Applying contingent valuation method for estimating willingness to pay to control urban flooding in ho chi minh city

Research problem and objectives In summary, to deal with the urban flooding issue, not to mention other smaller projects, there are two main projects have been approved and in the proce

UNIVERSITY OF ECONOMICS INSTITUTE OF SOCIAL STUDIES

VIETNAM – NETHERLANDS PROGRAMME FOR M.A IN DEVELOPMENT ECONOMICS

APPLYING CONTINGENT VALUATION METHOD FOR ESTIMATING WILLINGNESS TO PAY

TO CONTROL URBAN FLOODING

IN HO CHI MINH CITY

BY

NGUYEN DUY CHINH

MASTER OF ARTS IN DEVELOPMENT ECONOMICS

HO CHI MINH CITY, DECEMBER 2014

UNIVERSITY OF ECONOMICS INSTITUTE OF SOCIAL STUDIES

VIETNAM – NETHERLANDS PROGRAMME FOR M.A IN DEVELOPMENT ECONOMICS

APPLYING CONTINGENT VALUATION METHOD FOR ESTIMATING WILLINGNESS TO PAY

TO CONTROL URBAN FLOODING

IN HO CHI MINH CITY

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

MASTER OF ARTS IN DEVELOPMENT ECONOMICS

By

NGUYEN DUY CHINH

Academic Supervisor:

DR TRUONG DANG THUY

HO CHI MINH CITY, DECEMBER 2014

ABSTRACT

Despite the implementation of many large scale projects involving in controlling urban flooding in Ho Chi Minh City, the issue of urban flooding has been a long-lasting issue for inhabitants in HCMC There are many causes for the problem, both objective and man-made This study conducts a contingent valuation (CV) study to find out the willingness to pay (WTP) for the controlling of urban flooding issue The

CV survey was done with the direct survey instrument on 180 households in HCMC Double-bounded dichotomous choice question was also used Non-parametric and parametric estimates for mean WTP are VND 464,654 and VND 380,000 per each household respectively Bootstrapping procedure further solidifies these results

LIST OF FIGURES

Figure 1 Measure of change in human welfare 4

Figure 2 WTP and WTA in the case of perfect substitutability 6

Figure 3 WTP and WTA in the case of imperfect substitutability 6

Figure 4 Total economic value 8

Figure 5 Monthly mean rainfall in HCMC and mean sea water level at Vung Tau 16

Figure 6 Number of flooding locations in HCMC from 2008 to 2014 18

Figure 7 Open-bid responses from the pilot survey 27

Figure 8 Project 1547 overview 31

Figure 9 Protection for zone I 32

Figure 10 Protection for zone II 33

Figure 11 Distribution of age in the sample 38

Figure 12 Monthly income distribution and House condition 39

Figure 13 Income and ownership of households with ‘no’ response at VND 50,000 41

Figure 14 Survivor function of non-parametric estimation 42

Figure 15 Monthly income classified by education groups 49

Figure 16 Crystal Ball bootstrapping result for non-parametric WTP 67

LIST OF TABLES Table 1 Total economic value of the project 1547 10

Table 2 Three types of commonly used elicitation questions 20

Table 3 Details of 18 sets of bid used in the survey 27

Table 4 Distribution of observations across areas 28

Table 5 Main contents of the questionnaire 29

Table 6 Interval distribution of WTP responses 34

Table 7 First bid question response statistics 40

Table 8 Second bid question response statistics 41

Table 9 Variables specification 44

Table 10 Logistic regression result 46

Table 11 Unrestricted model and Restricted model comparison 48

TABLE OF CONTENTS

CHAPTER 1: INTRODUCTION 1

CHAPTER 2: LITERATURE REVIEW 3

2.1 Willingness to pay and willingness to accept 3

2.2 Total Economic Value 7

2.3 Utility Theory and the Utility Difference Approach 10

2.4 Contingent Valuation Method 12

2.5 Some relevant CVM case studies 13

CHAPTER 3: METHODOLOGY 15

3.1 Current flooding and flooding control state in HCMC 15

3.2 The procedure of conducting a CVM study 19

3.3 Survey procedure of the study 23

3.4 Main contents of the questionnaire 29

3.5 Description of the hypothetical project 30

3.6 Non-parametric estimation technique 33

3.7 Parametric estimation technique 35

3.8 Confidence intervals of mean WTP 36

CHAPTER 4: ANALYSIS RESULTS 38

4.1 Descriptive statistics 38

4.2 Non-parametric estimation result 42

4.3 Parametric estimation results 43

CHAPTER 5: CONCLUSION 51

5.1 Main findings and policy implication 51

5.2 Limitations of the study 52

REFERENCES 54

APPENDIX 59

CHAPTER 1: INTRODUCTION

Ho Chi Minh City (HCMC) is one of the largest city and an important economic center in Viet Nam With the population of 7.8 million people and density of 3,721 people per kilometer square (Wikipedia, 2011), the development of infrastructure cannot catch up with the rapid urbanization rate of HCMC The inconsistency in the urban development in HCMC caused various problems, one of which is the exceedance in capacity of the urban sewer and drainage system, one of the causes for major urban flooding (Hoc, 2008) Moreover, the instability of the weather also further aggravates the flooding issue through the heavy rainfall and high rainfall level

