higher environmental quality in central China, this paper
employed choice experiment method to estimate the willingness to pay of households to
improve MSWM services. Based on household survey data from two cities, the estimation
results reveal that households, in general, are willing to contribute extra money for the
improvement of MSWM services and there is considerable preference heterogeneity across
households and cities. Households in Nanchang tend to focus on frequency of waste
transportation while households in Anqing prefer frequency of waste collection. Household
income, household size, householder age, and especially environmental attitude, are
important sources of preference heterogeneity. On an average, households in Nanchang and
Anqing are willing to pay 25.632 CNY and 13.275 CNY per month for the improved MSWM
services respectively.
1. Introduction
Municipal solid waste management (MSWM) is a major challenge in urban areas throughout the
world, especially in the rapidly growing cities of developing countries [1, 2]. As the largest
developing country, rapid economic development, rising urbanized population and changed life style
have substantially accelerated the volume of municipal solid waste (MSW) generated in China [3, 4].
Almost two thirds of cities are besieged by garbage, most MSW is disposed in landfills and the
remainder is covered or heaped in China [5, 6]. Open-air stinking wastes that have not yet been
disposed of in many residential areas provide breeding grounds for houseflies, mosquitoes, vermin,
and mice. Environmental pollution-related and infectious diseases arising from poor waste
management still account for a non-negligible position in ill-health and death. Waste has become a
severe threat to public health and environmental quality, and even a serious social problem.
Unfortunately, constrained by the limited financial resource from government alone, the capacity for
waste management has not kept pace with the growth rate of waste generation. As safe disposal in
1990 was highly limited, the safe disposal rate reached only 53% in 2006 [7]. Accomplishing an
effective MSWM system should be a priority for the governments of all cities in China in the years to
come [8].
382
Zhang, Z.
Estimating Willingness to Pay for Improving Municipal Solid Waste Management Using a Choice Experiment: A Case Study of Central China.
In Proceedings of the International Workshop on Environmental Management, Science and Engineering (IWEMSE 2018), pages 382-392
ISBN: 978-989-758-344-5
Copyright © 2018 by SCITEPRESS – Science and Technology Publications, Lda. All rights reserved

A considerable number of studies have attempted to analyse MSWM in China [7-16]. Most of
these studies either are correlated with the current situation, challenges and suggestions in regard to
MSWM systems across regions, or focus on evaluating the MSWM services of different areas from a
macro perspective. These important studies indicate the direction for the development of MSWM
through a supply-driven approach. However, to what extent does current waste management system
provide services that matched households‘ preferences? It‘s a critical issue that local authorities
should design the MSWM systems considering the preferences or willingness to pay (WTP) of
households for the characteristics of services and the technology restrictions of service providers.
Recently, some researches have turned to analyse the MSWM services using the demand-responsive
approach. For instance, Chu and Xi [17] found that households have different preferences over
municipal solid waste collection (e.g., collection frequency and location of containers). Even though
they have to pay extra money privately, most households are willing to improve MSW source
separation facilities [18]. In a closely related paper, Wang and He [19] employed a contingent
valuation method to estimate the households‘ WTP for an improved MSWM program. They found
that a household in Eryuan (a county in southwest China) is willing to pay approximately 17.1 CNY
per month for the improved MSWM program on average. This important work, however, does not
disentangle what characteristics or attributes of MSWM services households are more willingness to
pay and are there preference heterogeneities on improving MSWM services across households and
areas.
This study extends the literature by applying a stated preference choice experiment (CE)
technique to estimate households‘ willingness to pay for improving MSWM services in two cities of
central China. Specifically, the detailed objectives of this research are listed below.
(1) Elicit households‘ valuation for the improved MSWM services and implicit price for each
attribute.
(2) Identify the factors that influence households‘ WTP for improving MSWM services.
(3) Carefully examine the preference heterogeneities across households and areas and explain
the potential source of heterogeneity.
