Valuation of recreational service by the mangrove ecosystem in thanh phu natural reserve

UNIVERSITY OF ECONOMICS INSTITUTE OF SOCIAL STUDIES HO CHI MINH CITY THE HAGUE VIETNAM THE NETHERLANDS VIETNAM - NETHERLANDS PROGRAMME FOR M.A IN DEVELOPMENT ECONOMICS VALUATION OF

UNIVERSITY OF ECONOMICS INSTITUTE OF SOCIAL STUDIES

HO CHI MINH CITY THE HAGUE

VIETNAM THE NETHERLANDS

VIETNAM - NETHERLANDS PROGRAMME FOR M.A IN DEVELOPMENT ECONOMICS

VALUATION OF RECREATIONAL SERVICE BY THE MANGROVE ECOSYSTEM IN THANH PHU

NATURAL RESERVE

BY

NGUYEN THI DIEU THUY

MASTER OF ARTS IN DEVELOPMENT ECONOMICS

UNIVERSITY OF ECONOMICS INSTITUTE OF SOCIAL STUDIES

HO CHI MINH CITY THE HAGUE

VIETNAM THE NETHERLANDS

VIETNAM - NETHERLANDS PROGRAMME FOR M.A IN DEVELOPMENT ECONOMICS

VALUATION OF RECREATIONAL SERVICE BY THE MANGROVE ECOSYSTEM IN THANH PHU

NATURAL RESERVE

A thesis submitted in partial fulfilment of the requirements for the degree of

MASTER OF ARTS IN DEVELOPMENT ECONOMICS

ABSTRACT

This study is part of a research project conducted to estimate environmental service value of the mangroves in Thanh Phu ward, Ben Tre province, South Vietnam to inform policy makers in applying the Payment for Forest Environmental Services program The study set out to measure recreational value of the forest through ecotourism activities using Contingent Valuation Method with double bounded dichotomous choice model The result shows that willingness to pay is affected by income and whether respondent thinks that environmental issues in Vietnam are properly addressed Mean WTP calculated using parametric and nonparametric method

is US $15.62 and US $12.61 respectively, which yields aggregate WTP at US

$2,656,204 with a lower bound of US $2,146,273

Keywords: mangroves, ecosystem service valuation, Contingent Valuation Method,

double-bounded dichotomous choice model

Abbreviations: CVM – Contingent Valuation Method, WTP – Willingness to pay,

PFES – Payment for Forest Environmental Services

ACKNOWLEDGEMENT

My thanks goes to Dr Pham Khanh Nam for giving me the chance to participate in the study which this thesis is part of, for his constant supervising and for his invaluable advice in helping me finish this thesis

I am also grateful to Dr Truong Dang Thuy for initiating my interest in the method used in this thesis and also for his advice

I thank my mother for her support to make it possible for me to write this thesis

I thank everybody at VNP for creating an academic environment where I learn skills and knowledge to write my thesis Specially, I would like to thank my classmate Nguyen Duy Chinh for giving me a lot of help and support when I work on the thesis

TABLE OF CONTENTS

ABSTRACT i

ACKNOWLEDGEMENT ii

TABLE OF CONTENTS iii

LIST OF FIGURES vi

LIST OF TABLES vi

CHAPTER 1 INTRODUCTION 1

1 PROBLEM STATEMENT 1

2 RESEARCH OBJECTIVES 2

3 RESEARCH QUESTIONS 2

4 SCOPE OF STUDY 3

5 THESIS STRUCTURE 3

6 CHAPTER SUMMARY 4

CHAPTER 2 LITERATURE REVIEW 5

1 ECOSYSTEM SERVICES VALUES 5

2 MEASUREMENT OF WELFARE CHANGES 6

3 REVIEW OF EMPIRICAL STUDIES 10

a Overall valuations of mangrove ecosystem services 11

b Valuations of recreational service for estuarine and marine ecosystems 12

c Valuations of recreational services as environmental goods using CVM 13

d Estimates of recreational services of mangroves in Vietnam and in other countries 14