In the effort to address the flooding circumstance, the Government has approved two major plans The first one is the plan of urban drainage improvement and sewer development of HCMC, which was approved by Decision 752 in 2002 The second

is the MARD1 construction plan, which was approved by Decision 1547 in 2008 Both projects have the same goal to help alleviate the flooding in HCMC, but they solve in different approaches While the former focuses on rehabilitating and developing the inner urban drainage system, the latter seeks to build large scale hydraulic constructs around the city However, due to the diversity in causes of flooding in HCMC, in order to completely solve the problem, the combination of both projects is required

Research problem and objectives

In summary, to deal with the urban flooding issue, not to mention other smaller projects, there are two main projects have been approved and in the process of implementing, which are urban drainage improvement and sewer development of HCMC (project 752) and hydraulic construction plan (project 1547) However, project 1547 is facing difficulties in appealing investment and cost underestimating, which may delay the project until 2025 Although a specific number of expected

1 MARD stands for Ministry of Agriculture and Rural Development

outcomes of the two projects were not given, CBA of these projects expected to relieve all the tangible damage caused by flood in 50 years, which means no heavy flooding will be occurring in the next 50 years (Steering Centre for Urban Flood Control Program, 2013)

Are all the aforementioned measures effective or not? Inspection of the all the projects upon completion will be required to answer the question in the future Although a CBA analysis for project 1547 has already been conducted However, the estimating of the benefit of the project in the CBA is based on the ‘flood risk approach’ which did not consider the demand and willingness to pay (WTP) of the inhabitants in HCMC In addition, the results of the study may serve as a guide for making decisions involving potential urban flooding project implementation and fee collection in HCMC as well as in other areas Furthermore, potential projects involving flood control in HCMC may also require WTP as an important indicator to estimate the benefit of the project Therefore, this study intents to:

 Evaluate inhabitants’ level of awareness about flood risks in HCMC

 Determine the aggregate WTP for a hypothetical anti-flooding project

 Find out the factors governing the WTP of HCMC inhabitants

In order to answer these questions, Contingent Valuation Method (CVM) will be applied in conjunction with the direct interview survey instrument to measure the willingness to pay for the elimination of urban flooding in HCMC The result in the study may be used in policy making for deciding whether or not a project would be feasible to be conducted in the future It may be also helpful for the process of inspecting the effectiveness of the project 1547 and 752

CHAPTER 2: LITERATURE REVIEW

Benefits and costs of a public good or an environmental good are difficult to determine since public goods or environmental goods usually do not have a price Moreover, the costs and benefits are mostly dependent on individuals’ preferences Given a public good is provided, the individual’s benefit can be measured by measuring how much that individual is willing to give up to obtain that public good

On the contrary, the individual’s cost when a public good is lost is measured by the value of something else that individual would accept to compensate for the loss

In terms of welfare, money is usually used as a standard measure In that case, the measure of benefit is the willingness to pay (WTP) to obtain the benefit or willingness

to accept (WTA) to compensate for the lost The measure of cost is WTP to avoid the cost and WTA to tolerate the same

Many techniques have been invented with the purpose of estimating the WTP and WTA for the provision or decrease in public goods Bateman et al (2002) classified techniques into revealed preference techniques (RP) and stated preference techniques (SP) Contingent Valuation Method (CVM) and Choice Modelling (CM) are commonly used SP techniques by researchers to estimate WTP and WTA While CM involves in inferring a value from the change in attribute levels of a public good, CVM directly collects individuals’ stated WTP by eliciting a scenario for respondents This chapter will present some basic concepts of economic valuation, which are WTP, WTA, Total Economic Value (TEV), Utility Theory, the Contingent Valuation Method and several CVM empirical applications

2.1 Willingness to pay and willingness to accept

Pearce (1997) defined willingness to pay is the monetary valuation that was placed

by an individual for a good or service WTP is constrained by ability to pay so that any people with higher income will value goods or services more highly than that with lower income Field (1997) and Pearce (1992) also pointed out that willingness

to accept is not constrained by the individual’s income, as is willingness to pay Thus, when people are asked willingness to accept questions, their answers are usually higher than their willingness to pay for the same item

According to Bateman et al (2002), the concept of WTP and WTA can be graphically illustrated using indifferent curve The vertical axis represents the expenditure in money unit of an individual on the private good (y) The horizontal axis represents the quantity of a public good (x) The indifferent curve I and I’ represent two linked combinations which have the same level of utility of two distinct individuals, with I has a lower level of utility than I’

Figure 1 Measure of change in human welfare

Source: Bateman et al., 2002

If there is a change in the quantity of a public good, the value of the change can be measured using four measures, two of which are WTP and WTA First, suppose there

is an increase in the quantity of public goods from x0 to x1 and the individual’s initial consumption point is at A With the increase in the quantity of public good, the

individual can enjoy x1 of the public good, but the private consumption is reduced by

BC, in other words, the new consumption point is now C The amount BC is defined

as the WTP for the increase in the public good (or the compensating variation for the increase in the public good)

Second, if the initial consumption of the individual is at B and there is a decrease in

quantity of public goods from x1 to x0, but the individual enjoys a higher level of private consumption, y1, moving the consumption point from B to D The amount

DA is the amount of private consumption needs to be compensated for the loss in the public good, specifically in this case, the WTA for the reduction in the public good

is DA (or the compensating variation for the reduction in the public good)