The contribution of this study to the literature is twofold. Firstly, although six studies have
employed CE method to estimate households‘ WTP for improving MSWM services [20-25], to our
best knowledge, this is the first CE study taking preference heterogeneity into consideration on the
economic value of MSWM services through random parameter logit model. Secondly, the CE
presented in this paper is the only one study that evaluates households‘ WTP for improving MSWM
services in mainland China. Consequently, facing severe challenge of increasing MSW and budget
constraints, the non-market valuation can timely provide the authorities with demand-side
information to prioritise the MSWM policy instruments which are preferred by most residents.
2. Methods
2.1. The choice experiment
The choice experiment method is consistent with utility maximisation and demand theory [26], which
has been widely used in non-market valuation. The CE method assumes the observable utility
function is linear in parameters and additively separable, and the probability that one individual
chooses a particular alternative is a function of the attributes of that alternative and the characteristics
of that individual [27]. Different assumptions of the distribution of random error term yield different
choice models. When assuming the random error term is independently and identically distributed
(IID) across alternatives and individuals, standard multinomial logit (MNL) model is obtained. But
when IID assumption is violated, random parameter logit (RPL) model can be used.
Once parameters of the above choice model are estimated, implicit price for a change in the level
of a single attribute can be obtained by dividing the coefficient of that attribute by the coefficient of
Estimating Willingness to Pay for Improving Municipal Solid Waste Management Using a Choice Experiment: A Case Study of Central
China
383

payment attribute [28]. When all attributes have increased their levels from initial state to subsequent
state, overall WTP for an improved scenario can be calculated.
2.2. Experiment design
This process began with gathering opinions from waste management official, experts in
environmental issues and urban residents along with reviewing the literature. After a collection of
background information, a series of focus group discussions were conducted to identify potential
perspectives upon the attributes of MSWM and their levels as well as other related issues. Four
attributes of MSWM are included in final CE survey (see Table 1). Generally, MSWM also has an
attribute of waste disposal, but in pretest survey local households universally express little concern
about waste disposal and think waste disposal are too far away from them and would not affect their
quality of life. The suggested attribute levels are displayed in Table 1.
Table 1. Attributes and their levels of MSWM.
Attributes
Levels
Waste classification
No waste classification needed
a
Waste classification and free containers from government
Frequency of waste collection
Once a day
a
Twice a day
Frequency of waste transportation
Once a day
a
Twice a day
Fee
5 CNY per month
a
10 CNY per month
15 CNY per month
Note:
a
The current level of each attribute.
Once attributes and attribute levels were determined, a fractional factorial was conducted [27, 29].
Using the orthogonal design in SPSS 20.0, 12 choice sets were left after dominant and implausible
alternatives were removed from the design, but orthogonality was reserved. The 12 choice sets were
averagely divided into two blocks, and each respondent was exposed to one version only. Each
choice set contains two alternatives and an option to keep the status quo.
A pilot test was conducted with a group of 20 respondents to check for respondents‘
understanding on the MSWM, choice context, adequacy of the attributes and levels considered, and
other factors such as the wording, format and length of the questionnaire. Apart from choice
experiment module, the final version of questionnaire included another two modules. Before choice
experiment module, questionnaire contained questions on the respondents‘ knowledge, attitudes and
awareness toward environment and MSWM in general. During choice experiment period, we
reminded respondents that: please choose each choice set separately; when choose alternative A or B,
you have to pay an incremental monthly fee for the improved MSWM services in addition to your
existing payment; the money you spend on this additional fee would not be available for other
household expense. And the last module consisted of items on socio-economic information of
respondents.
2.3. Study sites and data collection
The study sites were two inland cities (Nanchang and Anqing), which are situated in central China.