4 CHAPTER SUMMARY 15

CHAPTER 3 METHODOLOGY 16

1 CONTINGENT VALUATION METHOD 16

a Steps in implementing CVM 16

b Elicitation format for the WTP amount 17

2 SCENARIO AND PAYMENT VEHICLE 19

3 MODEL OF WILLINGNESS TO PAY 21

4 ESTIMATION OF WILLINGNESS TO PAY 22

a Parametric estimation 22

b Nonparametric estimation 24

5 DATA 27

6 CHAPTER SUMMARY 28

CHAPTER 4 ANALYSIS RESULTS 29

1 OVERVIEW OF THANH PHU AND THE NATURAL RESERVE 29

a Thanh Phu district and tourism development 29

b Thanh Phu Natural Reserve 31

2 PRELIMINARY STATISTICS 35

a Descriptive statistics 35

b Correlation of coefficients 40

c Willingness to pay by bid levels 43

3 NON-PARAMETRIC ESTIMATION 45

4 DETERMINANTS OF WILLINGNESS TO PAY 45

5 PARAMETRIC ESTIMATION 47

6 AGGREGATE WILLINGNESS TO PAY 47

CHAPTER 5 CONCLUSIONS AND SUGGESTIONS 49

1 CONCLUSIONS 49

2 POLICY IMPLICATIONS 50

3 METHODOLOGICAL REMARKS 51

4 SUGGESTIONS FOR FURTHER RESEARCH 52

5 CHAPTER SUMMARY 53

BIBLIOGRAPHY 54

APPENDICES 62

LIST OF FIGURES

Figure 1 Schematic representation of the ecosystem services 6

Figure 2 Extract of bid questions from questionnaire 20

Figure 3 Administrative map of Ben Tre 29

Figure 4 Map of Thanh Phu Natural Reserve’s current situation 32

Figure 5 Shares of occupational groups in the sample 37

Figure 6 Respondents’ knowledge about the situation of mangrove area in Vietnam 39 Figure 7 Graph of survivor function after the tenth iteration 45

Figure 8 Bootstrapping result for value of non-parametric WTP 73

LIST OF TABLES Table 1 Bid groups 20

Table 2 Classification of flora species in Thanh Phu Natural Reserve: 33

Table 3 Some rare and threatened species in Thanh Phu Natural Reserve 34

Table 4 Description and summary of respondents’ socio-economic characteristics 36

Table 5 Ranking of importance of social issues in Vietnam 38

Table 6 Ranking of importance of ecotourism benefits to respondents 39

Table 7 Correlation of coefficients 41

Table 8 Percent of ‘yes’ responses for different bid levels 44

Table 9 Regression result of the bivariate probit function 46

Table 10 Estimate of mean WTP and confidence interval 47

Table 11 The first iteration of calculating TSCA 71

Table 12 The second iteration of calculating TSCA 72

CHAPTER 1 INTRODUCTION

1 PROBLEM STATEMENT

Vietnam is one among the countries most vulnerable to climate change and the mangroves are one effective protection against sea level rise Mangroves also provide many other services including support for fishery and aquaculture, provision of food and raw material, carbon sequestration, water supply and regulation and recreational activities However, the mangroves are reported to have declined rapidly due to deforestation, from 408,500 hectares in 1943 to 155,290 hectares in 2000 of which only about 21% is of high-quality natural forest (Sam, D.D., N.N Binh, N.N Que, and V T Phuong, 2005)

Funding is necessary to protect the mangroves, yet it is often lacking Drawing on the concept of payments for environmental services, a solution to generate and distribute cash for conserving ecosystems, the government has developed the national program of Payment for Forest Environmental Services (PFES) (Hawkin, 2010) Specifically, the Prime Minister has issued decree 99/2010 to guide the application of PFES to five groups of services provided by forests: soil related, water supply related, carbon sequestration, recreation based on natural landscape and finally, aquaculture-related

In order to decide the levels of payment and also to make informed decisions of economics planning, it is important that policy makers know of the value of the services provided by mangrove forest This study is part of an assignment that arises from that need, seeking to gauge the economic value of the mangrove ecosystem in Ben Tre province

socio-Every day hundreds of tourists go to the beaches in the coastal district of Thanh Phu,

seafood and learning to do the work on shrimp farms (Le Luong, 2015) These are supported by the presence of the mangroves besides its own green lush, which can interest tourists to pay to spend time in, learn from and support That way, Thanh Phu could follow Can Gio and capitalize on the current crowds of tourists to develop eco-tourism centering on the mangroves

Also, the mangrove ecosystem in Thanh Phu is officially recognized as a natural reserve and put under care of the authority The combination of mangrove forest presence, stewardship and available crowd of tourists makes Thanh Phu Natural Reserve an ideal place for carrying out evaluation of the recreational service using Contingent Valuation Method

2 RESEARCH OBJECTIVES

Firstly, the study will estimate the monetary value that tourists coming to Thanh Phu are willing to pay for the tourism service of the mangroves in Thanh Phu Natural Reserve, Ben Tre

Secondly, the author will attempt to identify factors affecting this willingness to pay

3 RESEARCH QUESTIONS

The study seeks to answer the following questions:

- How much the tourists coming to Thanh Phu are willing to pay for the tourism value of the mangroves in Thanh Phu Natural Reserve?

- What are the determinants of this willingness to pay?

