Estimating the benefit from the algal bloom reduction – an application of the contingent valuation method
Master’s Thesis Anna-Kaisa Kosenius University of Helsinki Department of Economics and Management Environmental Economics May 2004
HELSINGIN YLIOPISTO HELSINGFORS UNIVERSITET UNIVERSITY OF HELSINKI Tiedekunta/Osasto Fakultet/Sektion Faculty
Laitos Institution Department
Faculty of Agriculture and Forestry
Department of Economics and Management
Tekijä Författare Author
Kosenius, Anna-Kaisa Työn nimi Arbetets titel Title
Estimating the benefit from the algal bloom reduction – an application of the contingent valuation method Oppiaine Läroämne Subject
Environmental Economics Työn laji Arbetets art Level
Aika Datum Month and year
Sivumäärä Sidoantal Number of pages
Master’s Thesis
May 2004
65 + appendices
Tiivistelmä Referat Abstract
The purpose of this study is to estimate the benefit from the reduction in harmful algal blooms (HABs) for tourism sector in the coastal city of Hanko facing the Gulf of Finland. The thesis is a case study of the European Union funded project ECOHARM that studied the impact of harmful algal blooms for tourism sector in the European Union. The contingent valuation method (CV, CVM) was applied to elicitate visitors’ willingness to pay for algal reduction. The answers were analyzed with the standard probit model and the average willingness to pay for algal bloom reduction was used to calculate the aggregate benefit for the tourism sector in Hanko. The results show that the proposed bid has negative effect on the visitors’ willingness to pay for the algal reduction while income, the importance of the algal blooms as the coastal problem, the vacation expenditure and the level of satisfaction of the respondent with the seawater quality have positive effect. In addition to willingness to pay, the study revealed information on tourism in Hanko and the visitors’ experience on harmful algal blooms. Slightly more than half (55.6 %) of the respondents had personal experience on harmful algal blooms and 40.8 % of the respondents stated algal blooms in the sea as the most important coastal problem in Hanko.
Avainsanat Nyckelord Keywords
algal bloom, contingent valuation, water quality, tourism Säilytyspaikka Förvaringsställe Where deposited
Faculty of Agriculture and Forestry, Department of Economics and Management Muita tietoja Övriga uppgifter Further information
HELSINGIN YLIOPISTO HELSINGFORS UNIVERSITET UNIVERSITY OF HELSINKI Tiedekunta/Osasto Fakultet/Sektion Faculty
Laitos Institution Department
Maatalous-metsätieteellinen tiedekunta
Taloustieteen laitos
Tekijä Författare Author
Kosenius, Anna-Kaisa Työn nimi Arbetets titel Title
Haitallisten leväkukintojen vähentämisestä koituvan hyödyn arviointi ehdollisen arvottamisen menetelmällä Oppiaine Läroämne Subject
Ympäristöekonomia Työn laji Arbetets art Level
Aika Datum Month and year
Sivumäärä Sidoantal Number of pages
Pro gradu -tutkielma
Toukokuu 2004
65 + liitteet
Tiivistelmä Referat Abstract
Tutkimuksen tarkoitus on arvioida haitallisten leväkukintojen (harmful algal blooms, HABs) sosioekonomisia vaikutukselle matkailusektorille Suomenlahden rannikolla Hangossa. Pro gradu –työ on osa ECOHARM-projektia, jossa selvitettiin haitallisten leväkukintojen vaikutusta matkailulle Euroopan Unionin alueella. Leväkukintojen haittojen selvittämisessä sovellettiin ns. ehdollisen arvottamisen menetelmää (contingent valuation, CV). Matkailijoilta kysyttiin heidän maksuhalukkuuttaan leväkukintojen määrän vähenemisestä. Maksuhalukkuusvastaukset analysoitiin probitmallilla ja saadun keskimääräisen maksuhalukkuuden avulla laskettiin leväkukintojen vähenemisestä koituvan kokonaishyödyn määrä matkailusektorille Hangossa. Tulosten mukaan matkailijoiden maksuhalukkuuteen vaikuttaa negatiivisesti heille esitetty rahasumma (bid) ja positiivisesti henkilön tulot, se mitä tärkeämmäksi vastaaja koki leväongelman rannikolla, lomailuun käytetty rahasumma ja se mitä tyytyväisempi vastaaja oli meriveden laatuun. Maksuhalukkuuden lisäksi tutkimus paljasti tietoja matkailusta Hangossa ja matkailijoiden leväkokemuksista. Hieman yli puolella (55,6 %) matkailijoista oli henkilökohtaista kokemusta joko myrkyllisistä tai ei-myrkyllisistä leväkukinnoista, ja 40,8 % vastaajista piti leväkukintoja pahimpana rannikon ongelmana Hangossa.
Avainsanat Nyckelord Keywords
leväkukinta, vedenlaatu, ehdollinen arvottaminen, matkailu Säilytyspaikka Förvaringsställe Where deposited
Maatalous-metsätieteellinen tiedekunta, Taloustieteen laitos Muita tietoja Övriga uppgifter Further information
PREFACE
This Master’s Thesis was started during my stay in Environmental Economics and Natural Resources Group at Wageningen University as an exchange student during the study year 2002-2003 and as a student assistant during the summer 2003 in the EU funded project ECOHARM. The work was completed in the Department of Economics and Management at University of Helsinki during the winter 2003-2004. Thanks to the ‘colleagues’ in both places for enjoyable working atmosphere.