Nanchang is the capital of Jiangxi province, which has a higher level of economic and social
development than Anqing. However, in those areas, the capacity of MSWM cannot keep pace with
IWEMSE 2018 - International Workshop on Environmental Management, Science and Engineering
384

the sharp increase of municipal solid waste generated. The limitation on financial budget and the
increasing real operational cost of MSWM have made it tougher to dispose of solid waste in many
communities. The heaped waste that has not received timely treatment poses a serious threat to
environmental quality and human health, especially in summer when the speed of waste rotting is
particularly fast. As a result, the demand of households for improving the MSWM services is
dramatically rising.
On July 5, 2013, after a systematic training session was held for the interviewers, the survey was
put into field. We employed face-to-face personal interviews as our survey mode in each city. In
order to limit interviewer bias the interviewers followed a random route procedure to select
respondents. The starting point was randomly determined by dicing and then every third household
was visited. Since waste is collected and paid for at the household level, we chose the household as
the unit of analysis and selected the head of household as respondent in each household. Overall, 240
attempts were made and 182 interviews were completed on August 20, 2013. After removing 7
protest bids (3 respondents indicated ―the government should pay for this‖, 3 respondents indicated
―I don‘t believe the policies would actually happen‖, and 1 respondent indicated ―the fee rises too
much‖), the remaining 175 questionnaires were used in the following analysis (see Hanley et al.) [30].
More information on the number of samples in each city was reported in Table 2.
Table 2. The number of samples in each city.
City
Population
a
Attempts
b
Interviews
c
Effective samples
Nanchang
5.04 million
120
86
82
Anqing
5.31 million
120
96
93
Note: a The data are from China’s sixth national census (2010); b The nu mber of p laned samples; c
The number of completed interviews.
3. Results and discussion
3.1. Social, economic and attitudinal characteristics of the respondents
Table 3. Descriptive statistics of respondents (N=226).
Variable
Nanchang
Anqing
Nanchang
mean
a
Sample
mean
Sample
std. dev.
Anqing
mean
a
Sample
mean
Sample
std. dev.
Gender
Male
52%
54%
0.50
51%
53%
0.50
Female
48%
46%
49%
47%
Marital
status
Married
78%
77%
0.42
81%
84%
0.34
Single
22%
23%
19%
16%
Age
35.49
38.46
12.46
37.28
40.57
12.37
Education
9.98
10.07
3.86
8.05
8.37
3.20
Household income
8.63
b
8.71
7.02
6.49
c
5.89
3.29
Household size
3.39
3.93
1.16
3.27
3.97
1.24
Environmental knowledge
–
2.01
1.19
–
1.43
1.32
Environmental awareness
–
3.74
0.44
–
3.55
0.50
Neighbors‘ environmental
awareness
–
2.80
0.82
–
2.69
0.79
Source: a China’s sixth national census, 2010; b Nanchang Statistics Bureau, 2014; c Anqing
Statistics Bureau, 2014.
Estimating Willingness to Pay for Improving Municipal Solid Waste Management Using a Choice Experiment: A Case Study of Central
China
385

The descriptive statistics of the respondents were presented in Table 3. Across the two sub-samples,
the social and economic characteristics of each sub-sample were similar to the city averages with the
exception of age and household size in both sub-samples and household income in Anqing sample.
The former was mainly due to the fact that the population mean was calculated based on all the
people in city while the sample only included heads of households. With respect to household size,
the sample mean was higher in each city because a certain proportion of persons in respondents‘
households were from countryside or even did not have ―hukou‖ (household registration). The
slightly lower income level of Anqing sample may be explained by the big gap between the rich and
the poor in Anqing and the sample was happening to consist of more poor people.
As regard to attitudinal characteristics, the environmental knowledge of respondents in Nanchang
was considerably more than that in Anqing. The respondents‘ environmental awareness and their
neighbors‘ environmental awareness in Nanchang were also higher than those in Anqing, but the
differences were small.
3.2. Estimation results of choice experiment
Initial descriptive statistical analysis showed that the sample was representative of the population.