4 SCOPE OF STUDY

The study is conducted for the mangroves in and tourist population coming to Thanh Phu District, Ben Tre Province, South Vietnam The sample population comprises of individuals among beach goers in Thanh Hai commune

The services provided by the mangroves to human are many and can be categorized under the main four groups of provisioning, regulating, supporting and cultural The study only set out to measure the category of tourism service, which belong to the cultural group

Willingness to pay is calculated for 2014 and used for the next several years as report shows that the number of tourists coming to Thanh Phu keeps increasing with improvements in infrastructure and unfolding services (Quoc Vinh, 2015)

5 THESIS STRUCTURE

The thesis remains in four chapters Chapter 2 reviews literature about how tourism fits

in the framework of ecosystem services values, how economists measure welfare brought about by environmental goods and previous related studies Chapter 3 discusses the methodology employed in the study: the steps of contingent valuation and elicitation format, hypothetical scenario and payment methods, construction of willingness to pay function, nonparametric and parametric methods for estimation of willingness to pay and finally, elements of data collection Chapter 4 discusses the findings of the study from knowledge about the situation of mangroves as tourist attraction to the statistics of data to results of nonparametric and parametric estimation Chapter 5 presents the conclusions of the study, comments about the methodology used

in the study, suggestions to policy makers and possible directions for future study

6 CHAPTER SUMMARY

The chapter first discuss the benefits of and threats to mangrove ecosystems and counteraction by the government, which eventually leads to the study Reason for choosing the site were also discussed After that, the research objectives and research questions are outlined Next the scope of the study is specified Finally, part 5 lays out the structure of the thesis

CHAPTER 2 LITERATURE REVIEW

1 ECOSYSTEM SERVICES VALUES

Ecosystems are “dynamic complexes of communities of animals, plants and microorganisms and the inorganic environment interacting as a functional unit, with human being an integral part” (UNEP, n.d.)

Among the Millennium Ecosystem Assessment categories, the mangroves belong to the Coastal Ecosystems group, being within an “area between 50 meters below mean sea level and 50 meters above the high tide level or extending landward to a distance

100 kilometers from shore” (UNEP, n.d.)

According to the UK NEA (n.d.), ecosystem services are “the benefits provided by ecosystems that contribute to making human life both possible and worth living” Examples includes benefits such as provision of material and food, regulation of water, shoreline protection and carbon sequestration, and non-material ones such as cultural heritage and aesthetic experience The term ‘services’ is often meant to include both the tangible and intangible benefits that humans obtain from ecosystems

These service are categorized into four groups: provisioning services, regulating services, cultural services and supporting services Tourism and recreation, which in this case are the same, are cultural services (UK NEA, n.d.)

This study is interested in measuring value of one form of tourism, ecotourism This is the service that the local government intends to extract from the mangroves and can be

defined as "responsible travel to natural areas that conserves the environment, sustains

the well-being of the local people and involves interpretation and education" (TIES, 2015)

Figure 1 Schematic representation of the ecosystem services

Source: http://www.unep.org/

2 MEASUREMENT OF WELFARE CHANGES

This study uses contingent valuation method to estimate the economic value of an ecosystem service According to Bockstael and Freeman (2005), CVM has theoretical foundation in the neo-classical welfare economics It presumes that economic activity

is to promote the welfare of the individuals, and each individual is the best judge of their own well-being Therefore, economic values of goods and services, whether market or nonmarket, should be based on their effect on human welfare

The theory has been developed to measure change in individual’s wellbeing to estimate the welfare effect of change of goods and services, be it a change of price or quantity

In doing so, the assumptions are made that individuals have well-defined preferences among alternative bundles of goods and their preferences have the attribute of substitutability, that is, people can make trade-offs between bundles of goods and services to keep their utility unchanged

With money as the common unit of exchange, that substitutability means the compensating increase and decrease in goods and services while making trade-offs can

be stated as changes to exogenous income

In order to illustrate the evaluation of change in levels of environmental goods, consider a simple case of abstraction Let D be a vector of human-induced impacts on the environment, Q be the vector of environmental outcomes that affect humans Q may directly enter individual’s utility function or indirectly through function Z(Q) of household or firm production process The artificial distinction between D and Q serves to distinguish between the relationships described physically or biologically and those that are fundamentally economic in nature

Let W be the level of economic welfare, the model of the case can be expressed as:

Going into details of (3), to define a measure of the welfare change bought about by an environmental change, we first consider the standard case where individual’s utility is a

function of only private purchased goods with no uncertainty The individual is subject

to income constraint and seek to maximize utility:

The measures of welfare change being considered here are based on compensation principle Utility change per se is known to be neither measurable nor additive for each individual and so cannot have unique money measures However, welfare measures can be defined by the monetary amount needed to compensate for the change in question, holding utility constant at the level of the status quo

To measure the welfare effects, economists use the concepts of compensating and equivalent variation Compensating variation (CV) is the amount of money needed to

be taken away to income to leave individuals with the same base utility after a change

in price, quality or quantity of goods Equivalent variation (EV) is the money needed to

add to income to give individual the same utility as after the change without making the change

𝑉(𝑃1, 𝑄1, 𝑀1− 𝐶𝑉) = 𝑉(𝑃0, 𝑄0, 𝑀0) (7)

𝑉(𝑃1, 𝑄1, 𝑀1) = 𝑉(𝑃0, 𝑄0, 𝑀0+ 𝐸𝑉) (8)

Defined as in (7) and (8), CV and EV have the same sign with the welfare effect

In developing mechanism for measuring CV and EV in case of change in P or Q, we seek to express these measures in terms of the expenditure minimization function, which is the dual of (4) In the case as in this study, Q enter the utility function directly: 𝑚𝑖𝑛 𝑃′𝑋 𝑠 𝑡 𝑈(𝑋, 𝑄) = 𝑈0 (9)