The third and the fourth measure are equivalent variation for the increase in the public good (equivalent gain) and equivalent variation for the decrease in public good (equivalent loss) Theoretically, they are equal to WTA and WTP, respectively, in value, but are distinct concepts elicited by different types of questions Equivalent variations are obtained by asking ‘how much would be good/bad as gaining/losing public good X’ instead of ‘how much would pay/compensate for the gain/loss public good X’

In theory, WTP and WTA have to be equal or only some insignificant divergence is allowed, but major differences between WTP and WTA have been illustrated by many studies Brown (2000) pointed out the difference between WTP and WTA still exists if the following conditions are not met: (i) no income effect; (ii) no transaction cost; (iii) perfect information about goods and prices; (iv) a market bring back a set

of truthful references In practice, the gap between WTP and WTA always exists because it is unrealistic to meet all the conditions in a single study case Nas (1996), Hanley (1997), and Hanemann (1984) stated that the divergence between WTP and WTA originates from the substitution between private and public good, so the equality between WTP and WTA is only achieved in the case of perfect substituted

environmental goods and services The substitution effect can be illustrated in the figure below

Figure 2 WTP and WTA in the case of perfect substitutability

Source: Hanley (1997)

Figure 3 WTP and WTA in the case of imperfect substitutability

Source: Hanley (1997)

2.2 Total Economic Value

The net sum of all relevant WTAs and WTPs is defined as the total economic value (TEV) of any change in welfare from a policy or a project The TEV is commonly classified into use and non-use value Use value is classified into actual value and option value, non-use value can be disaggregated into existence value, bequest value and altruism value The following figure demonstrates TEV and its categories

First, use value consists of two components Actual use relates to the actual use of the good or service (for example, a visit to a park, the prevention of potential damage of

an urban flooding prevention project) Option value refers to the willingness to pay

to preserve the good for using it in the future Amongst stated and revealed preference techniques, there are two commonly used techniques to value the use value of an environmental good, Travel cost method and Hedonic pricing method

Travel cost method is traditionally used to evaluate the recreational value for recreational sites, park, hunting sites by estimating the cost of related prerequisites for the trip such as gasoline costs, the opportunity cost of travelling, entry fee, service fee in the site, or in general, ‘travel cost’ This travel cost is heavily dependent on the distance to the site Specifically, the demand curve of total visits to the site with respect to added travel cost is derived from the survey data of visiting rate from different zones from the site, zonal population and travel cost from the zones After having obtained the demand curve, the value of the site can be inferred by calculating the consumer surplus of the curve Since the heavy dependence of TCM on the ‘travel cost’, which can be regarded as a directly revealed preference for an individual’s recreation value Moreover, total welfare of a site does not consist only recreational value, TCM is concluded to have the ability to measure use value, or more specific, actual use value, of TEV only

Another popular method of valuing the use value of an environmental good is the Hedonic Pricing Method (HPM) HPM based on the concept that the price on the market for a good comprise the price of associated amenity For instance, the price of

a specific house on the market could have the value of the beautiful surroundings internalized in HPM attempts to modelize the price of a market good (for example, house price) against related characteristics, in which have the amenity of interest From the function, the marginal value of the amenity, in other words, the ‘implicit price’ of the amenity, can be inferred A demand curve for that specific amenity is then constructed, and consequently, the economic value of the amenity However, HPM was not developed to capture the total economic value of an environmental good, since the implicit values internalized in a market good may not include non-use value For example, the value of preserving the existence of surroundings without being benefited from it (existence value) could not be implied in any market good price Therefore, it can be concluded that HPM and TCM are only suitable for

Figure 4 Total economic value

Source: Bateman et al., 2002

Second, the non - use value relates to WTP to keep some good exists, though there is

no actual or possible use The non - use value consists of existence, altruism and bequest value Existence value refers to the WTP to have a good exists and the individual eliciting WTP has no use for himself or for someone else The reasons for maintaining the good could be the ‘feeling of responsibility’ for the good or the

‘concern’ for the existence of a good (for instance, a rare species of animal) Altruistic

Total economic value

Altruism Bequest

value relates to the WTP to keep the good maintained to others in the current generation Similarly, bequest value may come when the person concerns that the good should be available to others in future generations

Covering TCM and HPM weakness of not being able to measure non-use value, Contingent Valuation Method (CVM) is one of the methods used to value the non-use value According to Mitchell and Carson (1989), CVM has many advantages over other methods, one of which is the ability to capture the existence value CVM involves in describing a hypothetical scenario for the respondent to state their WTP for it Therefore, it is suitable for the valuation of environmental goods with an easy-to-perceive aspect, such as the cleansing of a river or the provision of electricity in a remote area Due to the difficulty in capturing the actual use value of such goods, plus respondents of a CVM study rely on their own preferences, which is ‘contingent’ upon the hypothetical good, to give out their WTP, CVM is unable to capture the use value of the good However, CVM can be used to measure the non-use value effectively

According to Steering Centre for Urban Flood Control Program (2013), the total economic value of the project 1547 could be summarized as following

Table 1 Total economic value of the project 1547

Actual use  Flood damage prevention (the damage could be cleaning

cost, infrastructure damage, loss of productive land, crops, livestock, human lives, ecological goods…)

 The feeling of safety

 Cultural heritage and aesthetic

2.3 Utility Theory and the Utility Difference Approach

Following economic theory, Bateman et al (2002) defined the general form of indirect utility function as:

V(Y, P, S, Q)