Variables used in the choice models and their definition and coding were presented in Table A2.
Note that the software used to undertake these estimations was Nlogit 4.0.
3.2.1. The results of multinomial logit (MNL) model. Two different MNL models were estimated for
Nanchang and Anqing data sets (see Table 4). The first model was a basic specification that showed
the importance of the choice set attributes in explaining respondent‘s taste for different options, and
the second model, on the basis of the first model, additionally included socio-economic and
attitudinal variables by interactions with alternative specific constant (ASC).
Table 4. Parameter estimates from MNL model for Nanchang and Anqing data sets.
Variable
Nanchang
Anqing
Model 1
Model 2
Model 1
Model 2
ASC
1.628**
1.494
2.437***
-2.137
Wc
0.862***
0.898***
0.112*
0.075**
Fwc
0.322**
0.315**
0.620
0.625
Fwt
0.621***
0.573***
0.578**
0.353**
Fee
-0.368***
-0.365***
-0.524***
-0.504***
ASC*Gender
-0.175
0.013
ASC*Age
-0.236
-0.421*
ASC*Mar
-0.629
0.118
ASC*Edu
0.019
0.083*
ASC*Hinco
0.059**
0.092**
ASC*Hsize
0.110
0.093
ASC*Eknow
0.167*
-0.021
ASC*Eaware
1.363***
0.599**
ASC*Naware
0.381***
0.303*
Observations
492
492
558
558
Log-likehood
-320.870
-310.638
-421.159
-377.917
Pseudo R-square
0.252
0.276
0.227
0.306
Note: * significance at 10%; ** significance at 5%; and *** significance at 1%.
In model 1 for each data set, the coefficients on ASC and all attributes have expected signs and
are significant at the 10% level or less except the insignificant coefficient on ‗frequency of waste
IWEMSE 2018 - International Workshop on Environmental Management, Science and Engineering
386

collection‘ from Anqing data set, which indicates that respondents across the two cities, in general,
are concerned about the improvement of attribute levels and a relatively reasonable experimental
design we have. The model fit of model 2 for each data set is improved by incorporating the
interaction terms. Model 2 is highly consistent with model 1 in terms of signs and significance of
attribute coefficients. It reveals that assuming other factors remain constant, improving the levels of
‗waste classification‘, ‗frequency of waste collection‘ and ‗frequency of waste transportation‘ would
increase respondents‘ utility while increasing the fee would decrease their utility. But respondents
from different data sets have different preferences for attributes, e.g., respondents in Nanchang tend
to focus on ‗waste classification‘ and ‗frequency of waste transportation‘ while respondents in
Anqing are inclined to ‗frequency of waste collection‘. It should be noted that the coefficient on
‗frequency of waste collection‘ has the largest value but is not significantly different from zero at the
10% level in Anqing data set, and its accuracy and reliability deserve the following further study.
The interaction terms have different signs and significance across the two data sets. Model 2 for
Nanchang manifests respondents who have higher household income, more environmental
knowledge, higher awareness of environmental protection, or more positive subjective perception of
the environmental behaviors of their neighbors are willing to pay more for improving MSWM
services, ceteris paribus. As to Anqing data set, education, household income, environmental
awareness, and subjective perception of the environmental behaviors of neighbors have significantly
improved households‘ willingness to contribute money. These findings are consistent with previous
studies [20, 21, 31, 32]. However, older people in Anqing are more reluctant to pay, which is
intuitively correct since older people have low level of environmental awareness and are more
conservative in spending in China.
Despite widely used, MNL model has severe limitations with respect to the well-known
assumption of independence of irrelevant alternatives and its ability to capture random taste
heterogeneity across individuals [33, 34]. Hence, we will move to apply a less restrictive and more
flexible model to the following analysis.