My special thanks belong to Sara Scatasta, Markku Ollikainen, Eija Pouta, Ekko van Ierland, Marko Lindroos and Chiara Lombardini-Riipinen for supervision, advice, comments, help and support during the thesis process.
Helsinki, May 10, 2004
Anna-Kaisa Kosenius
Table of Contents 1. Introduction ............................................................................................................ 1 1.1 Background and objective of study.................................................................... 1 1.2 Contingent valuation methodology.................................................................... 3 2. Literature review.................................................................................................... 7 3. Theoretical background....................................................................................... 11 3.1 Utility maximization ........................................................................................ 11 3.2 Welfare change measurement .......................................................................... 16 4. Statistical model ................................................................................................... 19 4.1 Random utility model....................................................................................... 20 4.2 Modeling decisions .......................................................................................... 22 4.3 Willingness to pay estimation .......................................................................... 23 5. Data ....................................................................................................................... 27 5.1 Questionnaire ................................................................................................... 27 5.2 Descriptive statistics......................................................................................... 30 6. Results ................................................................................................................... 35 6.1 Motivations for willingness and reluctance to pay .......................................... 35 6.2 Parametric willingness to pay estimation......................................................... 39 6.3 Discussion ........................................................................................................ 44 6.4 Comparison of results ...................................................................................... 48 6.4.1 Previous studies......................................................................................... 48 6.4.2 Case studies of ECOHARM project ......................................................... 50 7. Methodology discussion and conclusions........................................................... 54 7.1 Critical view ..................................................................................................... 54 7.2 Counter-arguments........................................................................................... 56 7.3 Conclusions ...................................................................................................... 58 References ................................................................................................................. 61 Appendix 1. Annotated questionnaire.................................................................... 66 Appendix 2. Parametric model ............................................................................... 81 Appendix 3. Interviews in Hanko in July 2003 ..................................................... 87 Appendix 4. Occurrence of harmful algal blooms in Hanko in July 2003.......... 90
1
1. Introduction 1.1 Background and objective of study The occurrence of harmful algal blooms (HABs) is caused by the large amount of nutrients in seawater. The term ‘harmful’ refers here to the scale of occurrence of algal blooms when a natural algal phenomenon turns to be harmful for marine ecosystem or human beings in one way or another. The biological impacts of HABs on marine ecosystems have been studied for rather long. The insufficient knowledge of the socio-economic impacts of HABs restricts the estimation of cost-effective actions to monitor, manage and mitigate negative effects of HABs. From a socioeconomic perspective, HABs may have an impact on commercial fisheries, set constraints on recreation and tourism and cause risks for human health. Further, harmful algal blooms withdraw resources from the economy due to monitoring and management costs.
Harmful algal blooms are classified in four different types: seafood toxic blooms (STB), fish killing blooms (FK), high biomass non-toxic blooms (HBNT) and high biomass toxic blooms (HBT). There are two main types of HABs in the Gulf of Finland: blue green algae blooms and brown algae blooms. Blue green algae may be classified as HBNT as well as HBT, while brown algae may be classified as HBNT and FK. (HAEDAT 1998)
The nature of the occurrence of harmful algal blooms differs around Europe. In the North Sea, the HAB problem is caused by the invasion of exotic species. (Nunes and van den Bergh 2002, 1) In the Baltic Sea, the algal problem is due to anthropogenic nutrient load into the sea.1 Besides the external load, there is a considerable internal loading triggered by poor oxygen conditions. (Pitkänen et al 2001, 195) However, the complicated link between the eutrophication of marine and coastal waters and the formation of harmful algal blooms is still not completely uncovered. 1
The nutrient load of the Gulf of Finland is 2 to 3 times the average of the Baltic Sea relative to the surface area of the Gulf. At the end of the 1990s, the external nutrient load to the Gulf was in total 120 000 tons of nitrogen and 7000 tons of phosphorus per year. (Pitkänen et al 2001, 195)
1
The European Union project ECOHARM studied the socio-economic impact of HABs on recreation and tourism applying the contingent valuation method (CV or CVM). The case studies were conducted in Finland, France, Ireland and Italy. The CV questionnaire was designed at Wageningen University in the Netherlands in spring 2003, the interviews were conducted in Hanko and the other survey locations in summer 2003 and the Finnish data were analyzed and reported during winter 2003-2004 at University of Helsinki. The city of Hanko was chosen as a representative case from Finland due to the HAB occurrence and the dependence on coastal tourism. Moreover, it represents the small European coastal town.