The function V(.) can be described as the utility a household can acquire from the income (Y) with the prices of goods (P) (Higher or lower prices enable households to have more goods and thus, raise the utility) and the level of non-market good (Q) (Higher or lower level of non-market good represent an improvement or a decline) The utility is also assumed to be dependent on other socioeconomic factors (S)

Assuming there is a provision of non-market good that raise the level of non-market good from initial Q0 to Q1 The household’s utility will be pushed up to a higher level:

V(Y, P, S, Q) < V(Y, P, S, Q1)

With a higher level of non-market good, the household has to pay an amount to achieve a better well-being, but the higher the amount, the less utility the household will enjoy due to the reduction in dispensable income So the maximum amount the household willing to pay for the improvement can be described as the amount that will bring the household’s utility back to the state without the improvement, or mathematically:

V(Y, P, S, Q0) = V(Y − C, P, S, Q1)

With C is the household’s maximum WTP, or in other words, the compensating variation for the change in the well-being C can be written as a function of other parameters in the model Moreover, C is constrained by the income, which means the maximum WTP cannot exceed income:

C = C(Q0, Q1, Y, P, S) = WTP ≤ Y

Assuming a household indirect utility function has the linear form With the assumption of the price of market goods and socioeconomic factors of that household fixed The function of the household utility before the change (status quo) in the non-market good can be written as follows:

C =α1−α0+ μ1−μ0

β

The amount α1−α0+ μ1−μ0 can be interpreted as the utility difference that comes from the provision of the non-market goods This observable portion of the utility difference can be further parameterized as a function of household other socioeconomic characteristics (X1, X2,…, Xk) in a simple linear form:

α1− α0 = β0+ β1X1+ β2X2+ ⋯ + βkXk

2.4 Contingent Valuation Method

The contingent valuation method is defined by Mitchell and Carson (1989) as the method that collects people’s WTP for an improvement using survey questions The first CVM study was employed by Davis (1963) to estimate the benefits of outdoor recreation in Maine backwoods area The CVM is based on complex surveys to obtain information from respondents about their preferences for environmental amenities A hypothetical scenario or market is constructed to describe changes in goods or services Respondents are asked how much they are willing to pay to have the goods

or services improved or how much they are willing to accept to have the goods or services given up Then the data obtained from a sample of the target population are analyzed to infer the value of the good or service

CVM was stated as one of the four basic methods for valuing the environmental benefits of a proposed regulation Bateman and Willis (1999) stated that the CVM has the ability to quantify certain types of benefits such as non-use and passive-use benefits, which is the superiority of CVM over travel cost (TCM) and hedonic pricing (HPM) method By the end of 1970s, CVM was recognized officially by the US Water Resources Council as an approved valuation technique CVM has been applied broadly in many countries since 1980 (Bateman and Willis, 1999), especially in developing countries (Do and Bennett 2009; Hoa and Ly, 2009) However, through many decades, many debates took place concerning the validity of this method Later

on, Cummings et al (1986) and Mitchell and Carson (1989) have raised the debate into its peak through their publications

Carson et al (1992) conducted a well-known study to evaluate the losses from the Exxon Valdez oil-spill disaster in Prince William Sound in Alaska, US The validity

of the CVM used in the study has been questioned by the company which was responsible for the disaster, Exxon Company Numerous adjustments have been made

to the study, the result was then finally accepted by the Exxon and US Government The estimate for the damage of the disaster was $2.8 billion (based on median WTP

of $31 per household)

2.5 Some relevant CVM case studies

In 1988, Thunberg conducted the first major study of WTP that focused on reducing flood risk The objective of the study was to find out the WTP to avoid trauma or psychological damage due to floods 142 respondents were scheduled by telephone and surveyed using personal interview in Virginia, US The proposed good to be valued in the study was the lowered probability of flooding (20% compares to 40%)

in the landowner’s house, respondents also were given a payment card format answer The study offers respondents with two alternatives of payment vehicles, lump-sum payment and annual payment A follow-up question and an opportunity to ask about the good were also given to respondents after the proposal to ensure the respondents had a clear understanding of the project to minimize bias The result showed a WTP

of $ 314 of lump-sum payment for the change However, the important contribution

of the study was the development of a WTP model for flood risk reduction

Shabman and Stephenson (1996) did several studies based on the results of Thunberg (1988) The authors aimed to compare HPM, CVM and the PDA (Properties damage avoided method) estimates from various studies together and with actual WTP choices from the authors’ own CVM survey The elicitation question type used in the study was the payment card question with a follow-up confirmation question to ensure there would not be hypothetical bias in responses The payment vehicle used

was special assessment annual payment paid by property owners The comparison showed that the hedonic pricing produced a higher mean estimate ($ 1,333) than CVM ($ 314) and the PDA method was between these two methods ($ 597) The study also pointed out that the WTP in the hypothetical scenario ($ 124) is far lower than the WTP when the actual choice ($ 93 each year) is faced by the respondents

Zhai et al (2007) applied CVM to estimate WTP for flood reduction in Japan The study surveyed a large sample size with the payment card elicitation question to find out the interest of people in various flood control measures To reduce bias, the careful survey procedure was conducted, including choosing a large sample (962 households chosen), gifts from the program and most importantly, a cover letter and

a thank letter with the signature of the Project Director, this will convince the respondents the survey will be the actual referendum for policy making WTP for different anti-flooding measures (early warning systems, structural measures) ranged from the average of ¥ 2.887 to ¥ 4.861 WTP for flood control measures may depend

on income, ‘preparedness’, acceptability of flood risk, perception of other risk and provision of environmental information