3.2.2. The results of random parameter logit (RPL) model. As RPL model tends to be unstable and
identification issues arise when all coefficients vary over the population, the coefficient on fee is
fixed while other coefficients are allowed to vary in our analysis (see Goett et al., Revelt and Train)
[35, 36]. Considering some respondents may be satisfied with the status quo, e.g., respondents need
not have to spend time and energy on waste separation nor have to pay extra money, we specify the
coefficients on ‗waste classification‘, ‗frequency of waste collection‘ and ‗frequency of waste
transportation‘ to be normally distributed. Attributes showing insignificant standard deviation are
then respecified as nonrandom parameters. At last, ‗frequency of waste collection‘ and ‗frequency of
waste transportation‘ are specified as random parameters and other variables remain nonrandom
parameters in estimation models for the two data sets. To identify the potential sources of
heterogeneity, random parameters are interacted with socio-economic and attitudinal variables.
Initially, all possible interactions are put into estimation model for each city and finally only
interactions that are significant at the 10% level or less are retained. The estimation results are shown
in Table 5.
The estimation of RPL model results in considerable improvement of model fit over the MNL
model. All parameter estimates related to the attributes and ASC are statistically significant and have
the expected signs including the ‗frequency of waste collection‘ for Anqing data set, on which the
coefficient is insignificant in the MNL model. The ASC and ‗waste classification‘ are nonrandom
parameters for the two cities, implying that respondents generally agree to improve the services of
MSWM at the expense of extra spending and waste separation, which is out of our expectation. One
possible explanation is that, along with the increasing negative impacts of waste on residents and
Estimating Willingness to Pay for Improving Municipal Solid Waste Management Using a Choice Experiment: A Case Study of Central
China
387

vigorously publicizing waste separation from government and environmental organizations,
households are gradually aware of the necessity of improving MSWM services.
For Nanchang data, the RPL model reveals that, even though the respondents derive significantly
positive utility from ‗frequency of waste collection‘ and ‗frequency of waste transportation‘ on
average, the standard deviations for these two attributes are both significant, indicating that there are
heterogeneous preferences for these two attributes. The interaction results show that respondents with
higher household income demand MSWM program with higher level of ‗frequency of waste
collection‘. Those respondents having higher awareness of environmental protection prefer MSWM
program with higher level of ‗frequency of waste collection‘ and ‗frequency of waste transportation‘.
Finally, respondents having more environmental knowledge or more positively evaluate their
neighbors‘ environmental behaviors attach higher utility to MSWM program which has higher level
of ‗frequency of waste transportation‘. On the whole, it is environmental concern, rather than
household income, that contributes to explain the heterogeneous preferences for MSWM attributes.
Table 5. Parameter estimates from RPL model for Nanchang and Anqing data sets.
Variable
Nanchang
Anqing
Mean
Std. dev.
Mean
Std. dev.
Random parameters in utility function
Fwc
4.054**
0.827*
3.732*
2.721**
Fwt
4.636**
1.528***
1.689***
2.169**
Nonrandom parameters in utility function
ASC
1.686*
5.574***
Wc
1.056***
0.355*
Fee
-0.446***
-0.855***
Potential sources of heterogeneity
Fwc: Hinco
0.046*
0.165*
Fwc: Hsize
0.305*
Fwc: Eaware
0.703*
0.274***
Fwc: Age
-0.372*
Fwt: Edu
0.203*
Fwt: Hsize
0.546*
Fwt: Eknow
0.552**
Fwt: Eaware
2.561***
1.023**
Fwt: Naware
0.753*
0.387*
Observations
492
558
Log-likehood
-302.824
-366.307
Pseudo R-square
0.295
0.327
Note: * significance at 10%; ** significance at 5%; and *** significance at 1%.