Hanko is located on the southern coast of Finland in the Hanko Peninsula and the city faces the Gulf of Finland from south, west and north. Hanko has 10 000 inhabitants and 130 kilometers of coastline, of which 30 kilometers are beaches. For the economy of Hanko, the tourism industry is of great importance. In the summer season from June until August, Hanko is one of the premier seaside resorts of Finland. The largest guest harbor in Finland, the Eastern Harbor, is located in Hanko. The economy of Hanko in the year 2000 was based on services, trade and transport (58,9%), industry and building (39,4%) and agriculture and forestry (0,6%). The proportion of the services, trade and transport sector has been growing slightly during the period of 1998-2000 from 53,7% to 58,9%. (Hanko 2003)
Most of the workers in the tourism industry are seasonal workers mainly employed in the summer season. In the Eastern Harbor, there are about 7200 boats staying for night during the summer season, in addition to the popular Hanko Regatta weekend, when the number of boats is about 650 (Personal communication).2 During the year 2002, there were 47 466 nights spent in all accommodation establishments in Hanko. (Statistics Finland 2003, 52)
The objective of the study is to quantify the impact of harmful algal blooms on tourism sector in Hanko. The benefit from the algal bloom reduction is estimated using the contingent valuation method that elicitates people’s willingness to pay for 2
Personal communication. Harbor Master Thomas Levin. July 9, 2003. Eastern Harbor, Hanko.
2
the reduction. A parametric model is developed to reveal the factors behind the willingness to pay, and based on this model the aggregate benefit for the tourism sector from the algal bloom reduction is calculated. Besides recreation and tourism effects, the human health aspect is taken into consideration with the questions concerning the potential risk of getting shellfish poisoning from eating contaminated mussels.
This study is the first one to elicit people’s willingness to pay for the reduction in the harmful effects of HABs in Finland. The approach differs from the scientific studies carried out in other European countries since they estimated the benefits from nutrient reduction. The difference between ‘reduction in nutrients’ and ‘reduction in harmful effects of HABs’ is that the occurrence of HABs indicates eutrophication that is caused by nutrients in seawater.
1.2 Contingent valuation methodology In our study, the improvement of seawater quality is evaluated as an example of an environmental improvement. The existence of harmful algal blooms (HABs) in the seawater is regarded as a public bad and the reduction in HABs as a public good. The typical characteristics of a public good are nonrivalry and nonexcludability. Nonrivalry means that a person’s benefit from seawater quality improvement does not reduce other people’s benefit from the improvement. Nonexcludability means that no one's access to benefit from seawater quality improvement can be restricted regardless of whether they choose to pay for it or not. (See Bateman et al 2002, 18)
Since there is no market for the public good, we create an artificial one by contingent valuation method (CV or CVM) in order to simulate the behavior of the individual in the market. We describe the environmental change as a commodity that the respondent can buy in the market. The environmental policy needed to ensure the environmental change is described as well, and the cost of the policy implementation for the respondent is presented as the bid that is proposed to him / her. The respondent either accepts or rejects the bid, and his / her willingness to pay (WTP) can be derived based on these answers.
3
CVM is a stated preference (SP) technique and the idea is to ask people to state their valuations because we lack relevant information on the economic value of the public good. When some relevant information is available, the alternative for the SP technique is the revealed preference (RP) technique. It reveals people’s valuations for the public good from actual behavior, for example, prices paid for houses (hedonic pricing, HPM) or expenditures used in recreation (travel cost method, TCM). (Bateman et al 2002, 21)
The CVM is a widely used method to determine the value of environmental changes. The advantage of CVM over revealed preference methods is that CVM allows the evaluation of the environmental policy already in the planning step. The results of the CV study can be used in cost benefit analysis (CBA) when determining whether the project is worth implementing. (Bateman et al, 2002) Another advantage is that the CVM is the only method able to estimate the nonuse values. However, a critical view concerning the nonuse value estimation and the use of CV results in CBA is presented in section 7.1.
Our CV survey estimates the total economic value of damages caused by harmful algal blooms. Total economic value consists of use values (direct and indirect use values and option value) and non-use values (existence values and bequest values). The use values to be evaluated by CVM are for example the effects on human health, e.g. skin allergies and gastrointestinal disorders (direct use value) and the effects on marine ecosystem, e.g. the loss of local marine living resources diversity (indirect use value). The examples of damages in the non-use value category are the risk of the loss of the legacy of marine living resources for future generations (bequest value) and the risk of the loss of existence benefits (existence value). (Nunes & van den Bergh 2002, 3-4)
4
In contingent valuation studies, the value of the public good is measured through welfare change of the individual. There are four equivalent welfare measures: -
compensating variation and equivalent variation (when we deal with price changes) and
-
compensating surplus and equivalent surplus (when we deal with change in environmental quality). (Markandya et al 2002, 297)
Measures of Welfare Change due to
Price Change
Price Decrease
Max WTP CV
Min WTA EV
Quantity / Quality Change
Price Increase
Min WTA CV
Improvement
Max WTP EV
Max WTP CSU
Min WTA ESU
Degradation
Min WTA CSU
Max WTP ESU
Figure 1-1. The choice of the proper welfare measure. WTP = willingness to pay, WTA = willingness to accept compensation, CV = compensating variation, EV = equivalent variation, CSU = compensating surplus, ESU = equivalent surplus. (Markandya et al 2002, 298)
As we see from figure 1-1, the choice of the proper welfare measure depends on: 1) the nature of the change we value (the price / quality / quantity change), 2) the direction of change, and 3) the concept used in the study (willingness to pay or willingness to accept). The ‘path’ used in this study is marked by black arrows and yellow boxes.
5
We measure the welfare change due to seawater quality improvement and use the willingness to pay (WTP) concept instead of the willingness to accept (WTA) concept. The choice between WTP and WTA is the question of property rights. We assume that people do not have a right for the environmental improvement and they have to pay to ensure it. Thus, the welfare measure we use is compensating surplus (CSU).