Substituting the cost minimizing demands for X into the objective function yields the expenditure function:

In the case of an improvement in welfare as in this study, EV is the WTP for the hypothetical increase in the environmental service From (15) and (16), we can derive WTP as a function of P, Q, and M

Random utility model

The model for utility used in this study is the random utility model (RUM) suggested

by Hanemann (1984) where uncertainty and individual’s characteristics are taken into consideration Applying RUM in the case of improved mangrove ecosystem service, the utility of stakeholder i for an improved ecosystem service is:

𝑉𝑖𝑗 = 𝑉𝑗(𝑀𝑖, 𝑧𝑖, 𝜀𝑖𝑗)

Where:

j=1 denotes the improved state and j=0 is the status quo

𝑉𝑗(𝑀𝑖, 𝑧𝑖, 𝜀𝑖𝑗) is a function of 𝑀𝑖, 𝑧𝑖, 𝜀𝑖𝑗 with 𝑀𝑖 being stakeholder’s discretionary income, 𝑧𝑖 vector of stakeholder characteristics and attributes of the resource and 𝜀𝑖𝑗being unobserved preferences (the uncertainty to researchers) Analogously to the defining of WTP in the previous math abstraction, the real WTP is also assumed to be a function of 𝑀𝑖, 𝑧𝑖, 𝜀𝑖𝑗

The CV bidding question will force the respondent to choose between improved ecosystem service at bid amount t and the status quo The data will be analyzed to derive parameters for the WTP function

3 REVIEW OF EMPIRICAL STUDIES

Strictly similar studies using CVM to study recreational value of mangroves were hardly available Related studies is found in four strands: overall valuations of

mangrove ecosystem services, valuations of recreational value for estuarine and marine ecosystem, applications of CVM in valuation of recreational services and finally, estimates of recreational services of mangroves in Vietnam and in other countries

a Overall valuations of mangrove ecosystem services

Overall valuations of mangrove ecosystem services use a diverse collection of methods These include, but not limit to, obtaining information of economic benefits rather than economic value (provision and cultural services) (Uddin, de Ruyter van Steveninck, Stuip and Shah, 2013); marketed substitutes/alternative supplies (firewood), valuing the marginal productivity of the resource net any human effort (fisheries), preventive expenditures or damage costs avoided (shoreline stabilization, erosion control, and control of storm surges), CVM (existence, bequest, and option values) (Gunawardena and Rowan, 2005); combination of earth-observation- and household-survey-based analyses (Kuenzer and Tuan, 2013; Quoc Vo, Kuenzer and Oppelt, 2015) and value transfer (Malik, Fensholt and Mertz, 2015)

Services valued also vary among studies Timber provision are estimated by all studies reviewed (Gunawardena and Rowan, 2005; Udin et al., 2013; Kuenzer and Tuan, 2013; Quoc Vo et al., 2015, Malik et al., 2015), while more complicated services are included and estimated differently by different studies Gunawardena and Rowan (2005) and Kuenzer and Tuan (2013) value fishery both inside the forest and nearby coast with fishes living part of their lives in the forest On the other hand, Uddin et al (2013), Quoc Vo et al (2015) and Malik et al (2015) value fishery inside mangroves only The estimations often include coastal protection and recently carbon sequestration while leaving out recreation and biodiversity benefits (Gunawardena and Rowan, 2005; Kuenzer and Tuan, 2013; Quoc Vo et al., 2015, Malik et al., 2015) Indeed, Gunawardena and Rowan (2005) acknowledge the need for insights into the interaction

between ecological and socio-economic links in order to obtain a fairly accurate value

of the mangroves

Besides studies valuing individual mangrove sites, there are meta-analysis studies conducted to find way to apply value transfer on mangrove ecosystems (Brander et al., 2012; Salem and Mercer, 2012) One of these is the paper by Brander et al (2012) for mangrove ecosystems values in Southeast Asia The authors study 130 value estimates, mostly for mangrove in Southeast Asia The services valued are material extraction (fuel wood, food or thatch), water quality maintenance or improvement, support to commercial fisheries and shoreline protection Other services such as recreational opportunities, biodiversity benefits and carbon sequestration were not included due to scarcity of studies estimating them The mean and median values found are 4185 and

239 US$/ha/year These values vary greatly due to factors including the site physical characteristics and the socio-economic characteristics of the stakeholders Finally, one factor positively influencing the value of a mangrove forest is adjacency to other mangrove forests

bio-b Valuations of recreational service for estuarine and marine ecosystems

Brander, Van Beukering and Cesar (2007) use meta-analysis to study the recreational value of coral reefs 166 studies were collected and 52 were statistically adequate for meta-analysis yielding 100 value observations The results show that larger dive sites and smaller number of visitors make sites more attractive to recreationists Different methods gives very different estimated values with contingent valuation produce significantly lower results Also, quality of a large part studies are low which may account for the variation in value estimations When applying value-transfer, the average transfer error is 186% which is not sufficiently reliable for the method to be used