In Anqing, the RPL model shows that the mean values and standard deviations for ‗frequency of
waste collection‘ and ‗frequency of waste transportation‘ are significant, suggesting the presence of
heterogeneity in preference for these two attributes. In addition, the standard deviation parameter for
‗frequency of waste transportation‘ exhibits considerable variability, meaning that some respondents
derive negative utility from higher level of this attribute. Respondents who have higher household
income, or who are younger, prefer MSWM program with higher level of ‗frequency of waste
collection‘. Whereas those who receive higher education, or who have more positive subjective
perception of the environmental behaviors of their neighbors tend to choose MSWM program that
has higher level of ‗frequency of waste transportation‘. Finally, household size and environmental
awareness significantly increase the preference of respondent for MSWM program with higher level
of both ‗frequency of waste collection‘ and ‗frequency of waste transportation‘. These findings are
similar to other scholars‘ conclusion that household size and environmental awareness effectively
IWEMSE 2018 - International Workshop on Environmental Management, Science and Engineering
388

improve the WTP of households for MSWM program with sufficient waste separated collection [18,
37]. Nevertheless, in contrast with the findings in Nanchang, environmental knowledge has played an
insignificant role in explaining attributes preference in Anqing, indicating that only high levels of
environmental knowledge matter.
3.3. Welfare estimates
Once we have obtained the results of choice model, welfare measures can be estimated. The implicit
prices for MSWM attributes are calculated using the results of MNL models (Model 2) and RPL
models (see Table 6). And the confidence intervals for the implicit price measures have been
calculated using the delta method.
Table 6. Implicit prices for attributes (in CNY) and 95% confidence intervals (95%
level).
Attribute
Nanchang
Anqing
MNL
RPL
MNL
RPL
Wc
2.460
(-0.508, 5.429)
2.368
(-0.542, 5.278)
0.149
(-1.930, 2.227)
0.415
(-1.299,
2.129)
Fwc
0.863
(-1.604, 3.330)
9.090
(-1.091, 19.270)
—
4.365
(-0.511,
6.079)
Fwt
1.570
(-0.245, 3.385)
10.395
(-0.194, 20.983)
0.700
(-0.601, 2.002)
1.975
(-7.843,
11.794)
Note: —The parameter estimate is not significantly different from zero at the 10% level;
Confidence intervals in parentheses.
As can be seen in Table 6, compared to the estimates of implicit prices derived from the MNL
models, the resulting implicit prices derived from the RPL models are consistently larger except the
implicit price for waste classification attribute in Nanchang data set. A similar result has also been
reported by Revelt and Train [36], and Sillano and Ortúzar [38], which is related to the fact that RPL
model decomposes the unobserved utility and normalizes parameters on the basis of part of the
unobserved utility. Furthermore, we can find larger confidence intervals for implicit prices in the
RPL models, corresponding to those in the MNL models, reflecting the substantial variations in
respondents‘ preferences for these attributes.
Table 7. Willingness to pay for the improved MSWM services (in CNY) and confidence
intervals (95% level).
Nanchang
Anqing
WTP for MSWM
25.632
(23.834, 27.431)
13.275
(9.648, 16.902)
Considering the MNL models have restrictive assumption and their model fits are worse than the
RPL models‘, our following analysis is based on the results of RPL models. An implied ranking of
attributes in terms of respondents‘ preference for each data set can be derived. ‗Frequency of waste
transportation‘ is the highest ranked attribute in Nanchang while ‗frequency of waste collection‘ is
the highest ranked attribute in Anqing. As ‗waste classification‘ requires households to spend time
and energy on waste separation at the household level, it is the lowest ranked attribute in the two data
sets. Furthermore, as showed in Table 7, on an average, households in Nanchang are willing to pay
Estimating Willingness to Pay for Improving Municipal Solid Waste Management Using a Choice Experiment: A Case Study of Central
China
389

25.632 CNY per month for the improved MSWM services while the WTP of each household in
Anqing is 13.275 CNY per month. The big gap on WTP for the improved MSWM services between
Nanchang and Anqing is caused by the imbalance development of the society and economy across
regions in China. In addition to relatively affluent income, more importantly, households in
Nanchang have a higher level of environmental concern, which has an appreciable impact on
increasing households‘ WTP for improving MSWM services (see Table 3, 4 and 5). In addition, we
also compared our WTP estimates with previous researches in other countries or areas (see Table 8).