The welfare theory behind the compensating surplus is explained in section 3.2 after the presentation of the individual’s utility maximization process in section 3.1. The previous CV studies in Europe concerning seawater quality improvements are reviewed in chapter 2. The statistical model used in the analysis is presented in chapter 4. The data description and the results are in chapters 5 and 6. The critical discussion and the conclusions are provided in chapter 7.
6
2. Literature review In Europe, impacts of seawater quality improvements have been estimated from the viewpoints of introduction of non-indigenous species (Nunes & van den Bergh, 2002) and nutrient reduction (Markowska & Zylicz, 1996; Gren, Söderqvist & Wulff, 1997; Söderqvist & Scharin, 2000). The study of Nunes and van den Bergh (2002) was the first one in Europe that measured the tourism-related impact of a reduction in HABs, while the other studies are not especially related to tourism.
Nunes and van den Bergh (2002) conducted a joint travel cost (TC) and contingent valuation (CV) study in Zandvoort beach by the North Sea in the Netherlands. The main cause of the occurrence of micro-algae along the Dutch coast is the ballast waters from the ships in Rotterdam harbor. People were asked about their willingness to contribute to the planned algae monitoring program, while the ship companies would finance a treatment plant for ballast water disposal and partially finance the algae monitoring program as well. Due to the implementation of the program, the risk of a biological pollution incident along the North-Holland coast and beaches would decrease with 90 %.
Respondents were asked whether they would vote for or against the introduction of a biological pollution annual tax on seawater that would be collected for a period of two years. The estimates for mean WTP and median WTP range from 58.2 Euro to 58.8 Euro and from 40.1 Euro to 47.4 Euro per person per year for two years, respectively, dependent on the model specification. (Nunes & van den Bergh 2002, 23)
In the Baltic Sea area, the studies estimating people's WTP for nutrient reductions have been conducted at both local (Söderqvist & Scharin, 2000) and Baltic Sea wide level (Markowska & Zylicz, 1996; Gren et al, 1997). In Finland, Kiirikki et al (2003) studied the effect of the nutrient reduction in the Gulf of Finland, but the study does not contain any quantitative information on the harms caused by eutrophication.
7
In the study of Gren et al (1997), the Baltic Sea wide overview of benefits from nutrient reduction was presented based on studies of Markowska and Zylicz (1996) conducted in Poland and Gren et al (1997) conducted in Sweden. The results of the Polish study were regarded as representative for the transition economies around the Baltic Sea (Estonia, Latvia, Lithuania, Poland and Russia), while the Swedish results were regarded as representative for the market economies around the Baltic Sea (Finland, Germany and Sweden). The estimated benefit from nutrient reduction was then compared to the estimated costs in order to find a cost-effective way to obtain the environmental target.
The scenario in the Swedish study consisted of a large-scale international action plan where cost-effective actions would be financed by an extra environmental tax for households. The 20-year plan was to reduce the eutrophication level of the Baltic Sea to the level it sustains. The scenario of the Polish study was almost identical. (Gren et al 1997, 136-139)
The mean WTP for nutrient reductions in Sweden was asked in two magnitudes. The total effect was estimated to be about 110 Euro (SEK 1000)3 and the recreational effect was about 88 Euro (SEK 800)2 per person per year for twenty years. The rather high amounts can be explained by high awareness of eutrophication as a very serious problem. More than 90 % of the respondents had heard about eutrophication, 85 % knew the effect of eutrophication on algal blooms, and 30 % of the respondents had personal experience on effects of eutrophication. The estimated mean WTP in the Polish study was about 33 Euro (SEK 300) per person per year. (Gren et al 1997, 137)
Söderqvist and Scharin (2000) estimated the regional WTP for reduced eutrophication in the Stockholm archipelago. The respondents were asked to suppose an abatement program according to which the farmers and the sewage treatment plants in three counties close to Stockholm archipelago have to contribute to measures against the nutrient emissions, which would result in increased prices of agricultural products and tap water. The benefit from the policy would be a one3
Exchange rate used 1 SEK = 0.10953 Euro. September 5, 2003.
8
meter increase in the sight depth in ten years.
The sight depth in the inner
archipelago at present is about 1 meter in summertime. The mean WTP per adult resident in the counties of Stockholm and Uppsala was estimated to be from 48 to 79 Euro (SEK 436-725) per year for ten years. (Söderqvist & Scharin 2000; 11,18)
Kiirikki et al (2003) studied the potential effects of the reduction of anthropogenic nutrient load in the Gulf of Finland. They focused on clarifying the state of the St. Petersburg water sector and to quantifying the cost and the load decreases due to the planned water protection measures. In addition to these, the study revealed information on recreational use of the Gulf of Finland and the nuisance caused by eutrophication, and people’s opinions about where to target water protection measures.
