Willingness to pay for the conservation of yew trees in vietnam

The subject of this research is to measure the existence value of yew trees using contingent valuation method CVM with single-bound dichotomous choice DC.. In answering the above questio

UNIVERSITY OF ECONOMICS INSTITUTE OF SOCIAL STUDIES

VIETNAM - NETHERLANDS PROGRAMME FOR M.A IN DEVELOPMENT ECONOMICS

WILLINGNESS TO PAY FOR THE

CONSERVATION OF YEW TREES IN VIETNAM

BY NGUYEN THANH TUAN

MASTER OF ARTS IN DEVELOPMENT ECONOMICS

HO CHI MINH CITY, SEPTEMBER 2013

HO CHI MINH CITY THE HAGUE

VIETNAM THE NETHERLANDS

VIETNAM - NETHERLANDS PROGRAMME FOR M.A IN DEVELOPMENT ECONOMICS

WILLINGNESS TO PAY FOR THE

CONSERVATION OF YEW TREES IN VIETNAM

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

MASTER OF ARTS IN DEVELOPMENT ECONOMICS

By NGUYEN THANH TUAN

Academic Supervisor:

DR TRUONG DANG THUY

Yew tree has not only value in use of wood products but also in existence of itself which provides valuable opportunities to other fields of research such as biology, chemistry, pharmacology The number of Vietnamese yew trees, however, has reduced year after year and today there are only approximately 250 natural individuals locating in Daklak, Vietnam According to the IUCN and Viet Nam Red Data Book, Vietnamese yew trees are critically endangered The subject of this research is to measure the existence value of yew trees using contingent valuation method (CVM) with single-bound dichotomous choice (DC) The study finds out that the mean WTP for conservation of Yew in Vietnam is approximately VND 73,000 per household; and household heads with different socio-economic characteristics such as age, income, occupation, and education have significantly different levels of WTP

Acknowledgement

I would like to thank my supervisor for his instruction and comments through my research Especially I am deeply impressed by his enthusiasm with all students working with him That

is one of the motivations to help me accomplishing the thesis

I also want to say thanks to all of professors, lecturers, and other people in this program Without knowledge from their classes and tutorials, my thesis cannot be possible Eventually, I would like to express my gratitude to my family, my friends, and my company

who did support me during the time of studying and working on my thesis

Nguyen Thanh Tuan

Chapter 1: Introduction 1

Chapter 2: Literature Review 6

2.1 A brief history of CVM 8

2.2 Economic theory of CVM 10

2.3 Basic elements of CVM 12

2.4 Empirical applications 16

2.5 Empirical studies In Vietnam 17

Chapter 3: Methodology and Data 19

3.1 WTP estimation 19

3.2 Survey design 21

Chapter 4: Results and Discussions 26

Chapter 5: Conclusion 53

Table 1: Survey mode 13

Table 2: Statements 33

Table 3: Response of WTP questions – at the first time 36

Table 4: Response of WTP questions – at the second time 36

Table 5: Calculated mean WTP – non-parametric estimate 43

Table 6: Definitions of variables included in the regression 44

Table 7: Results of simple regression with constant and BID 45

Table 8: Regression results 47

Table 9: Summary of parametric estimated WTPs for the case of dependent variable - Y22 51

Table 10: summary of estimated WTPs for the case of dependent variable – Y22 52

List of figures Figure 1: Distribution of respondents’ age 26

Figure 2: Cross - tabulation of marital status and group of age 27

Figure 3: Occupation by gender 28

Figure 4: Levels of education 29

Figure 5: Gender in levels of education 30

Figure 6: Distribution of reported household income 30

Figure 7: Country facing problems 31

Figure 8: Environmental issues 32

Figure 9: Ranks of proposal species 33

Figure 10: Attitude toward endangered species 34

Figure 11: Survivor function for the second time WTP question 37

Figure 12: Reasons of why respondents dislike the proposed payment vehicle – electricity surcharge 40

Figure 13: Discussion time to answer questionnaire 41

Figure 14: Survivor function after certainty adjustment for the first time WTP elicitation 42

Figure 15: Survivor function after certainty adjustment for the second time WTP elicitation 43

Chapter 1: Introduction

Yew (also known as water-conifer) is - the scientific names Glyptostrobus pensilis - listed in the IUCN and Vietnam red data Books as endangered situation (CR)1, and considered as the living fossil of gymnosperm industry The Decree 32/2006/ND-

CP of the Vietnamese government has stated that this kind of tree is classified into group IA: strictly banned from exploitation and use for commercial purposes

According to the reports of Forest Protection Department (FPD) of DakLak province, this species of yew trees in the world now exists only in Ea Ral (Ea H'Leo district), Trap K'sor (Krong Nang district) and a few dozen has been scattered in Krong Buk district, DakLak province (Bao Huy, 2010) This ancient species facing extinction needs urgent solutions for propagation, conservation and development

Yew wood is un-attackable to termite, warping, and fragrant Therefore, it is normally used to produce furniture, household goods, musical instruments, etc The breathing roots which are smooth, porous, and light have been used to produce cork, hat Moreover, its branches, leaves, and ripe cones are used as arthritic medications, pain reducing, and skinny tightening It also has fine grain, many patterns, and especially the longer soak in the water or mud, the shinier it is Additionally, yew shape is beautiful, then, one could grow yews for ornamentation, or for protection the coastal land from erosion

Population and habitats of yew tree in Daklak, Vietnam:

The number of yew currently is 255 individuals2 which only locate in Daklak province (Bao Huy, 2010) There are 219 individuals of yew in Ea Ral, Ea H’leo