c Valuations of recreational services as environmental goods using CVM

Numerous published studies can be found following this strand These papers can employ CVM alone (Shultz, Pinazzo and Cifuentes, 1998; Huhtala, 2004; Jim and Chen, 2006; Lee, Lee, Mjelde, Scott and Kim, 2009; Dong, Zhang, Zhi, Zhong and Li, 2011; Yamazaki, Rust, Jennings, Lyle and Frijlink, 2013), CVM in combination with expenditure data (Bergstrom, Stoll, Titre and Wright, 1990) or CVM separately and in comparison with travel cost method (Navrud and Mungatana, 1994) Researches are carried out in both developing countries (Navrud and Mungatana, 1994; Shultz et al., 1998; Jim and Chen, 2006, Dong et al, 2011) and developed countries (Bergstrom et al., 1990; Huhtala, 2004; Lee et al, 2004; Yamazaki et al., 2013) with factors significantly affecting willingness to pay reflecting the idiosyncrasies of the cases For example, while estimating the demand for recreational service of wildlife viewing in Lake Nakuru National Park of Kenya, Navrud and Mungatana (1994) find that recreational demand of international visitors is price inelastic as expected of the category of specialized, high quality viewing which this case fit into However, recreational demand of residents is elastic as this is considered a luxury for Kenyans Another example is the valuation of recreational services by parks In Guangzhou, China where people are accustomed to paying for access to greenspaces and seeing outdoor recreation in greenspaces as important leisure pursuit, only income significantly affect willingness to pay (Jim and Chen, 2006) However, in Finland where access to parks is free and income is high implying ample supply of recreational activities, 60% of the respondents indicated cultural and ecological conservation as the motivation for willingness to pay while less than 20% state opportunity to use the parks

as their reason for payment and income is not statistically significant as an impact on willingness to pay (Huhtala, 2004)

Issues of the CVM methods are discussed and sometimes addressed Although CVM is designed to capture both use and non-use value, Navrud and Mungatana (1994) found estimation result of CVM lower than that of Travel Cost Dong et al (2011) also comment that their result using CVM is low compared to other studies One early study (Shultz et al (1998) acknowledge the limitations of not including potential visitors; lacking detailed scenarios and information framing; and cultural-strategic biases when conducting surveys in a developing country On the other hand, a recent study by Yamazaki et al (2013) has tested and controlled for response biases, namely anchoring bias and structural shift

d Estimates of recreational services of mangroves in Vietnam and in other countries

A few studies estimating recreational services of mangroves in developing countries has been published Specifically, Bennett and Reynolds (1993) do research for the mangroves in Sarawak, Malaysia; Ahmad (2009) for the mangroves in Larut Matang, Malaysia; Uddin et al.(2013) for the Sundarban Reserve Forest, Bangladesh; and Kuenzer and Tuan (2013) for the Can Gio Mangrove Biosphere Reserve, Vietnam Methods used include obtaining secondary information of income collected from tourists (Uddin et al, 2013; Bennett and Reynolds, 1993), travel cost method (Kuenzer and Tuan, 2013) and contingent valuation method (Ahmad, 2009)

For comparison, the economic value of recreational service per ha by the mangrove forests in Sarawak, Larut Matang, Sundarbans and Can Gio are US$423.92 (for the year 1989), US$23.8 (2007), US$0.07 (2001-2002 to 2009-2010) and US$2960-US$4992 (2013) respectively The first and fourth numbers, being high in totally different levels compared to the other two numbers, is the income of already quite developed tourism dependent on mangroves in Malaysia and Vietnam The second number is the WTP for hypothetical ecotourism services provided within another

mangrove forest also in Malaysia and the third, lowest estimation is for official revenue collected only by the Forest Department of Sundarban Reserve Forest where real market values do not exist Adding to that the conclusion that values using CVM often yield lower results as discussed in earlier section, it seems that the value estimated by contingent valuation method is a conservative number that only inform about the initiated interest of the population in the environmental service Due to resource constraints, other significant economic benefits are overlooked which will arise with proper investment and development from government

4 CHAPTER SUMMARY

The chapter begins by providing definitions of key concepts including ecosystems, mangroves, environmental service and ecotourism After that, the theoretical foundation of welfare economics is presented which trace the connection from changes

to environmental goods to WTP measurement in random utility model Going on, empirical literature is reviewed along four strands: overall valuations of mangrove ecosystem services, valuations of recreational value for estuarine and marine ecosystem, applications of CVM in valuation of recreational services and finally, estimates of recreational services of mangroves in Vietnam and in other countries