The survey was conducted in the county of Eastern Uusimaa and the questionnaire included questions concerning the attitudes on where the nutrient reduction should take place and who should be responsible for financing. The reduction of nutrient load from St. Petersburg and other parts of Russia were ranked as the most important sources followed by Finnish industry, Finnish coastal towns and Finnish agriculture. About two thirds (70 %) of respondents stated that the society and polluters should pay the costs of reduction. Almost a half (48 %) of respondents supported the idea of investing Finnish money in the neighboring countries Russia and Estonia, while 32 % of respondents were against it. (Kiirikki et al 2003, 36-39)
Concerning the nuisance caused by eutrophication, the respondents were asked to state whether they found the effects of eutrophication very harmful, harmful or not harmful at all. The fouling of beaches and turbidity of water were most often stated as harmful or very harmful, while the toxicity of algae was most often experienced very harmful. The fouling of fishing gear or boats was less often stated to be harmful. Blue green algae blooms have had rather large media-coverage, and a limited number of people had personally experienced algal blooms in open sea. In general, people in Finland seemed to be well aware of the eutrophication problem. These qualitative assessments can be used as guidance in the discussion about benefits from nutrient reduction. (Kiirikki et al 2003, 37)
9
The numerical results of our study can be compared to Nunes and van den Bergh (2002), Gren et al (1997) and Söderqvist and Scharin (2000). Information on people’s recreational behavior and experience of nuisance caused by HABs can be compared to the study of Kiirikki et al (2003).
10
3. Theoretical background
This chapter presents theoretical background of contingent valuation method (CV or CVM). Theoretical background lies on the consumer theory presenting how a rational individual maximizes her utility in the market. In chapter 4, economic theory is linked to the statistical analysis by the random utility model (RUM).
3.1 Utility maximization Sugden (1999, 133) and Hanemann (1999, 43-51) present the theory of public goods and the utility maximization process of an individual as follows. In the simplest form of the consumer theory, there is a market of two goods. One is a composite private commodity and another a public environmental good. Each individual chooses the amount of her consumption of the goods according to her preferences such that her utility is maximized. The choice is restricted by a budget constraint, so that she has a certain amount of money available in the market.
Individuals’ preferences are represented by a utility function
[1]
U i = U i ( xi , E )
i = 1,..., n ,
where i refers to an individual, n to the number of individuals in the market, Ui to utility of the individual i, E to the quantity of the public good available for every individual, and xi to the quantity of the private composite good of the individual i. As we can see from equation [1], there is no subscript i concerning the environmental good E. The amount of the environmental good E cannot be divided between the individuals because the same amount supplies benefits for all individuals (nonexcludability). This characteristic makes the individual i to take the quantity of E as given in her utility maximization process. Moreover, she takes the price of the
11
private good denoted by p as given and maximizes her utility subject to the budget constraint that sets the amount of money used in the market less or equal to the disposable income y.
We study the utility maximization of a representative individual in two markets, the real market of conventional consumption good and the counterfactual market comprising also a ‘tradable’ public good. First, the utility maximization of the individual i in the real market situation (we omit the subscript i):
[2]
Max U ( x, E ) x
subject to px ≤ y .
From the first order conditions of this problem a demand function can be derived
[3]
x = d ( p, E , y ) ,
where the demand of the good x is a function of its own price p, the quantity of the environmental good E and the income y.
However, when attempting to evaluate the environmental improvement in order to be able to derive the corresponding demand function for the environmental good E, we have to imagine a counterfactual situation where the individual actually is able to choose the level of E given its price denoted by π. (Hanemann 1999, 50)
The representative individual maximizes her utility
[4]
Max U ( x, E ) x,E
subject to px + πE ≤ y
and the Lagrangian function is
[5]
Max L = U ( x, E ) + λ ( y − px − πE ) . x ,E
12
From the first order conditions
[6a]
∂L = U x − λp = 0 ∂x
[6b]
∂L = U E − λπ = 0 ∂E
[6c]
∂L = y − px − πE = 0 ∂λ
where U x and U E represent the corresponding partial derivatives, the following demand functions can be derived:
[7a]
x = dˆ x ( p, π , y )
and
[7b]
E = dˆ E ( p, π , y ) .
The demand is a function of prices and income. The hats refer to the counterfactual demand functions, and in the further analysis it is useful to distinguish between the real and counterfactual market.
From [6a] and [6b] follows an important feature in utility maximization:
[8]
Ux UE = =λ, π p
where λ interprets the marginal utility of income of the individual. According to the consumer theory, the maximum utility for the individual is attained when her budget is allocated such that the ratios of marginal utility of the good and the price of the good are equal across the goods. In the optimal point where the marginal utility of income equals across the goods the individual is indifferent in whether to choose an additional unit of one good or another.
13
According to the conventional neoclassical economic theory, the individual’s preferences over the composite private good x and the public good E are represented by strictly convex downward sloping indifference curves (see figure 3-1 below).
The slope of the indifference curve is found by solving the total differential
[9]
dU =
∂U ∂U dx = U E dE + U x dx = 0 . dE + ∂E ∂x
From [9] follows that
[10]
dx / dE = −(U E / U x ) ,
which implies that the slope of the indifference curve, dx / dE , is negative, and reflects the marginal rate of substitution (MRS), U E / U x . The MRS indicates the rate by which the individual i is willing to trade between the private good x and the environmental good E.
Rearranging [8] gives
[11]
Ux p = , UE π
where the right hand side is the slope of budget line and the left hand side is the MRS. The graphical illustration (figure 3-1) of this is familiar: the budget line BB is tangential to the indifference curve I0 at the optimal consumption point a.