CHAPTER 3: METHODOLOGY

Based on the referred literature, this chapter will present the context in which the study is conducted, the application of CVM in the context of valuing the WTP for controlling urban flooding, some justifications about selecting the appropriate elements in CVM, the main contents of the questionnaire, description of the hypothetical scenario, and lastly, the estimation and bootstrap techniques used in the study will also be introduced

3.1 Current flooding and flooding control state in HCMC

Recently, flood controlling has been one of the biggest issues of Ho Chi Minh City authorities Not only in low ground areas, but also in the central districts, inhabitants have to face frequent flooding during the rainy season (Dung, 2011) Causes of flooding are tides, high discharges from Saigon and Dong Nai rivers, flooding from the Mekong Delta, heavy raining or a combination of these factors Moreover, the insufficient capacity and inefficiency of the drainage system, elevations of the city and land subsidence, inadequate spatial planning, policies and regulations, lack of public awareness and participation has aggravated the flooding problem (Steering

Centre for Urban Flood Control Program, 2013)

Current flooding situation in Ho Chi Minh City has caused significant economic damage and impacted the health of HCM City inhabitants The estimated amount of economic damage caused by flooding every year is ranged from six thousand billion VND to 22 billion VND (Steering Centre for Urban Flood Control Program, 2013) The situation is getting worse by three factors which are partly man-made: i) sea-levels and rainfall instability which are affected by the climate change ii) land-subsidence resulted from numerous high-rise buildings and underground water exhaustion and iii) Sewer rehabilitation and flood protection is falling behind the rapid urbanization process (Truong, 2010)

Figure 5 Monthly mean rainfall in HCMC and mean sea water level at Vung Tau

Source: Steering Centre for Urban Flood Control Program (2013)

Figure 5 shows the average rainfall in HCMC and sea level at Vung Tau2 The rainfall stays consistently high from May to October but the sea water level is only at its peak

in October and November The occurrence of flooding is often lies within May and October, which is appropriate to the high rainfall season However, the peak of the flooding is in September and October, which is the months with highest sea level Therefore, the flooding problem can be concluded to have the high rainfall as the main cause while the high sea level, although may not be the directly influencing factor, further aggravates the long lasting flooding problem (Phi, 2009)

In 2001, plan of urban drainage improvement and sewer development of HCMC was approved (JICA 1999) by the Decision 7523 of the Government This is a big scale project involving in improving of 186 drainage routes, 132 of which have the support

2 Due to the lack of sea level tracking stations in HCMC Measurements from the Vung Tau station were used

3 This project have been renamed after Decision 752, to Project 752

from ODA sources from Japan and the rest of them funded by the city budget The total investment cost is approximated to be 40,380 billion VND (Decision 752/QĐ-TTg, 2001) The implementation of the project will take at least 20 years to complete with the target year at 2020 The completion of this project will drastically help the city in reducing the flooding caused by rain by enhancing water circulation from urban areas to canals, increasing hydraulic capacity of the urban drainage system, therefore help alleviating the inundation in the city when heavy rains occurred

Apart from sewer rehabilitation and improvement of urban drainage system, which was also essential for the urban flood controlling and urban development In order to deal with the high rainfall and the tidal level, which are the main cause for the flooding problem in HCMC, there have been many hydraulic plans proposed to the People’s Committee of HCMC and the Government Three of which are recommended by the Steering Center of The Urban Flood Control Program: The Soai Rap barrier, the MARD plan and a variant of the original MARD plan

In 2008, the Government issued Decision 1547 regarding planning and implementation of the MARD plan Due to the enormous scale of the project 1547, which is named after Decision 1547, this hydraulic construction project4 is split into

46 items Accordingly, main items consist of 12 sluice gates and 172 kilometers of dykes, will be built, enclosing entire HCMC and some area of Long An Province The project also involved in building 8 bridges and rehabilitating 27 small canals with the total estimated total investment cost of 11,531 billion VND However, the project only serves as a tidal controller role and gives better water discharges from city canal

to the Sai Gon river, while controlling urban flooding, especially flooding caused by heavy rains, also relies on the capacity and efficiency of the urban drainage and sewer system in the city, which has been constantly being done by the city since 2001 with the project 752 Without the enhanced sewer and drainage system, the system itself

4 The term ‘Hydraulic construction project’ is the name for projects involving in construction of large-scale structures like big dykes, dams… It is distinguished from ‘Irrigation’ or ‘Sewer and drainage rehab’ projects

only functions as an anti-tidal-flooding measure Currently, due to the lack of fund, only one sluice and 33 kilometers of dykes have been built (Decision 1547/QĐ-TTg,

2008)

In 2012, the city conducted the project aiming to rehabilitate the Tham Luong – Ben Cat – Rach Nuoc Len canal route, one of the most important canal routes in HCMC This is one item in the project 1547 The project focuses on improving the inundation condition of the Go Vap, Binh Thanh, Tan Binh, Tan Phu, Binh Tan districts and Binh Chanh by dredging the whole canal, building various sluices, sewers and water processors along the canal

Figure 6 Number of flooding locations in HCMC from 2008 to 2014

Source: Steering Centre for Urban Flood Control Program (2013)