1

CR – according to classification of IUCN and Vietnam red data book, CR means critically endangered The order of classification is that Extinct (EX), Extinct in the wild (EW), Critically endangered (CR), Endangered (EN), Vulnerable (VU), Lower risk (LR), Data deficient (DD), and Not evaluated (NE) Find more details in Vietnam Red Data Book, Part II Plants (Dang et al., 2007)

2

According to the managers of the conserving yew tree project in Daklak for period 2011 – 2015, this number was updated as 162 individuals in Aug, 2012

district; 31 individuals in Trap K’sor, Krong Nang district; and 5 old trees in Cu Ne, Krong Buk district

The existence of yew has not only meaning to itself, but also to the whole area of primary forest with biodiversity in which they locate According to Bao Huy (2010), with the biological characteristic that they have just lived in the seasonal wetlands, yews only exist in some specific areas in Vietnam or in the world With the thousand-year existence, yews and biological system in the area of conservation have been a very valuable place for researches If there is no action or intervention from the government from now, the risk of permanent loss of the thousand-year primitive forest is inevitable by the growing and strengthening effects of timber extraction and yew-wood trading activities That would be a large detriment for studies have not been done as well as the discovery of biodiversity here

Figure 1C (in Appendix C) illustrates the current distribution of yew individuals in Trap Ksor (Krong Nang) Each individual has been numbered as shown in the figure The distribution is dispersed widely in a large area In Trap Ksor (Krong Nang, Daklak), on March 24, 1987 the local government of Daklak province established a Decision about scheming and forbidding Trap Ksor forest area, and identified that this zone had become a protection area of yew trees’ gene The total area is about 98.6 hectares and under the control of 2 rangers and 3 contractual officers The surrounding area is agricultural land, separated from the core zone by simple barbed wire The current problems with this preservation area are that there are not sufficient human resources for protection purpose, low budget (about 16 million VND per year), and long distance from here to the managerial office (10 kilometers) This explains why there are still a lot of timber extraction activities here Moreover, the agricultural activities of farmers in surrounding area have reduced the water resource that supports for growing and fire protecting of yew (Bao Huy, 2010)

Figure 2C (in Appendix C) describes the current distribution of yew population in

Ea Ral (Ea H’leo) In comparison with population in Trap Ksor, the individuals’ location is quite close This would be a good condition in planning this area for conservation program In Ea Ral (Ea Hleo), the Yew management station has been established and under control of Ea Hleo forest protection unit (FPU) in 1994 But according to the manager, there has been not an official decision for this establishment The number of rangers here is just five persons Similarly to Trap Ksor, this lack of management gives more opportunities to illegal loggers The statistic number of FPU shows that during 2009 there have been 22 violating cases caught and prosecuted (Bao Huy, 2010)

The number of yew trees in Cu Ne (Krong Buk) is just five individuals In addition, these trees are very old and without top of trees All of them locate far away from the residential area, then, the officers face many considerably difficulties in protecting activities (Bao Huy, 2010) According to scientists (Bao Huy, 2010; Tran, 2012 in Thai, 2012), the number of current yew trees is going to be reduced in the near future due to the increasing illegal logging and trading

Current efforts of conserving yew tree in Daklak province: In 2011, the

government of Daklak province has approved the project of conserving habitat and yew trees in DakLak for period 2011 - 2015 The project was designed with funding from the state budget and partly from the contributions of the people and the international donor organizations However, this project had not been implemented until August 2012, and at that time, only Ea Ral area was taken over by the project managers The remaining areas in Krong Buk and Krong Nang have just been handed over in March 2013 This slow procedure is due to lack of personnel and funding3 According to the project design, the Ea Ral area needs at least 7 people on patrolling duty However, three of them are general staff, and handle concurrently

3

Mr Phuoc – current director of Yew conservation program – answered reporter of Da Nang Polices Newspaper on May 20, 2013 Online version is available at http://cadn.com.vn/new/65_10471_-tha-p-tu- ng-ba-u-va-t-thu-y-tu-ng.aspx

jobs as official and patrolling staff in reality Infrastructure and equipment are not fully equipped This has made limited in ability to protect yew populations in Ea Ral, especially patrolling activities in protected areas With flooded and marshy properties, the movement of patrol is very difficult In heavy rain conditions, this problem is even more difficult Another concerning problem is the area around the belt has not been cleared; there are many individuals of yew being on the boundary and adjacent areas of agricultural crops of the local residents Therefore, it is very easy for illegal loggers to intrude the reserves At the same time, farming operations, irrigation of the surrounding population directly affect water resources and

ecological conditions of the yew populations

As mentioned above the number of yew individuals in Ea Ral area was counted as

219 at the time of conducting the project research by a group of researchers from the Tay Nguyen University in 2010 However, until the project management was officially formed and accepted the transfer of management, this number is only 140 yew trees The main reason is due to destruction of illegal loggers If the loss of control is continued, there will be more individuals of yew ruined annually The reduced number of yew also means that the resource for research and breeding will decrease Reducing the number of individuals and increasing of competing species which have been dominated habitats, plus the impacts of unfavorable natural elements from the surrounding agricultural activities will directly affect the habitat

of yew environment This increases the risk of loss of yew in the future

Research problems

In summary, the lack of resources for research and management has constrained the conservation of yew trees Moreover, the low ability and quantity of rangers make this task limited One important thing should be considered is that during 35 years the scientists who were tracking these areas did not recognize any regeneration of young yew tree by seed, just only a few plants regenerated by buds Most individual

Annually plants still flower and fruit set, but the results are flat-seeds (Bao Huy, 2010)