CHAPTER 3 METHODOLOGY

1 CONTINGENT VALUATION METHOD

CVM is a survey based evaluation method where respondents are asked to state their willingness to pay (WTP, or willingness to accept, WTA) for the conservation (or loss)

of some environmental services The name “contingent” indicates that the respondents are being presented with a hypothetical situation of the environmental goods being evaluated CVM belongs to the category of stated preference methods as it is based on people telling what they would pay and not what is observed from what they actually

do

CVM is a popular method of evaluation, having decades of application and development It is proposed by Ciriacy – Wantrup (1947) and first applied in an empirical study by Davis (1963) to estimate value of a wilderness area to hunters and tourists In 1998, there was a notorious accident to the environment in the US, the Exxon Valdez oil spill CVM was then use to quantify the damages by asking WTP to prevent a similar accident The number is huge and sparked controversy Later Arrow

et al (1993) was commissioned to investigate whether CVM is a valid method of evaluation They concluded that the method can produce reliable estimates with the condition that surveys are carefully designed and implemented The method then gain breakthrough in popularity and at this time there is already a significant body of works applying CVM

- Describe the site, the services being valued, the relevant population

- Specify time span of evaluation

- Establish value to be estimated and unit of measurement

Step 2: Make preliminary decisions for the survey

- Decide on the sample size, mode of survey, who will be surveyed, who will pay for the good, who will use the good and other related issues

Step 3: Design the questionnaire

- Conduct interviews and focus groups to explore the situation and how to design questionnaire, which generally comprises of: Introduction, socio-economic questions, Scenario, WTP/WTA elicitation format, payment vehicles

- Pretest the questionnaire and make necessary revisions

Step 4: Implement the actual survey

- Decide the sampling technique, how, when and where to run the interviews

- Train enumerators

- Run the interviews

Step 5: Analyze data and report results

- Enter data: identify responses that may not express respondent’s value for the site, deal with possible non-response bias

- Analyze WTP/WTA with appropriate statistical techniques

- Extrapolate the benefit or loss for the population

- Compute annual net benefits and total value of the services

b Elicitation format for the WTP amount

There are many ways to elicit WTP amount, directly or indirectly Major ones include: Open-ended, Single-bounded Dichotomous choice (Take-it-or-leave-it), Double bounded dichotomous choice, Modified dichotomous, Payment card method, Iterative bidding games… In their book Carson and Hanemann (2005) recapped the development of elicitation methods as follow:

The first survey use the "bidding game" format where respondents receive series of questions with increasing or decreasing amount of payment until the respondent switch their answer from "yes" to "no" or vice versa This method leads to concern about the influence of starting bids on the final estimated WTP

Later the open-ended question was used where respondents get to decide their maximum payment, but this approach produced a lot of "don't know" question

To fix that, Mitchell and Carson (1981) use cards with different payment amounts from which respondents can choose

To allow for the variation in WTP, Bishop and Heberlein (1980) seek to bound WTP rather than trying to get out the exact WTP amount The method involves asking yes/no vote to alternative amounts randomly assigned to respondents This is known as the single-bounded dichotomous choice question

The single-bounded dichotomous question, however, provides little information about the WTP of each respondent and requires large sample to obtain expected precision To overcome this limitation, Carson (1985), Hanemann (1985), Carson and Steinberg (1990) and Hanemann, Loomis and Kanninen (1991) have developed the double bounded approach which can generate both a lower and an upper bound on each respondent's WTP It is done by asking binary question conditional on the first WTP question If the first is answered with a "yes", the second will increase the amount and

vice versa This is the approach that the author plans to use for this research and the corresponding response probability models

2 SCENARIO AND PAYMENT VEHICLE

The base scenario is of tourists arriving at the beach, doing typical activities like swimming, playing sports, taking picture and having seafood then go home or to a nearby attraction The beach is not too magnificent but rather unspoiled This is very common experience except that the seafood is fresh and priced inexpensively

Some tourists are then approached and requested to help answer a questionnaire In the questionnaire, the tourists will be introduced with a scenario The scenario starts with a brief about the situation and value of the mangrove forest A hypothetical ecotourism trip into the mangrove forest is then presented The trip includes cruising into and learning from a tourguide the features of the mangrove forest The Mangrove Forest Management Board will plan the route, provide boat and trained tourguide for the trip The payment method is through a price for the ticket for the trip Respondents are informed that profit will go into supporting the conservation of the forest Each respondents will be asked two questions to elicit his or her willingness to pay as:

Figure 2 Extract of bid questions from questionnaire

To determine the bid levels, a group of enumerators were sent to interview tourists for possible levels of bid A later focus group between enumerators and experienced practitioners of environmental valuation generate 7 bid levels: 50,000; 100,000; 150,000; 200,000; 300,000; 400,000; 500,000 From these bid levels, 18 sets of bid questions were constructed as follow:

Table 1 Bid groups

Group No of

tourists

1st bid (VND)

Higher bid (VND)

Lower bid (VND)

7 Would you pay a fee of <bid level> to have that trip?

 Yes  go to question 8  No  go to question 9

8 If the fee is <higher bid level>, would you pay it to have that trip?

9 If the fee is <lower bid level>, would you pay it to have that trip?

Group No of

tourists

1st bid (VND)

Higher bid (VND)

Lower bid (VND)

3 MODEL OF WILLINGNESS TO PAY

The respondents are asked whether they are willing to pay a bid amount t1 and then bid amount t2 in the elicitation questions The assumption is that they will answer yes when their real willingness to pay is higher than that amount: WTP > t