Figure 6 shows the number of flooding locations in HCMC from 2008 to 2014 which reflects the effort of HCMC to reduce the urban flooding over 5 years A significant amount of flooding locations has been reduced, from 126 locations in 2008 to 11 locations in 2013 Currently, the central area of HCMC is almost alleviated from flooding However, the remaining 10 locations are still impaired heavily with flooding caused by rain and require complete improvement of the drainage system of that area

3.2 The procedure of conducting a CVM study

According to Bateman et al (2002) A typical CVM work procedure follows the steps:

 Validity and reliability testing

 Aggregation and reporting

First, the research question must be specified and a hypothetical market or scenario

is established The plausibility of the hypothetical market or scenario have to be ensured

Second, a choice is made from different types of survey: face-to-face survey, telephone survey or mail survey The face-to-face survey is the most preferred and proves to be the most efficient method, however, it may be a costly method in terms

of time and money (Bateman et al., 2002)

Third, we need to determine how many observations need to be collected, where and when the data must be collected and what method should be used in order to have a good sample

Fourth, the most important part of the procedure is to design the questionnaire This involves in eliciting the scenario used for asking the valuation question, choosing what kind of valuation question (such as single-bounded dichotomous choice, double-bounded dichotomous choice or open-ended choice question, the payment vehicle for

the hypothetical market or scenario and how the questions will be elicited to the respondents The table below presents three popular types of valuation question and its details

Table 2 Three types of commonly used elicitation questions

Open-ended question

Single-bounded dichotomous choice question

Double-bounded dichotomous choice question

Description Respondents are asked to

state a single amount of WTP for the scenario

Respondents are asked to give a discrete choice (yes or no) to a pre- specified amount of WTP

Respondents are asked to give a discrete choice to

a specified amount, then another higher (lower) amount will be given if the respondent answer yes (no) to the initial question

WTP obtained Continuous WTP data Discrete choice WTP

data

Interval WTP data

Advantages The exact value from

respondent, accurate if the respondents know their WTP well

Straightforward to respondents

Simplicity in WTP aggregating

Easier for respondent to answer since only a single yes-no question involved

Incentives compatible (Carson & Groves, 2007)

Easier for respondents since it requires respondents to give the yes-no answer twice Higher statistical efficiency than single- bounded question Incentives compatible (Carson & Groves, 2007) Disadvantages Too difficult for most

respondents (NOAA, 1993)

High non-response rate High possibility of WTP understating due to the lack of knowledge about proposed scenario (Mitchell & Carson, 1989)

Possibility of yea-saying and nay-saying

Large sample required

First bid question may impact the answer on the following question More complicated econometric techniques required

Other than these elicitation methods, there are also other methods such as bidding game format or payment card Bidding game is the format that the respondents will

face with multiple rounds discrete choice questions Optionally, a final question, which is an open-ended WTP question could also be included in the survey The main disadvantage of this format is the starting bias, which is the influence of the starting bid value to the proceeding answers, and the possibility of producing a number of outliers The bidding game is also impossible to apply with the mail survey instrument and instruments without the assistance of the interviewers (Bateman, et al., 2002) Payment card is the method which presents to the respondents a number of monetary amounts, usually begins with the lowest amount Then the respondents are asked to decide on each of the amounts as the amount goes up The payment card is a more developed version of the bidding game and open-ended format, since it reduces significantly the number of outliers and provides better estimates However, this method is very time-consuming and still difficult to respondents as it requires respondents to think very carefully to make a decision, besides, if the respondents have little knowledge of what being proposed and their actual WTP, this method may still produce invalid estimates

Payment vehicle is also important to a CVM study Payment vehicle is the channel through which the money will be collected from the respondents Inaccuracy in responses may occur when the actual WTP of the respondents lies between the bid and the expected cost of respondents when a compulsory payment vehicle is applied However, the use of voluntary payment vehicle (voluntary donation) also has the disadvantage of encouraging respondents to overstate their WTP when asked because when the payment is collected, respondents can pay any amount they want (Ivehammar, 2009) In summary, there is no ‘best’ payment vehicle, the choice of payment vehicles differs depending on the nature of the study and the reasonability

of the vehicle in a specific case (Cummings, Brookshire, & Schulze, 1986)

Fifth, the testing of the questionnaire must be conducted The testing must include a focus group interview, which is the process of interviewing a small group of respondent in an unstructured manner Focus groups are the most important

qualitative research procedure (Malhotra, 1996) It is also necessary to pre-test the revised questionnaire in a small number of respondents (pilot survey), the pilot survey stage holds the purpose of fine-tuning the survey and training the interviewers (Bateman et al., 2002) Then the actual survey is then carried out

Sixth, after having obtained the data, econometric techniques are used to estimate the WTP against other independent variables The complexity of the process is dependent

on the data For instance, interval data from double-bounded dichotomous type of valuation question need more complex techniques to estimate than continuous data

The next step is to test the validity and reliability of the result Is the survey understandable for respondents from a wide range of backgrounds and education levels? Or whether the variables correct and suitable for the measurement? Or whether other valuation methods produce the same result? (Bateman et al., 2002)