Research questions

The above problems raise a question of whether resources should be devoted to conserve yew trees A full cost-benefit analysis is required to answer this question This study contributes by measuring the economic benefits of the conservation of yew trees This research analyzes attitudes and preferences on the issue of Vietnamese yew trees through a public survey This paper’s aims are to (i) assess the public awareness of yew trees; (ii) to answer the question “Are people willing to pay for the conservation of Vietnamese Yew?” If yes, how much are they willing to pay?; and (iii) to determine the impacts of socio-economic characteristics on the respondents’ WTP

In answering the above questions, this study is able to partially measure the economic value of conserving yew tree in Daklak, which will be helpful for policy makers to decide on the conservation of yew trees, especially in the funding for a project if it would be conducted

Chapter 2: Literature Review

Most of goods and services from natural resources and environment are considered public goods, such as the forest, breathable atmosphere, water, recreational sides, and wildlife Unlike marketed goods which have market prices, these public goods have neither market nor price Public goods and services are not sold or bought in markets; and markets do not regulate any activities of supplying or changing them

A system of decentralized market is not believable to conduce to the optimization of using natural resources (Freeman, 2003, chapter 1, p.6) Meanwhile, the measures of welfare from these providing or changing of environmental goods and services in term of money are very crucial for efficient use of natural resources Up to now, there are many methods that have been developed to monetize the welfare generated

by these goods and services

One of the primary elements to reach the efficient uses of natural resources is to assure that the valuation of environmental assets is appropriate Asheim (2000) in Amirnejad et al., (2005) states that the concept of environmental valuation brings on questions about the ability to account for the environmental value in terms of finance According to Howard and Farber (2002), efforts to assess the monetary value of ecosystem services act as various roles in management of the links between natural systems and human At the macro level, there is the contribution of ecosystem valuation into the construction of indicators of human welfare and sustainability At the micro level, the studies of valuation are based on both the structure and function, and the varied and complex roles of ecosystem in supporting human welfare to reveal information

The total economic value of a natural resource consists of use and non-use values Use value includes actual value (for instance, relaxing in a national recreational area) and option value (value that is not planned to use at the present but preserved the existence to use in the future as an option) Non-use value relates to the WTP to

Non-use value includes altruistic value, existence value, and bequest value Altruistic value is welfare generated from the fact that a good or service is available

to other people in the society Bequest value is the welfare from knowing that goods and services are available to the future generation The existence value reflects the value that an individual is willing to pay to maintain existence of goods, given that individual knowing there is no value of actual use or planned use for anyone (Bateman et al., 2002).This study aims to measure the existence value of yew tree in Vietnam by estimating the WTP of yew tree conservation

Several techniques have been developed to elicit the attitudes and preferences for, and thus monetize the total economic value of public goods such as environmental services, forest, and recreation These techniques are usually classified into revealed preference method and stated preference method However, the most popular technique to measure existence value is the stated preference (SP) Researchers could apply this method to ask people about their WTP for public goods, or their willingness to accept (WTA) a decrease in public goods, or to suffer the bad services, based on a hypothetical setting Among stated preference methods, contingent valuation method (CVM) and choice modeling (CM) are popular to analysts In CVM, respondents are asked to give their WTP or WTA for a change in supplying or quality improving of natural and environmental resources In CM, respondents will face a series of choice sets In each set, they have to choose the preferred alternative among several alternative goods, or rank the alternatives (Mogas, Riera & Bennett, 2006) Bateman et al (2002) suggests that if it is required the WTP for the total values of environmental good or service, then the contingent valuation method (CVM) should be chosen, instead of choice modeling (CM), because the latter is more suitable for measuring the WTP for one or some attributes

of that good

The following parts of this section will provide a brief history of CVM, economic theory of CVM, and basic elements of CVM

2.1 A brief history of CVM

Among the stated preference techniques, this paper applies the contingent valuation method CVM has been accepted by both academics and policy makers for the purpose of valuation of resources, environmental goods and services (Han, 2011) CVM has become the method to be used most widely for environmental valuation with stated preference information, with the value estimates obtained are contingent

on the information which has been early provided to interviewees in the survey

CVM was firstly introduced by Bowen (1943) and Ciriacy-Wantrup (1947), but Davis in 1963, was the first person who used CVM in an empirical study; in which

he conducted a survey within hunters’ sample and estimated the goose hunting’s benefits The two important components in the total economic values were gained after that in the environmental economics literature: option and existence values These values are not able to be estimated by the revealed preference methods such

as travel cost method (Smith, 1993); only the CVM or CM could capture the estimation of these values (Desvousges et al., 1993) In this line of thought, Hanemann (1994) states that this method is one of the methods that provide the standardization and flexibility to measure the economic values CV method is based

on questionnaire to estimate non-market goods’ economic values In the survey, people will be asked to give out monetary bids for hypothetical goods with the provided information of products However researchers in doing CVM need to consider its possible shortcomings Problems of CVM pointed out and criticized by Diamond and Hausman (Hanemann, 1994) are validity of surveys, surveys’ vulnerability to response effects, creating the values of survey process, respondents’ prior experience or training for valuing the environment, and unable verification of survey responses Hanemann (1994) argues that it is theoretically possible to establish experiments of CV scenario which avert problems in terms of economic models commonly occurred with observed data

There are some ways assuring the reliability of survey that are avoiding convenience sampling and self-administered surveys; confronting subjects with situation that is specific and realistic, instead of abstraction; applying closed-ended question; adding

an interrogating section at the bottom of questionnaire to check whether respondents understand and accept the important parts of the scenario; and applying appropriate statistical procedures in data analyzing