There is some space for increasing the level of garbage treatment fee, which is both the precondition
for sustaining MSWM and the incentive mechanism for residential waste reduction.
Table 8. Comparison of WTP estimates.
Study
Country and area
WTP per household per month
Share
a
Nanchang (present study)
Mainland China
25.632 CNY (USD 4.18)
0.35%
Anqing (present study)
Mainland China
13.275 CNY (USD 2.16)
0.27%
Ku et al. [22]
Korea
1178-1918 KRW (USD 1.24-2.02)
0.03-0.05%
Afroz and Masud [31]
Malaysia
22 MYR (USD 6.89)
1.7%
Fonta et al. [32]
Nigeria
230 Naira (USD 1.8)
1.82%
Afroz et al. [39]
Bangladesh
13 Taka (USD 0.18)
0.12%
Jin et al. [40]
Macao
67-81 MOP (USD 8.33-10.15)
0.36-0.44%
Note: a denotes the share of average annual household inco me; The exchange rate between the
country’s currency and dollar is the exchange rate at that time.
4. Conclusions and Implications
This paper was motivated by providing policy-makers with additional information to adopt
appropriate MSWM measures to improve the poor environmental quality of urban households in
central China. In this study, the CE technique was employed to elicit households‘ preferences for
various improved MSWM alternatives in two cities.
The results from MNL models suggest that even though households have different preferences for
MSWM attributes across the two cities, households in each city are generally willing to contribute
extra money for MSWM services associated with higher levels of attributes. In order to obtain more
reliable results and figure out the presence of unobserved heterogeneity in the preference for MSWM
attributes, RPL model was specified for the respondents from each city. The findings support that the
possible sources of heterogeneous preferences for ‗frequency of waste collection‘ and ‗frequency of
waste transportation‘ are multiple household characteristics. Social-economic characteristics and
especially environmental attitude are critical factors that stimulate households willing to pay extra
money for improving attribute levels of MSWM services.
We further obtained implicit price measures for MSWM attributes and WTP measures for the
improved MSWM services of the two cities. Households in Nanchang are willing to pay more for the
higher level of ‗frequency of waste transportation‘ while households in Anqing are willing to pay
more for the higher level of ‗frequency of waste collection‘. In general, households in Nanchang and
Anqing are willing to pay 25.632 CNY and 13.275 CNY per month for the improved MSWM
services respectively. This information is important for the Environment Protection Agency to
determine which MSWM options will provide the greatest benefits to the widest households in the
two cities where financial resources are limited and conflicting development interests exist. However,
it is important to note that actual decision is a more complex process, which needs more cost-benefit
information and valuation criterion.
Acknowledgement
This research was supported by the General Program of the National Natural Science Foundation of
China (Grant No.71773114).
IWEMSE 2018 - International Workshop on Environmental Management, Science and Engineering
390

References
[1] Han H, Zhang Z, Xia S and Li H 2018 J. Environ. Plan. Manage. 61 568-596
[2] Moh Y and Manaf L A 2017 Resour. Conserv. Recycl. 116 1-14
[3] Cheng H, Zhang Y, Meng A and Li Q 2007 Environ. Sci. Technol. 41 7509-7515
[4] Hong J, Li X and Cui Z 2010 Waste Manage. 30 2362-2369
[5] Chung S S and Lo C W 2008 Waste Manage. 28 272-281
[6] Wang L A ,Pei T Q, Huang C and Yuan H 2009 Waste Manage. 29 2203-2208
[7] Chen X, Geng Y and Fujita T 2010 Waste Manage. 30 716-724
[8] Zhang D Q, Tan S K and Gersberg R M 2010 J. Environ. Manage. 91 1623-1633
[9] Hong R J, Wang G F, Guo R Z, Cheng X, Liu Q, Zhang P J and Qian G R 2006 Resour.