One important issue worth noting when conducting a CVM study is the problem of

‘hypothetical bias’, which is the phenomena of stating a different WTP on the survey than the WTP that same respondent would be actually willing to pay due to the scenario in the survey being hypothetical The underlying theoretical reasons for hypothetical bias has not yet been clearly defined in the literature (Loomis, 2011) However, Loomis (2011) synthesized five remedial solutions for the hypothetical bias from other literature The first solution is to (i) convince the respondent that the survey result will the ‘advisory’ referendum for policy makers (Carson & Groves, 2007) In that same study, Carson and Groves (2007) also pointed out the use of binary choice (two choice) questions will result in incentive compatible responses, which will also help reducing hypothetical bias The second solution is (ii) the use of

‘cheap talk’ (Cummings and Taylor, 1999) The ‘cheap talk’ is actually a reminding

to respondent that they should state their WTP truthfully as if this is the real decision using their money However, the effectiveness of this method is still very doubtful The third solution proposed by Mitani and Flores (2010) is (iii) to eliminate completely the uncertainties of respondents about whether they should actually pay

or not and whether the proposed good will be actually provided or not According to Mitani and Flores (2010), there will not be hypothetical bias if the respondents are clear about the probability that the good will be provided and the probability the respondents will make the payment However, in order to apply this method, in some cases, it is required for the researcher to be untruthful in the questionnaire design since not all hypothetical good made up by the research can be assured to be provided based on the study result Following the theory of respondent uncertainty, Champ et

al (1997) proposed (iv) the use of a follow-up question after the dichotomous choice question to define the level of certainty of respondents’ answers This solution has proven to be effective by some studies Finally, in 1994, National Oceanic and Atmospheric Administration (NOAA) proposed (v) a method of cutting down the final WTP by a percentage (50% in that study) (NOAA, 1994) However, the underlying background for the ratio is still unclear

In the final step, the result need to be aggregated The simple way to aggregate the result is to multiply the mean or median WTP with the total population Another way

to obtain the welfare change for an entire population due to the project or policy change is to regress the WTP responses against independent variables to obtain the parameters Then the parameters are used to produce an aggregate measure of welfare change (Freeman, 1993)

3.3 Survey procedure of the study

Although there are many methods to evaluate public good, such as hedonic pricing and travel cost, CVM is especially suitable in the case of studies involved in non-marketed goods and services, which exists in this study (Bateman et al., 2002) Moreover, the CVM is more feasible when implementing in developing countries, where the response rate is high compared to developed countries (Grosh & Glewwe, 2001) According to Mitchell and Carson (1989), CVM has many advantages over other methods, one of which is the ability to capture the existence value, which mean the individual willingness to pay to preserve the good or service In addition, there

has not been any study measuring the WTP for urban flooding prevention in Vietnam yet Therefore, the CVM will be used to measure the WTP for HCMC flooding prevention

The actual survey will be in face-to-face interview format After providing information about the flooding situation in HCMC, question about characteristics of the respondent and the house, household expectation about flood situation will be asked, then a hypothetical project will be set up The proposed hypothetical project will be the project 1547 upon completion, which was approved in 2008 by the HCMC People’s Committee and the Government plus some brief explanation of the project

752 involving rehabilitating the urban sewer and drainage system will also be given Although these projects have actually been implemented already, but due to their slow progress, the completed projects are still chosen for the hypothetical scenario The hypothetical scenario will be described as having the capabilities of alleviating the urban flooding completely when finished After describing the hypothetical project, the respondents will be given an opportunity to question about the project Then the bid question is given, in which respondents will be asked how much he/she willing to pay if the project need contributions from HCMC residents

Regarding the survey mode, the study applied was the direct in-home face-to-face interview mode Bateman et al (2002) introduced seven advantages of the face-to-face mode over other modes of survey, which are (i) the assurance of the household members stated, (ii) the order of information unfolding to the respondent, (iii) the assistance of the interviewer to the respondent, (iv) the correction of interviewer with misunderstanding to the questionnaire, (v) the motivation of the interviewer, (vi) the recording of verbatim comments made by the respondents and finally, (vii) the interviewer can monitor the respondents’ confidence to the important questions, this

is especially useful in question regarding attitude of the respondent to a specific disaster in this study In addition, computer-assisted personal interviewing (CAPI) was also implemented Specifically, an electronic tablet with the built-in

questionnaire will be brought to each household and will be used by the interviewer

to record the answer This will help cutting down the time spent on data entering and minimize the possibility of error due to entering mistake if the survey was done by traditional paper-survey mode

For the bid question, the study uses double-bounded dichotomous contingent valuation to ask for the amount that the respondents are willing to pay for the hypothetical project Then from the data, the WTP for the HCMC flooding prevention

is inferred The double-bounded dichotomous bid question is chosen due to the high statistical efficiency could be achieved compared to open-ended or single-bounded dichotomous-choice method according to Bateman et al (1999) This type of bid question requires each respondent to identify two pre-specified amount that bound their maximum WTP (one amount is higher and one amount is lower than their maximum WTP) Specifically, each respondent will be asked an initial amount (initial bid), if the answer ‘yes’ / ‘no’ is given, a higher/lower amount (higher/lower bid) will

be asked From the answers (yes-no, yes-yes, no-yes, no-no) and corresponding bounds, the WTP interval from each individual is inferred in the data analysis