Survey responses could be significantly influenced by some small changes in the order or words used in the survey instrument (Schuman and Presser, 1981 cited in Hanemann, 1994) But in some cases, the effects such as order, shifting meaning, and framing effects can be controlled For example, one choses the sequence to create a cautious result or randomizes items’ order across interviews to deal with the sequence effects Or by using stringent tests with techniques of cognition, researchers can understand what the meaning of instrument to people is, and what the responses mean

CV has been asserted that its respondents place an unreal value on item, and the interview process generates the values that it is looking for to measure Hanemann (1994) stated that it is theoretically possible by debriefing section to identify that whether one had inattention or no focus and stated “hasty or ill-considered responses”, and researchers can discard them if desire

By eliciting the suggestion of NOAA Panel, Hanemann (1994) confirmed that prior experience or training is not relevant if the purpose of CV survey is to draw out preferences of people

To defend the idea that survey responses cannot be verified, Hanemann (1994) listed a lot of studies and concluded that “replication, comparison with estimates from other sources, and comparison with actual behavior where this is possible” are the ways to make validation of CV results

2.2 Economic theory of CVM

In term of welfare economics, there should be a public intervention when the total benefits of this intervention are greater than its costs However, the problem here is that how can we measure this overall benefits? The starting point is the estimation

of individual benefit which then will be used to aggregate to the relevant population

to come up with the social benefits The net change in income associated with the change of public goods provided in terms of quality or quantity is the accurate measure in the estimating procedure of an individual’s benefits The information observed from the CV survey will be employed to discover the willingness to pay (WTP) distribution for a change in hypothesized provision of environmental goods

A combination of the utility function in the economic theory and the error term in econometric theory is the framework of CV method In a discrete-choice CV survey, respondents are asked to vote for or reject a program that provides an environmental good or services, associated with a given monetary cost (T) Respondents are assumed to have a utility function, U, which is a function of income (A), and a set of conditioning factors (S): U (A; S) This means respondents know which factors are important to their utility levels But the true utility function may be unknown to the researchers; therefore, they have to build up a simplified model of real one Of course, it also captures important factors that are in establishing their welfare changes The analyst’s model is given by u(a; s; ɛ) The lower cases a, s are used to reflect the fact that including factors by analyst will not be exactly the same as considered respondent’s ones To take into account the differences, the researchers’ model contains a random element ɛ, which analyst cannot observe in the real utility function of respondents We can write the utility function as:

Uij(qi; Aj; Sj) = uij(qi; aj; sj) + ɛij (1)

Where i=1 (q1) is the condition that the environmental good or service, or an improvement in its quality, is supplied; and i=0 (q0) is the status quo Each

environmental good (yew trees in this study) with a payment of specified monetary amount, T People will accept the suggested amount if

uj(q1, aj-T ௝ ; sj) + ɛ1j ≥ uj(q0, aj; sj) + ɛ0j (2) and reject otherwise (Hanemann, 1984; Lee and Han, 2002)

However, the unobservable component ɛj is captured in this function Which means that analyst cannot observe it and is unable to predict the value taken by ɛ, but can only estimate the probability of it taking any specific value Therefore, the probability of a “yes” response could be estimated by the followings:

Pr(yesj) = Pr(uj(q1, aj-T ௝; sj ) + ɛ1j ≥ uj(q0, aj; sj) + ɛ0j) (3)

Now, if we assume the deterministic part is linear in income and covariates, then it results the linear utility function Hence the deterministic part can be re-written as:

௜௝( ௝)= ∑௠௞ୀଵ  ௜௞  ௝௞+  ௜( ௝) (4) + when i = 1 (q1 condition)

ଵ௝( ௝− ௝)= ∑௠௞ୀଵ  ଵ௞  ௝௞+  ଵ( ௝− ௝) (5) + when i = 0 (q0 condition)

଴௝ (  ௝ )= ∑௠௞ୀଵ  ଴௞  ௝௞ +  ଴ (  ௝ ) (6) Then the change in deterministic utility:

ଵ௝−
଴௝ =∑ (௠௞ୀଵ ଵ௞− ଴௞) ௝௞+  ଵ( ௝− ௝) −  ଴( ௝) (7)

Since the marginal utility of income is constant:

ଵ௝ −
଴௝ =∑௠௞ୀଵ  ௞  ௝௞ -  ௝ (8) The probability for utility function is then:

Pr(yesj) = Pr (ɛ ௝ < ∑௠௞ୀଵ  ௞  ௝௞ −  ௝ ) (9) where ɛ௝ ≡ ɛଵ− ɛ଴ The model can be estimated by assuming a distribution for ɛ௝ When a logistic distribution is assumed, the model can be estimated by a logit model

2.3 Basic elements of CVM

As described earlier, CV method is based on the information observed from a survey It measures value of a scenario, compared to status quo Therefore, it is basic but very important to the survey design including sampling, survey mode, scenario, elicitation format, and questionnaire structure Beside this, survey administration is also a crucial part in CVM

Once identifying the target population achieved, there will be a need to obtain a relevant sample represented for this population A sample is defined as the representative group for target population and a subject that survey will be implemented By sampling, instead of entire population, time and expenses are significantly saved Designing sample includes both the types and the number of interviewees The principle of sampling is that the target subset must be representative for the entire population and sufficiently large to produce unbiased and precise estimations Otherwise, there could be errors in sampling such as sampling error or sample selection bias To minimize the sources of these errors, an appropriate approach of sampling must be taken There are two main techniques of designing sample: non-probabilistic design and probabilistic design The former reflects that the probability of being chosen of interviewees is not fixed and decided

by the researchers, meanwhile in the latter, individuals have a fixed and non-zero probability of being chosen (Bateman, et al., 2002) Due to the importance of this, sampling is a crucial part of CV study