Conserv. Recycl. 49 129-146
[10] Jiang J, Lou Z, Ng S, Luobu C and Ji D 2009. Waste Manage. 29 1186-1191
[11] Li Z, Yang L, Qu X and Sui Y 2009 Waste Manage. 29 2596-2599
[12] Linzner R and Salhofer S 2014 Waste Manage. Res. 32 896-907
[13] Liu C, Wu X 2010 Waste Manage. Res. 29 371-378
[14] Ren X, Hu S 2014 Waste Manage. Res. 32 340-347
[15] Yuan H, Wang L A, Su F and Hu G 2006 Waste Manage. 26 1052-1062
[16] Zhu M, Fan X, Alberto R, He Q, Federico V, Liu B, Alessandro G and Liu Y 2009 Waste
Manage. 29 1227-1233
[17] Chu Z, Xi B, Song Y and Crampton E 2013 Habitat Int. 40 194-200
[18] Zhang W, Y Che, Yang K, Ren X and Tai J 2012 Waste Manage. Res. 30 1261-1271
[19] Wang H, He J, Kim Y and Kamata T 2014 Waste Manage. Res. 32 695-706
[20] Jin J, Wang Z and Ran S 2006 Waste Manage. Res. 24 301-309
[21] Karousakis K and Birol E 2008 J. Environ. Manage. 88 1099-1108
[22] Ku S, Yoo S and Kwak S 2009 Environ. Manage. 44 278-287
[23] Othman J 2007 Int. J. Manage. Stud. 14 189-212
[24] Pek C and Jamal O 2011 J. Environ. Manage. 92 2993-3001
[25] Sakata Y 2007 Waste Manage. 27 639-644
[26] Bateman I J, Carson R T, Day B, Hanemann W M, Hanley N, Hett T, Jones-Lee M, Loomes G,
Mourato S and Ozdemiroglu E 2003 Guidelines for the use of stated preference techniques
for the valuation of preferences for non-market goods (Cheltenham, UK: Edward Elgar)
[27] Louviere J J, Hensher D A and Swait J D 2000 Stated choice methods: analysis and
applications (Cambridge: Cambridge University Press)
[28] Boxall P C, Adamowicz W L, Swait J, Williams M and Louviere J 1996 Ecol. Econ. 18 243-
253
[29] Hensher D A, Rose J M and Greene W H 2005 Applied choice analysis: a primer (Cambridge:
Cambridge University Press)
[30] Hanley N, Colombo S, Mason P and Johns H 2007 J. Agr. Econ. 58 433-453
[31] Afroz R and Masud M M 2011 Waste Manage. 31 800-808
[32] Fonta W M, Ichoku H E, Ogujiuba K K and Chukwu J O 2008 J. Afr. Econ. 17 277-304
[33] Beharry-Borg N and Scarpa R 2010 Ecol. Econ. 69 1124-1139
[34] Train K E 1998 Land Econ. 74 230-239
[35] Goett A A, Hudson K and Train K E 2000 Energy J. 21 1-28
[36] Revelt D and Train K 1998 Rev. Econ. Stat. 80 647-657
[37] Basili M, Matteo M D and Ferrini S 2006 Waste Manage. 26 209-219
[38] Sillano M and Ortúzar J D D 2005 Environ. Plan. A 37 525-550
Estimating Willingness to Pay for Improving Municipal Solid Waste Management Using a Choice Experiment: A Case Study of Central
China
391

[39] Afroz R, Hanaki K and Hasegawa K 2009 J. Environ. Manage. 90 492-503
[40] Jin J, Wang Z and Ran S 2006 Ecol. Econ. 57 430-441
IWEMSE 2018 - International Workshop on Environmental Management, Science and Engineering
392