14
x
B
a
x
I0
B 0
E
E
Figure 3-1. The indifference curve I0 of the individual i, the budget constraint BB and the optimal consumption point a.
With the demand functions [7a] and [7b] derived from the counterfactual utility maximization process, we can form the counterfactual market indirect utility function
vˆ( p, π , y ) that corresponds to the utility function U ( x, E ) in [1] as follows:
[12]
[
]
vˆ( p, π , y ) ≡ U dˆ x ( p, π , y ), dˆ E ( p, π , y ) .
The corresponding real market indirect utility function comes from [3]
[13]
v( p, E , y ) ≡ U [d ( p, E , y ), E ] .
15
3.2 Welfare change measurement Before proceeding to welfare change measurement, the link between the real and counterfactual market indirect utility functions is presented as
[14a]
v( p, E, y ) ≡ vˆ[ p,πˆ ( p, E , y ), y + πˆ ( p, E , y )E ] .
The real market indirect utility function v( p, E , y ) corresponds to the counterfactual market indirect utility function vˆ( p,πˆ , y + πˆE ) . In the counterfactual market, the individual chooses both the amount of environmental good E and the private good x, and in order to afford both, the disposable income y must be supplemented by the amount of money needed to buy the environmental good E. This amount of money is the price of the environmental good times the amount of environmental good, πˆE .
The price that induces the individual to buy the amount E of the environmental good is denoted by πˆ , and can be found by solving the demand function for the environmental good derived from utility maximization E = dˆ E ( p, E , y + πE ) for
π = πˆ ( p, E , y ) with given p, E and y.
Correspondingly, the nonsupplemented counterfactual market indirect utility function equals to the real market indirect utility function from which the supplemented part of income is subtracted and the utility is maximized with given E
[14b]
vˆ( p, π , y ) = max E v( p, E , y − πE ) .
(Hanemann 1999, 50)
We study the welfare change from the environmental improvement for the individual using the real market indirect utility function v( p, E , y ) . Suppose that the amount of the public good changes from E0 to E1, E1 > E0 . Consequently, the utility of the individual changes from U 0 ≡ v( p, E0 , y ) to U 1 ≡ v( p, E1 , y ) . In order to measure the size of the utility change, Mäler (1971, 1974) proposed the measure of welfare
16
change, compensating surplus (CSU), which measures the individual’s maximum willingness to pay (WTP) for environmental improvement: v( p, E1 , y − CSU ) = v( p, E0 , y ) .
[15]
x
a
x0
b
CSU
U1 = I1
c
x1
U 0 = I0
0
E1
E0
E
Figure 3-2. Compensating surplus (CSU). (Markandya et al 2002, 305)
In figure 3-2, the initial situation is the optimal point a at the utility level U0 that corresponds to the indifference curve I0 in figure 3-1. The environmental improvement from E0 to E1 would bring the individual to the higher utility level U1 and to the consumption point b. Compensating surplus (CSU) is defined with the utility level U0 as the reference4, and in order to leave the individual at the utility level U0, the change from E0 to E1 has to be compensated by reducing the money available for the consumption of x as shown in [14b]. As the result, the consumption of x decreases from the initial level x0 to x1 and the final consumption point is c. In figure 3-2, the quantity of the private good x is measured in monetary terms (it is a
4
Equivalent surplus (ESU) defines the welfare change using the utility level U1 as the reference.
17
numeraire), and CSU can be regarded as the WTP of the individual for the change from E0 to E1. (Markandya et al 2002, 304) Compensating surplus CSU is a theoretical measure of willingness to pay. When conducting a CV study, we attempt to elicit the willingness to pay of the individuals for a proposed environmental improvement. The hypothetical market described in the CV scenario reminds the counterfactual market presented above. In the counterfactual market, the individual’s demand for E can be derived, and we can find the price by which the individual is willing to buy a particular amount of the environmental good (the improvement described in the CV scenario). In the counterfactual market, she will buy this if her income is supplemented by πˆE to compensate the loss in the consumption of the private good x (Bateman et al 2002, 25).
In the CV study, the respondents are asked whether they are willing to pay a particular sum for described environmental improvement. The respondents’ answers to the WTP question are analyzed by statistical method. The economic theory and the statistical framework meet in the random utility model (RUM) presented in the next chapter.
18
4. Statistical model
Statistically, the responses to the binary choice willingness to pay question are discrete dependent variables. We propose a bid that reflects the estimated cost from the environmental policy to the respondent who either accepts or rejects the bid. The answer depends on whether s/he wants to stay in the status quo or to move to the improved environmental state and pay a certain sum for it. The common structure of the binary response model is the following. The probability that the respondent accepts (rejects) the bid is expressed as the function H ( A, z )
[16]
Pr[YES ] = H ( A, z ) ≡ H ( A)
Pr[NO ] = 1 − H ( A, z ) ≡ 1 − H ( A)
where A refers to the proposed bid and z represents the parameters estimated from the data, for example, the sociodemographic characteristics of the respondent or the attitudes towards the program in question. (Hanemann & Kanninen 1999, 306)
In contingent valuation literature, a statistical model has been formulated through utility differences (for example, Hanemann 1984) or bid functions (for example, Cameron 1988). Generally, a bid function is an approximation to a functional form derived from a utility difference model. Thus, the bid function results from the utility difference problem solved by the respondent. The advantage of the bid function approach is that the utility difference approach starts by specifying the exact form of utility function and this may lead to complex specifications of the bid function. Another advantage of bid function approach is that the marginal effects are easier to calculate. The bid function approach is more simple and practical, but it is less explicitly connected to neoclassical utility maximization theory. (Bateman et al 2002, 184-191)
We follow the utility difference approach but choose the simplest functional form, the linear form, in order to avoid complexity. In any case, the functional form specified by the researcher is nothing but an approximation of the true indirect utility 19
function of the individual (Bateman et al 2002, 185). The different functional forms are discussed in section 4.2. In the next section, the WTP analysis is put into the random utility framework.