Another part of survey structure in CVM is choosing the survey mode There are

and face-to-face interviews which could be taken place in respondents’ homes or outside the home of respondents (Bateman, et al., 2002) The first has advantage in cost among these modes Due to the travel to respondents’ home, face-to-face mode has the highest expense and time consuming The mediate one, telephone interviews, is more expensive than mail surveys, but with the normally short calls its cost is still lower than face-to-face interviews However, the response rates are inverse to the cost-based ranking, which means that the face-to-face survey mode can touch the response rate at 70 per cent or even higher; while the rates of telephone and mail survey modes are normally at 60-75 per cent, and 25-50 per cent, respectively

In term of bias, the mail surveys can make self-selection bias when the large ratio of returned mails implies that they are interested in the survey’s topic Meanwhile the attendance of interviewers can also generate both negative and positive effects For example, respondents could feel uncomfortable to answer the sensitive questions related to income or political attitudes with the appearance of interviewers Or interviewees could react with the answers that make them nice to interviewers

The telephone survey mode can quickly get the results with calls but they are also easily refused There are some characteristics of these modes summarized in the following table

Table 1: Survey mode

Mail surveys Relatively inexpensive

Lack of interviewer bias Easier to answer sensitive questions

Can be completed at respondent’s own pace

Low response rates 50%

25-Self-selection bias Time-consuming Little control over who fills the questionnaire

Fixed question order

No clarification or probing possible

Restricts the use of visual aids

Respondent can alter earlier responses Telephone interviews Complex questionnaire

structures are possible Cheaper than personal interviews

Permits probing and clarification

Relatively quick to administer

Easy to monitor 60-75% response rates

No use of visual aids Restricts use of lengthy scales

Respondent may get tired Respondent may not answer sensitive questions Non-telephone or non-listed respondents not sampled

Face-to-face interviews Highly flexible

Complex questions and questionnaire structures are possible

Permits probing and clarification

Larger quantity of data can be collected Potential for extensive use

of visual and

Relatively expensive Interviewer bias Intercept surveys: samples normally not

representative and selection bias

self-Intercept surveys;

questionnaires have to be short

demonstration aids High response rates 70% + Greatest sample control Source: Bateman et al., (2002, p.106)

Each of them has both advantages and disadvantages that researchers should consider to select in order to come up with a qualitative data collection

In constructing questionnaire, the scenario and elicitation format are considerably important As any survey, CV survey also depends on the contexts, which means that with different prospects of the provided scenario and questions, the estimated values are contingent Bateman et al (2002) suggests that three main describing elements should be included in a valuation scenario: the policy (or program, or project) change of interest; the constructed market; and the payment method More details will be introduced in the section of survey design

After being provided with information about the goods and the scenario, respondents will be asked questions to decide the value of goods if they face the chances to reach it under some certain conditions In other words, that is the purpose

of measuring monetary values In principle, elicitation method could be performed

in various ways: the open-ended format, the bidding game, the payment card, the single-bounded dichotomous choice, and double-bounded dichotomous choice (Bateman, et al., 2002) The open-ended format is the most straightforward to apply for discovering values Respondents decide the values by themselves, thus the maximum WTP can be directly elicited from responses However, this method could lead to some problems, for instance, large number of non-responses, protest and zero answers, and outliers Moreover, this format is able to bring on unsound responses as argument of Mitchell and Carson (1989) in Bateman et al (2002) In the bidding game format, respondents will be repetitively asked whether they are willing to pay a certain amount Raising or lowering the amounts is based on

whether respondents have accepted the previous offers or not This iterative process will stop when a convergence of estimated WTP is achieved Payment cards present

to individuals a list of values (the values could be point estimate or a range of estimates) to make choice In a dichotomous or discrete choice CV format, respondents are asked to make decision of “yes” or “no” for voting a hypothesized change of goods provision

Therefore, a good design of scenario and an appropriate payment mechanism are important parts that will crucially contribute to the reliability and accuracy of responses

Because the CV method is based on survey, then the questionnaire structure or the ordering of questions is also an important stage in designing questionnaire Bateman

et al (2002) argues that the reasons of this importance are the effect of earlier questions on the answers of the later stage, and the encouragement to answer the rest questions

Beside the main and crucial parts above of survey designing, survey administration that consists of tasks from survey designing, pretest, to conducting survey needs crucially paying attention from the researchers, especially in the stages of pre-test and conducting main survey where the tasks are conducted by a group of people This allows the research to be on the right way as designed

2.4 Empirical applications

As mentioned above, this paper explores the attitudes and preferences towards the Vietnamese yew tree, and measures the WTP for the conservation of yew trees using CVM

To the best of my knowledge, there is no study valuing yew tree during last decade

by using CVM There are several studies applying CVM to measure the welfare changes of conservation of forest or avoiding deforestation Chopra (1993) stated

By using contingent valuation approach, Echeverria et al (1995) reported that the estimated WTP of existence value of cloud forest in Costa Rica is U.S

$121.05/respondent Kramer and Mercer (1997) applied the contingent valuation method to estimate the WTP valuation of U.S citizen on the world tropical rainforests On average, respondents presented a one-time of WTP around $21-31 per household to protect an additional 5% of tropical rainforests In recent years, there were studies about forest of Amirnejad et al (2005) and Andrea et al (2010)

By using CVM, Amirnejad and partners found that the estimated WTP for forests’ existence value was $2.51 per household/ month in Iran In the paper of Andrea et al., the mean WTP value of Swiss citizen for conservation of tropical forests was approximately CHF (Swiss franc) 110 per year