The payment vehicle of choice is also crucial to a CVM study (Morrison, Blamey, & Bennett, 2000) Payment vehicle is the method by which the money will be hypothetically collected from the respondents Two payment vehicles were put into consideration, one-time voluntary payment and one-time mandatory payment in electricity bill of the household Although a mandatory increase in electricity bills seems to be a more convenient method than a voluntary donation for urban households, Morrison, Blamey & Bennett (2000) pointed out that one of the reasons for the occurrence of vehicle bias is respondents’ doubts about payment being one-off The request for an increase electricity bill may raise doubts to the respondent that, although being clearly stated, that increase will not be charged one time Furthermore, Ryan (2006) also stated collective payment mechanism usually results in a somewhat

higher estimate of WTP compare to voluntary mechanism To avoid bias, one time voluntary payment vehicle was chosen in this study

To address hypothetical bias In the context of the study, three solutions were chosen The first remedy used was giving the respondents the certainty that the hypothetical project has actually been implemented by project 1547 and project 752 (Cummings

et al., 1995; Landry and List, 2007; Mitani and Flores, 2010), the survey will give some information regarding the year of starting, the scale of these actual projects However, due to the actual slow progress of these projects, the current progress of these projects was not given the respondents to avoid protest bids The second solution was the use of vocal confirmation message after the respondents have completed the bid question More specific, the respondents will be asked whether they are certain about the bid or not, if they are not, they will be required to rethink and change their bid answers And finally, several emphasis will be made by the interviewer throughout the survey reminding the respondents to answer the questions honestly

At the beginning of the study, a focus group discussion is conducted, the main purpose of the focus group discussion is to gain insights about the issue of urban flooding, what has to be considered in the questionnaire and basic bid range suggestions The focus group consists of 6 people who have resided in HCMC for more than 10 years Participants are asked to give an opinion on the understandability

of the proposed scenario and decide whether a specific question is appropriate or not

After the questions in the questionnaire have been constructed and revised by the focus group To pre-test the questionnaire, a pilot survey whose main purpose is to determine the bid levels and check the validity of the questions was initiated 20 random households were selected and surveyed, they were asked to give the specific bid (open bid) on the scenario instead of choosing dichotomous yes-no The result for the open bid question in the pilot survey is presented as follows

Figure 7 Open-bid responses from the pilot survey

According to the collected pilot-survey data, 18 sets of bid were designed, these sets

of bids would be used in the final survey

Table 3 Details of 18 sets of bid used in the survey

Set # Initial bid

(VND)

Lower bid (VND)

Higher bid (VND)

Number of households

is distributed across areas as following

Table 4 Distribution of observations across areas

observations

After the survey has been conducted, the mean and the WTP would be computed using two methods, non-parametric estimation and parametric estimation For non-parametric estimation, Bateman et al (2002) proposed a method to deal with double-

bounded data, Turnbull Self Consistency Algorithm (TSCA), which is the method will be used in this study While parametric estimation will conduct a logistic regression to estimate mean WTP

3.4 Main contents of the questionnaire

The contents of the questionnaire are summarized as follows (for the full questionnaire, see the Appendix)

Table 5 Main contents of the questionnaire

 House location: district and street

 Ownership: Private or rental

 Distance from river or canal: (1) less than 100m; (2) 100

m ~ 500 m; (3) 500 m ~ 1 km; (4) 1 km ~ 2 km; (5) more than 2 km

 Zone: whether the house is in zone I or not Flood perception

(Zhai el al 2007)

 Past flood experience: Yes or no

 Perception of future: Do you think flood situation will eventually get better in the future (yes or no)

 Perception of consequences of current flooding (yes or no)

 Natural disasters: Hurricane, thunderstorms

 Environmental risk: Air pollution, noise pollution, water pollution

 Urban risk: Robbery, traffic accidents, fire hazards Private anti-flood

measures

 Does the house has private anti-flood measures? (yes or no)

much the respondent is willing to pay for the proposed project

3.5 Description of the hypothetical project

The hypothetical project is based mainly on the project 1547 is originally proposed

in 2008 First, the figure shows the division of HCMC into 3 zones (see figure 8)

 Zone I (yellow zone): contains all of the left-hand side of the Saigon river plus Nha Be District (central HCMC districts and a part of Long An Province)

 Zone II (red zone): contains the area around Saigon river – Dong Nai river fork (District 2, 9 and Thu Duc District)

tri- Zone III (blue zone): contains Can Gio district only

The project focuses on protecting zone I and zone II, zone III is considered to be an important ecological and vulnerable area So no measures are implemented in zone III in the project

Figure 8 Project 1547 overview

Source: Steering Centre for Urban Flood Control Program (2013)

Second, the protection for the zone I consists of 172 km of dykes and 12 sluice gates, protecting urban and rural areas of HCMC, more specific, zone I, against tidal

flooding and river discharges It also serves the role of controlling the water level in the protected area by enhancing the outer drainage system Furthermore, the inner urban drainage will also be enhanced by rehabilitate 186 urban drainage routes

Figure 9 Protection for zone I

Source: Steering Centre for Urban Flood Control Program (2013)

Third, the protection for zone II focuses on the raising of the embankment of lowlands and building smaller dykes to enclose ecological and tourism areas Canals and main rivers will also be dredged to improve drainage capacity

Figure 10 Protection for zone II

Source: Steering Centre for Urban Flood Control Program (2013)

3.6 Non-parametric estimation technique

An advantage of non-parametric estimation over parametric regression models is that the former do not make any model assumption Parametric estimation relies solely on the data on the bid levels offered and households’ choices The double-bounded dichotomous question produces interval data with overlapping interval Bateman et