4.1 Random utility model The random utility model (RUM) was originally established by McFadden in the 1970’s. The idea is that the awareness of the respondent differs from the awareness of the researcher. During utility maximization, the respondent is fully aware of all possible factors affecting the choice and makes a choice based on the factors. A statistical model cannot reveal all the factors behind the utility maximization process. The ones that remain unobservable for the researcher are captured by the stochastic component ε that makes the statistical model more consistent with economic theory. (Hanemann & Kanninen 1999, 307)
As seen in section 3.2, the welfare measure compensating surplus (CSU) represents the maximum WTP of the respondent for the environmental improvement. The respondent can reach the same utility level U0 either with the initial environmental state E0 or with the improved environmental state E1 when the payment required to secure the change is reduced from his / her income. We add the stochastic term ε in equation [15] and omit the price of the private good p
[17]
v(E1 , y − CSU , ε 1 ) = v(E0 , y, ε 0 ) ,
where y is the disposable income and ε0 and ε1 represent the added stochastic term. As a result of the rational utility maximization, the respondent accepts the bid A if s/he is better of with the improved environmental state E1 and the payment required to obtain it than with the initial environmental state E0. This is captured in the utility difference function
[18]
v(E1 , y − A, ε 1 ) ≥ v(E0 , y, ε 0 ) .
20
As the researcher lacks some information about the preferences of the respondent, only probability statements can be made about whether the respondent accepts the bid or not. The probability of accepting the bid A is then
[19]
Pr[YES ] = Pr[v(E1 , y − A, ε 1 ) ≥ v(E0 , y, ε 0 )] .
From [17] we know that the welfare change measure compensating surplus CSU ( E0 , E1 , y, ε )
reflects the maximum WTP of the respondent for the
environmental improvement from E0 and E1 and from [18] we know that the respondent accepts the bid A if it is less than or equal to his / her WTP and rejects it otherwise. Based on these, the equivalent way to express [19] is
[20a]
Pr[YES ] = Pr[CSU (E0 , E1 , y, ε ) ≥ A] .
In the RUM model, CSU ( E0 , E1 , y, ε ) is treated as a random variable due to preference uncertainty captured in the stochastic term ε, and the researcher must assume the distribution of CSU ( E0 , E1 , y, ε ) . Then, the response probability formula [20a] can be expressed in the general form
[20b]
where
Pr[YES ] = 1 − GC ( A) ,
GC (A)
is the assumed cumulative distribution function (cdf) of
CSU ( E0 , E1 , y, ε ) . The equation [20b] corresponds to the general structure of the binary response model in [16] as Pr[YES ] = H ( A, z ) ≡ H ( A) = 1 − GC ( A) . (Hanemann & Kanninen 1999, 308; Haab & McConnell 2002, 25-26)
21
4.2 Modeling decisions For the analysis we have to make two modeling decisions: the functional form of indirect utility function and the distribution of the error term. (Haab & McConnell 2002, 25)
The generalized functional form that nests a number of models in the literature is the Box-Cox transformation of income5
yλ −1 + ε E , vE ( y, z ) + ε E = α E z + β λ
[21]
where E refers to the environmental state, εE is the random term, αE is an mdimensional vector of parameters, z is an m-dimensional vector of variables, and
α E z = ∑ k =1α Ek zk . The Box-Cox utility function captures the effects of income on m
indirect utility flexibly through the transformation parameter λ . If λ = 2 , the utility depends on the square of income, if λ = 1 , the utility depends linearly on income and correspondingly, if λ = −1 , the utility function is linear in the inverse of income, and if λ = 0 , the utility function is log-linear in income. (Haab & McConnell 2002, 40) We assume the special case of the Box-Cox utility function with λ = 1 , when the utility function is linear in the deterministic part, i.e. parameters z and income y
[22]
vE = α E z + β E y + ε E ,
E = 0 or 1 .
The linear model is a good approximation to any utility specification, but the drawback is that it assumes the constant marginal utility of income across the scenarios. (Haab & McConnell 2002, 35)
The second modeling decision is the distribution of the error term. The error term in the linear single-bounded discrete response probability model can have standard
5
The notation of the indirect utility function changes here such that v(EE,y,εE) = vE(y,z)+ εE.