2.5 Empirical studies In Vietnam

In recent years, there are more and more researches applied stated preference techniques in Vietnam, especially in health care and environmental field Dang & Chennat (2010) applied the CVM to estimate the loss of value water resources causing by pesticide pollution They employed payment cards as elicitation method, in-person interview, and payment fees as payment vehicle4 The results showed that the WTP for improving water quality is VND 98,520/year/household Pham, Le, Petrie, Adams & Doran (2008) conducted a research about the households’ WTP for

a motorcycle helmet in Hanoi, Vietnam with changes in price of a helmet The authors used CV survey with both discrete-choice and open-ended questions to elicit the WTP Their finding was that the estimated WTP in average for a helmet is VND 163,794 By using both CVM and choice modelling (CM) methods, Tran & Navrud (2007) estimated the social benefits of restoration and preservation programmes for the My Son (a world cultural heritage site) in Vietnam In the application of CVM, they employed dichotomous choice questions with four bid levels for both

4

This is not correct in term of definition of payment vehicle (PV) PV is the channel through which the money is collected, for example: income tax, property tax, electricity bill

subsamples of residents and foreigners They found out that the two methods produce almost the same estimate Instead of using WTP, Navrud, Tran, & Bui (2012) used willingness to contribute (WTC) labour to a flood prevention program

in a CV study The face-to-face survey was conducted in Quang Nam province, Vietnam with sample size of 706 households They suggested that this CV approach

of WTC labor is applicable to the estimation of total welfare loss of natural disasters Truong (2005) studied about the WTP for conservation of Vietnamese rhinoceros by employing CVM In the drop-off survey, the author employed the single - bounded questions in eliciting the WTP, and electricity bill as payment vehicle The estimated WTP is approximately VND 40,000 per household

In health care segment, CV is also applied to find out the demand for some types of vaccines Do, Whittington, Le, Utomo, Nguyen, Poulos, Dang, Kim, Nyamete, & Acosta (2006) aimed to discover the demand function for typhoid fever vaccines by using CVM with a sample size of 1065 households in Hue, Vietnam The authors gave evidence that estimated mean WTP for a single vaccine varies from US$2.30

to US$4.80 In another healthy study, Kim, Do, Poulos, Le, Cook, Nguyen, Nyamete, Deen, Clemens, Vu, Dang, & Whittington (2008) conducted a survey on

800 randomly selected respondents for purchasing an oral cholera vaccine in Hue, Vietnam Their finding was that the estimated median WTP for 50% effective for 3 years vaccine is approximately US$5

Chapter 3: Methodology and Data

This chapter introduces the methodology that this paper applied, especially the method to estimate the WTP including non-parametric and parametric estimation, and the data collected from the CV survey

3.1 WTP estimation

This section describes the procedure to estimate the non-parametric and parametric WTP that will be employed to estimate the WTP based on single-bounded CV questions

For the linear random utility model defined in equations (8) and (9), the willingness

to pay (WTP) can be identified as:

∑௠௞ୀଵ  ଵ௞  ௝௞ +  ଵ (  ௝ −  ௝ ) +  ଵ௝ = ∑௠௞ୀଵ  ଴௞  ௝௞ +  ଴ (  ௝ ) +  ଴௝ (12)

The left hand side of equation (12) is the utility of respondent j-th with the CV program and net of the required payment - ௝ The right hand side is the utility of the status quo of that respondent In general, the WTP is the amount of money that makes respondents indifferent between the proposed CV scenario and the status quo

ஒ has mean zero and variance஢మ

ஒ మ, then the expectation of WTP is:

, |, ,  = ∏ ே 1 + ቀି൫ఈ௦ೕ ିఉ் ೕ ൯ቁ ିଵ௬ೕ1 − 1 + ቀି൫ఈ௦ೕ ିఉ் ೕ ൯ቁ ିଵଵି௬ೕ

where N is the number of households in the sample (N = ∑୎୨ୀ଴N୨) The log of the likelihood function is used to calculate the maximum likelihood estimates from the above function Then, the log-likelihood function is

ln
௉൫௬ೕ ୀଵ൯

ଵି௉൫௬ೕୀଵ൯  =  +  ଵ  ௝+ ଶ ௝+ ଷ ௝+ ସ ௝+ ହ ௝+ ௝ (17) The parameters will be estimated by using maximum likelihood method The expected sign of ଵ is negative, because with the higher bid level people are less willing to vote for the program The expected signs of coefficients, ଶ, ହ are positive because respondents with higher educational level and positive attitude are expected to be more willing to pay Meanwhile, the sign ofଷ, and ସ could be negative or positive

3.2 Survey design

As mentioned in chapter 2, a well design of scenario and payment mechanism is really important to elicit precise and reliable responses (Bateman et al., 2002) According to Bateman et al (2002), there are three fundamental elements in a valuation scenario The first is to describe the policy change of interest Single impact policies and multidimensional policy are what researchers may be attracted

in valuing policy change If a policy has a single impact, the description of its change which is valued consists of several steps: attributes of the investigated good such as characteristics, benefits related to use or non-use, extent of geography, being rights of property should be introduced by a way that makes respondents easy to understand or conceive the meanings; it is recommended to describe the available substitutions for interest, and alternative possibility of expenditures that may

influence the elicited values; and, of course, it is needed to state the proposed change of policy and the according change of the attributes of interested good

For a multidimensional policy, when the total value of the policy is required, the scenario should introduce information about each component and all of its congruous elements, and also provide the relative changes to the status quo What if

it is only a specified change needed valuation, and it is just part of including policy which encompasses various simultaneous occurrences? If that is the case, it is needed to present that single change as part of wider package Then respondents get chances to consider all possible effects including substitution, complementary, and income effects between elements of that package In another situation, when the change being valued is part of sequent changes of a policy, the point at which it appears in the sequence will affect its value And this value becomes little by little smaller when its place is deeper and deeper in the sequence of valuation (Hoehn and Randall, 1989 cited in Bateman et al., 2002)