Real WTP of respondent i is assumed to be a linear function of explanatory variables and the error term:

KNOWLEDGE: whether respondent is correct about the situation of mangroves in

Vietnam

GENDER: whether respondent is male or female

MARISTAT: whether respondent is married or not

EDUC: respondent’s highest education attainment

HHSIZE: respondent’s household size

INCOME: respondent’s income level

TRANSMEANS: whether respondent travel to site by motorbike or car

DISTANCE: distance respondent has travelled to visit the site

4 ESTIMATION OF WILLINGNESS TO PAY

a Parametric estimation

The WTP amount in (17) is not directly asked of the respondent, but rather extracted by presenting bid levels for him or her to choose By modeling the probability of the dichotomous answers and using maximum likelihood to fit the data, we can estimate β and compute mean WTP (Jeanty, 2007)

In this case, the choice variable ANSWER is regressed on variables BID, ENVISSUE, ECOEXP, KNOWLEDGE, AGE, GENDER, MARISTAT, EDUC, HHSIZE, INCOME, TRANSMEANs and DISTANCE (18)

Following equation (16), this study set the assumptions that the error term ui follows the normal distribution and proceed to estimate the bivariate probit function as two separate probit functions, assuming that the WTP functions of the first and the second answer are different

After that, observation is made on whether the two probit functions are correlated and the hypothesis that these probit functions are the same is tested If it cannot be rejected, constraints will be set to have bivariate probit regression results in two similar probit function If it is rejected, the probit function of the first answer will be used for analysis

as the second answer is likely to be subjected to response biases such as anchoring and structural shift (Whitehead, 2000)

Finally, the Robb and Krinsky procedure is carried out to find mean WTP and its confidence intervals (Haab and McConnell, 2000, p.110 – 113)

Continuing with (18), for the probit models, the probability of answering yes given the respondent with vector of explanatory variables 𝑧′𝑖 is

Pr(𝐴𝑁𝑆𝑊𝐸𝑅 = 𝑦𝑒𝑠|𝑧′𝑖)= Φ(𝛽′𝑧′𝑖)

Where:

𝛽′ is the vector of estimated coefficients of (18), which is also the estimate of vector β of (17) adding coefficient for bid

Φ is the cumulative normal distribution

Using maximum likelihood estimation for probit function, we can find out 𝛽′ Estimate

of mean and median WTP is given by:

Mean (median) WTP = −Z̅β

β0Where:

Z̅ is the row vector of sample mean, including 1 for the constant and excluding bid

β is the column vector of estimated coefficients 𝛽′ excluding the coefficient of bid, which is also the vector in of coefficients in (16)

β0 is the coefficient of ‘bid’ variable

b Nonparametric estimation

It is known that estimate of mean WTP is very sensitive to the distribution assumption

of WTP, particularly if the pattern of responses is not well-behaved (Haab and McConnell, 2002) Therefore, it is useful to develop a least restricted estimation of WTP With only the assumption that WTP of a respondent is equal or greater than the bid which they agree to pay, the method of nonparametric assumption matches the need

Also, nonparametric calculation is conservative, using only the bids to which the respondents agree to pay as their WTPs Consequently, mean WTP estimated by

nonparametric method can be taken as the lower bound estimate of mean WTP (Bateman et al., 2002)

In order to calculate the mean WTP from double bounded data, Bateman et al (2002) suggests using the method of Turnbull Self Consistency Algorithm (TSCA) The TSCA WTP is calculated as the common Kaplan-Meier-Turnbull after reassigning all respondents into basic WTP interval bounded by two adjacent bid levels It is an iterative process as described below

Firstly, the initial estimate of the survivor function 𝑆̂(𝐵𝑗) is established

𝑆̂(𝐵𝑗) = 𝑛𝑗

𝑁

Where:

𝐵𝑗 is the boundary levels of intervals where respondents’ WTPs fall into, where

𝐵0 = 0 and 𝐵𝑗+1= infinity Otherwise 𝐵𝑗corresponds to the bid levels and there are J bid levels in the survey

𝑛𝑗 is the number of respondents with WTP greater than 𝐵𝑗

N is the number of respondents in the sample

Mean WTP estimate is given by the area under the survivor graph (with x-axis being 𝐵𝑗and y-axis being 𝑆̂(𝐵𝑗))

𝑀𝑒𝑎𝑛𝑊𝑇𝑃̂ = ∑ 𝑆̂

𝐽

𝑗=1

(𝐵𝑗)[𝐵𝑗 − 𝐵𝑗−1)

Then, the conditional probability of respondents who have WTP bounded by 𝐵𝑖 and 𝐵𝑘

to have WTP lying in basic interval j (bounded by 𝐵𝑗−1 and 𝐵𝑗) is:

𝑆̂(𝐵𝑗−1)− 𝑆̂(𝐵𝑗)𝑆̂(𝐵𝑖)− 𝑆̂(𝐵𝑘) i ≤ j – 1, k ≥ jAfter that, the recalculated number of respondents with WTP lying in basic interval j is:

Where ℎ𝑖𝑘 is the number of respondents with WTP with overlapping interval bounded

In order to check the accuracy of the estimates, confidence intervals are constructed by calculating WTP for different sets of data with observations randomly drawn from the original dataset The task is carried out using Crystal Ball software for Excel

There are five enumerators with two males and three females Before going out to do fieldwork, they are trained to conduct the interview in an appropriate manner The training is done with: a) distributing questionnaire and list of rules for good interview, b) answering questions from enumerators about the questionnaires and the rules and c) holding interview example with comment from trainer and other enumerators

The rules of good interview practice are taken from Whittington (2002) and contain the following main points:

- Read the questions clearly and exactly as printed to have all respondents answer

to the same questions

- Interview each respondents separately and stay absolutely neutral about their answers to ensure honest opinions are given

- Listen carefully to the answer to make respondents feel being heard However,

do not give advice on personal matters or suggests an answer

- Answer directly any questions the respondents may ask about the purpose of the survey

6 CHAPTER SUMMARY

This chapter presents the methodology of the study in five parts Part 1 discusses the adoption of CVM by researchers and policy makers, the required steps in conducting a study using CVM and the decision of elicitation format for this study Part 2 details the scenario and payment method which are important parts of the questionnaire Part 3 is about how willingness to pay can be modelled, which helps to understand how it is estimated by parametric and non-parametric methods as explained in part 4 Finally, part 5 informs about data collection for the study

CHAPTER 4 ANALYSIS RESULTS

1 OVERVIEW OF THANH PHU AND THE NATURAL RESERVE

a Thanh Phu district and tourism development

Thanh Phu is one of three coastal districts of Ben Tre, which is a province in the Mekong Delta Thanh Phu has seventeen communes and one town of the same name

As of 2013, the district has an area of 425.7 km² and population of 130,820 (Tiềm năng

và cơ hội đầu tư của huyện Thạnh Phú, 2014) Its capital lies in Thanh Phu town

Figure 3 Administrative map of Ben Tre

To the north of the district is Ham Luong river and Ba Tri district, to the west is Mo Cay Nam district, to the south is Co Chien river and Tra Vinh province and to the east

is the East Sea

Thanh Phu has great potential for tourism development with numerous tourist attractions These include the Ho Chi Minh Trail by Sea, 100-year-old Huynh Phu mansion, many trade villages, pristine alluvial beaches in Thanh Hai and the mangrove ecosystem However, with poor infrastructure and underdeveloped hospitality sector, the potential is still to be realized

Over the last five years (2011 – 2015), the government has made many efforts to address these issues Many historical-cultural relics such as the Huynh Phu mansion, Ong Tomb (Lang Ong), An Linh Pagoda has been renovated The long time project to renovate the Ho Chi Minh Trail by Sea is half way toward finishing the first phase (2013-2017) Traffic infrastructure has been improved by tarring the road from Cau Van to Con Chim T-junction and building Cau Van Bridge in place of the old ferry-boat which connect upper communes to the three coastal communes through Cau Van (Quoc Vinh, 2015)

Also, many investment promotion activities and training sessions have been carried out

to establish a good hospitality sector The provincial Information Center for Tourism Promotion works with the media to disseminate information about natural resources, tourist attractions and projects to attract investments Some government bodies work together to organize experimental community-, eco-tours to persuade travel companies

to bring tourists to Thanh Phu Training on tourism for administrators, foods for small businesses, first aid, rescuing and wave intensity observation were given Tourism advisory team, first aid team, waste collection team; assemblies inspecting service

price and food safety were formed to ensure satisfaction of tourists coming to the district (Quoc Vinh, 2015)

b Thanh Phu Natural Reserve

Designated by the government on a national level, Thanh Phu Natural Reserve has coordinates of 9048' - 9059'N and 106034' - 106041'E and is situated on the coastal area between Ham Luong and Co Chien estuaries The total area is 2,584 ha which comprises of a strict protection section, ecology restoration section I, ecology restoration section II and an administration – service section

Being located between estuaries and the sea, the area is strongly affected by both accretion and erosion The major type of soil are saline tidal and sandy ones There is

an intensive web of streams which serve traffic in the region However, due to alluvial accretion raising the bottom of streams, waterway traffic depends on the tides which occur twice daily The quick pace of the tide make transportation difficult

The area has a tropical climate with monsoon and two distinct rainy and dry seasons The rainy season lasts from May to October and the dry season lasts from November till April the next year Average temperature is 26.60C, fluctuating between peak of 28.40C in April and trough of 24.30C in December Average humidity is 83%, being highest in August till October (84%-94%) and lowest in February till March (65%-80%) Average rainfall is 1454mm, with about 126 days rainy days yearly Rainfall is distributed unevenly and concentrates in rainy season, being highest in October

As of 2000, there are 1,470 people living in the natural reserve and more than 6,935 people living in the buffer zone Main economic activities are agriculture, fishing and forestry

Figure 4 Map of Thanh Phu Natural Reserve’s current situation

Source: Mapping and Database Center – Forest Inspection and Projection Institution