At the beginning, focus group discussions (FGDs) were applied to explore the problem The participants were rangers or who has good knowledge about this field The first FGD was to identify the current situation of yew, and opinion of participants about how to develop the current program of conservation of yew tree

In the second FGD, the participants were ordinary people This task provided some suggestion about the bid levels would be used to elicit the respondent’s WTP

The most important thing in CV method is the scenario that provides information about the good being valued A brief introduction of Yew is presented in the first paragraph which also includes its use and non-use values Then it is the current situation of Yew and its conservation plan conducted by the local government The limitations of the current program are the basics of proposed program used in this

CV survey Respondents then will be asked to vote for this proposed program of Yew conservation with a specified amount of money; and it is a one-time payment

The questionnaire is constructed in 4 parts The first is the general problems facing the country; the second is the attitude toward conservation program and knowledge about selected endangered species; the third part describes the conservation plan for yews; a scenario will be described in this part to explain the problem more clearly and to take the answers about the willingness to pay; the final part consists of a series of questions about the socio-economic situation of respondents Before conducting the final survey, the questionnaire was pretested several times in order to check their availability and validity After revising, survey would be conducted as planned

The single-bounded questions were employed in the questionnaire to identify the amount of voting WTP from interviewees According to Freeman (2003), there are

at least three advantages of discrete choice (DC) format relative to the bidding game, and open-ended question: firstly, people feel as in real life when they decide

to buy or not a product in a supermarket given offered price Secondly, because that they just answer yes or no to the proposed question And it is simple to make decision, not like in an open-ended question that they could feel uncomfortable or refusal to give an answer Thirdly, this format could lead respondents to answer truthfully in survey, called the incentive compatibility The reason of choosing single-bounded DC instead of double-bounded DC is that the design of latter is more complicated and could lead to downward bias of estimated WTP due to the low designed range (Freeman, 2003) Follow-up questions would also be included

to find out the reasons of against WTP (including zero WTP) and of supporting program Respondents were asked these questions (DC and follow-questions) two times, the first was right after the scenario part, and the second took place at the end

of the third part of the questionnaire

The next step is to choose sample This paper employed the two-stage randomly sampling The sample size is 360 respondents with six bid levels5 The survey is

5

VND 5,000; VND 15,000; VND 30,000; VND 60,000; VND 250,000; and VND 500,000

conducted in Ho Chi Minh City There were five districts randomly selected including District 4, District 7, District 10, District 12, and Tan Binh District Then,

in each district five streets were randomly chosen6 On each street, interviewer started with a random address to implement the survey The rule is that each five households, takes one, and in case of building, office, apartment it is automatically skipped This process would be finished when interviewers have got enough number

of participated respondents on each street

The survey mode was applied is the drop-off mode, including some free conversation with respondents at the initial contact to introduce briefly the purpose

of survey, and at the time of picking questionnaire up to explain what respondents still misunderstand The drop-off method has been proved to increase the response rate relative to mail survey (Douglas et al., 1970; Stover and Stone, 1974; Lovelock

et al., 1976; Mangione et al., 1982; Olsen et al., 1998; Melevin et al., 1999; Steele et al., 2001; Riley and Kiger, 2002; Clark and Finley, 2007 in Allred and Ross-Davis, 2011) The questionnaire will be delivered to the respondents and picked up a few days after when respondents completed questionnaire This method combines characteristics of mail survey and personal interviews (Bateman et al., 2002) There are several advantages of drop-off in comparison with mail survey: potential personal contacts, verbal communication, and the ease of returning questionnaire The researcher has the chance to have personal contacts when he/ she deliver the questionnaire to respondents at their home (face-to-face contact) The purpose of study and the important participation of respondents are able to be explained at that time With the verbal communication, researchers have the opportunities to check whether chosen respondent are suitable or not in terms of criteria, for instance, age, owner relationship, etc Another noticed advantage of drop-off mode is the ease to return the completed questionnaire, because it will be picked up by researchers at designed time Even though there are still several disadvantages such as high conducting costs, possible inability to complete the questionnaire of respondents at

home due to respondents’ health, busy time, ability of literacy, this method has been continued to be applied in current researches of natural resources To make incentive to interviewees’ responses, there will be some gifts in cash7 According to Dillman et al (2009) in Allred and Ross-Davis (2011), respondents commonly finish questionnaire to gain rewards now or in the future Gifts in cash will give incentives to respondents in answering questionnaire This could reduce the bias in some cases that respondents have no incentive to complete questions, for example, they do not like the subject, to participate in the survey, or just recognize that there

is no benefit for them

The payment method if respondents vote for the conservation is the electricity bills This method is suitable because most households in Vietnam use payable electricity Meanwhile, other methods like water bill, tax are not appropriate Because households in country side do not buy water, they use own well for example; or people feel not comfortable in paying tax due to the difference in income which means that higher income people may pay more than lower income people Moreover, at this time people are encouraged to consume more by the government’s policy such as the exemption and reduction of personal income tax, so that it is impossible to get acceptance from people in voting for increasing their own tax

Beside the questionnaire, this research planned to apply a series of discussion with managers of the two Yew Reservations, and interviewing key informants (rangers)

7

Respondents have two options: receiving 20,000 VND or choosing a gift like pen or candy The latter is what questionnaire-completed respondents suggest

Chapter 4: Results and Discussions

In this section, the results of descriptive statistics and WTP estimation will be presented respectively The former part describes the descriptive statistics of the sample, including the analyze of WTP questions The latter consists of both non-parametric and parametric estimation of WTP

Figure 1: Distribution of respondents’ age

In total, the groups of age 21-30 and 31-40 are the highest with 26.6% and 33.65% respectively Across the groups of age, the ratio of married respondents is very high

in comparison to single and other marital status, except the first two groups under

20 and from 20 to 30 years old The married rate is 77% (Figure 2)

Figure 2: Cross - tabulation of marital status and group of age

From the Figure 3, approximately 31% of respondents reported that their major job

is doing own business (self-employment), and about 24% of them are private employment The rate of people working as laborer, mechanic, tailor, and skilled worker takes the third place with 15.87% of respondents The rate of people working for government is 12.7%

Figure 3 also shows that there is a huge difference between male and female ratio in this main occupation (32% and 18% respectively) However, the rate of male

working as their own business is less than that of female (28% and 33% respectively)

Figure 3: Occupation by gender Figure 4 illustrates the levels of education of respondents Almost people have gone

to school There is only 1.6% of respondents having no school So that, the average schooling years is very high (13.1 years) The high school and university levels have the highest rate, 35.7% and 33.1% respectively

Fisherman/Farmer Laborer/Mechanic/Tailor/Skilled Worker

Overseas Foreign Worker

Pensioner (Retired)

Other

Total Male Female

Figure 4: Levels of education The results in the Figure 5 show that the difference between male and female is strongly significant in both these levels There is more female than male in high school, but it is reversely in university level Across the rest the difference is not much significant

Figure 5: Gender in levels of education The statistical result in the Figure 6 describes the distribution of reported household income In which the highest level of income is the interval from 4 to 6 million VND per month In general we can see that average income of household may be around 10 million per month which prevails in this distribution

Average household size is 4.5 persons, this is higher than the overall level of Vietnam in 2010 (3.89 persons/household)8 In average, the number of income earner of each household is 2.48 persons Average household income is 9.27 million VND per month And monthly average income per capita is 2.05 million VND which is higher than that of the country in 2010 (1.387 million VND/month)9

Figure 6: Distribution of reported household income 16% of respondents stated that they have a better economic situation than others, and 80% of respondents reported that they are at average or lower than average level

100% of households use electricity provided by Electricity Company More than 90% of them pay fully electricity bill every month And the average bill of electricity is 662,000 VND/month which takes 7.14% of average household income

Attitude toward general problems and issues in Vietnam

Figure 7 gives the statistical result of problems that country has been facing Economic issues are ranked as the most important among the problems that Vietnam

is facing (36%) Followings are government and governance problems, and education

Figure 7: Country facing problems

Environment is ranked at the fourth place, more important than other issues such as health, foreign relations, violation and crime, or infrastructure However, the difference between the forth and the two previous ranks is not considerable (11%, 14%, and 15% respectively rank for environment, education, and government & governance) There are 66% of answerers reporting that the problems of the environment and natural resources are not properly addressed in Vietnam

Among environmental problems, air pollution is rated as the most important issue with a proportion of 43% of respondents (Figure 8) The next two are water pollution and deforestation, 25% and 11% respectively The loss of endangered species is ranked as the fifth It is also the last place (5%), because the rest three are not chosen (soil erosion, destruction of coral reefs, and other)

Figure 8: Environmental issues

Knowledge and attitude toward endangered species

Among six presented species, Dalbergia Tonkinensis Prain (Sua) is ranked as the most crucial species (31%) that need to be protected (Figure 9) This might be explained that the information about this kind of tree has been widely informed through public media in Vietnam during previous time Yew and Dalbergia (Cam) are the next two (25% and 17% respectively) The rank of Yew could be biased due

to the introduction of questionnaire Moreover, the number of respondents that have seen Yew is very low, about 18%; but the correct answer for question about the type

of Yew is just 9% of respondents In other words, most of respondents have not seen

the Yew and have not much knowledge about the Yew, but they still rank it as an important species that highly requires protection The plausible reason here is that people perceive the importance of Yew in the introduction part of questionnaire or the initial contact with interviewers

Figure 9: Ranks of proposal species

Table 2: Statements

Statement 1 There are more considerable environmental problems than conservations of

endangered species Statement 2 Poaching, exploiting, and trading endangered species should be punished by laws Statement 3 The conservation of endangered species should be a priority concern of the

government Statement 4 The endangered species are important even though we do not see or have an

interaction with them Statement 5 Everybody has duty to guarantee for existence together with mankind in the future of

current known plants and animals Statement 6 The government should increase more monetary resource to resolve the

environmental programs in Vietnam Statement 7 The government should increase taxes to pay more for endangered species

Statement 8 The government should pay attention to helping people before spending money on

endangered species conservation Statement 9 High-income household should be taxed more to pay for conservation of endangered

Pinus Yew Calocedrus Dalbergia

species Statement 10 People should contribute by cash to endangered species conservation

Figure 10 shows that people feel free and are generous when stating the general statements concerned to endangered species without any relation to government’s activities or monetary contribution In general, more than 60% of respondents agree

to the first 5 The common characteristics of these statements are that in general they describe the common attitude toward the environmental problems which could

be accepted by a large proportion of population; they do not relate to any specified contribution of individual or to governmental activities that may affect people’s benefits or obligations

Figure 10: Attitude toward endangered species

But with the statements related to government or themselves as in statements from 7

to 10, people react cautiously The actions like increasing taxes or contributing by cash payment from the government are not strongly supported Because respondents

In the second FGD, the. .. 5: Gender in levels of education The statistical result in the Figure describes the distribution of reported household income In which the highest level of income is the interval from to million