22
normal, standard logistic, standard lognormal, standard log-logistic or standard Weibull distribution. The normally distributed error term leads to the probit model and the logistically distributed error term to the logit model. (Hanemann & Kanninen 1999, 415)
We assume the standard normal distribution with mean zero and unknown variance
σ 2 , thus we use the probit model. In the case of the linear utility model, the probit model and the logit model lead to consistent willingness to pay estimates. The problem that likely arises with the combination of normally or logistically distributed error terms and the linear functional form is the violation for the constraints set to WTP. In most cases, WTP is bounded from below by zero and from above by income, but normally and logistically distributed error terms allow unbounded WTP measures. (Haab & McConnell 2002; 39, 88-89)
Paying attention on the distribution of the error term is important, because in the random utility model (RUM) the stochastic component is an essential part that interacts with the deterministic part and affects the welfare evaluation. In order to make the shape of the WTP distribution more flexible, one can add parameters to the probability model, use mixture models including a spike for example at zero or at the right tail of the WTP distribution, use heteroscedastic distributions or the models of preference uncertainty or thick indifference curves. One way to get around the difficulty of the WTP distribution decision is to estimate willingness to pay nonparametrically. (Hanemann & Kanninen 1999, 407) Additional ways to improve the accuracy of WTP estimates are discussed in section 7.2.
4.3 Willingness to pay estimation The expression for WTP is found from the difference of linear utility functions. We denote the status quo as E = 0 and the improved environmental state defined in the scenario as E = 1. Thus, the utilities of the representative individual in the status quo and after the environmental improvement are v0 and v1, respectively,
23
[23a]
v0 = α 0 z + β 0 y + ε 0
and
[23b]
v1 = α1 z + β1 ( y − A) + ε1 ,
where A is the bid proposed to the respondent.
Thus, the welfare change due to the environmental change is
v1 − v0 = α1 z + β1 ( y − A) + ε 1 − α 0 z − β 0 y − ε 0 [24]
= (α1 − α 0 )z + (β1 − β 0 ) y − β1 A + ε 1 − ε 0 = αz − β A + ε
where α ≡ α1 − α 0 and ε ≡ ε 1 − ε 0 . Moreover, we assume that the marginal utility of income is constant – ‘independent of income, constant across the states E0 and E1 and constant across observations’ (McConnell 1990, 21) – and so β 0 = β1 ≡ β . The differences between the error terms ε0 and ε1 cannot be identified, and we can treat them as a single random component ε ≡ ε1 − ε 0 . (Haab & McConnell 2002, 26; Hanemann & Kanninen 1999, 309)
The response probability formula for the respondent i becomes Pr[YES] = Pr (αz i - βAi + ε i > 0 ) [25]
= Pr (− (αz i - βAi ) < ε i )
= 1 − Pr (− (αz i - βAi ) > ε i )
= Pr (ε i < αz i - βAi ) .
We suppose the error term to have a standard normal distribution with mean zero and unknown variance, but we convert the error term ε to a standard normal variable with mean zero and variance one N(0,1). Denoting θ = ε / σ , [25] becomes
24
Pr (ε i < αz i - βAi ) [26]
αz β = Pr θ < i - Ai σ σ αz β = Φ i - Ai , σ σ
where Φ(⋅) is the cumulative standard normal distribution of ε.
The parameters are estimated by the maximization of the likelihood function
αz βA L(α , β y, z , A) = ∏ Φ i - i σ i =1 σ T
[27]
Ri
αz βA 1 − Φ i - i σ σ
1− Ri
,
where T is the sample size and Ri = 1 if the respondent i accepts the bid. The maximum likelihood estimates are calculated using the log of the likelihood function
[28] T αz βA ln L(α , β y, z , A) = ∑ Ri ln Φ i - i σ i =1 σ
αz βA + (1 − Ri ) ln 1 − Φ i - i σ σ
.
(Haab & McConnell 2002, 27-29)
The aim of the parameter estimation is to calculate WTP and to find the effects of covariates on WTP. In WTP calculation, the researcher faces two sources of variation: uncertainties concerning parameters and preferences. Moreover, individual covariates may vary across the individuals. The parameter uncertainty can be seen as the ratio of the parameter estimates
αz i βAi and . After the parameter estimation σ σ
step, we take the parameters as given and move to the WTP calculation step. (Haab & McConnell 2002, 34-35)
25
From the linear utility model, WTP can be found from [24] by substituting A for WTP and solving the equation for WTP, which results in
[29]
WTP = αz / β + ε / β =
αz + ε . β
The expectation of WTP for the respondent i with respect to preference uncertainty is then
[30]
Eε (WTP α , β , z i ) =
αz i β
for both the mean and the median WTP, because they are equal when we have the linear utility specification with the symmetric mean zero error. (Haab & McConnell 2002, 34-35)
When expanding the results of the sample to the whole population, we use the mean WTP and the mean vector for variables
[31]
Eε (WTP α , β , z ) =
αz , β
where αz = α 0 + ∑k =1α k z k and m is the number of explanatory variables and zk the m
sample mean for variable k. (Haab & McConnell 2002, 33-35)
The results from the standard probit model are reported in chapter 6.
26
5. Data
The questionnaire (see Appendix 1) used in the survey was designed in order to elicit the socio-economic impacts of HABs on the selected case study locations using the willingness to pay concept. The questionnaire was designed at Wageningen University during March-May 2003 and pretested in Riccione, Italy in June 2003. The format chosen to reveal people’s willingness to pay was the one-and-one-half bounded (OOHB) question format. The data were collected in case study locations Hanko (Finland), Riccione (Italy), Hyères Le pradet Conquieranne (France) and Galway (Ireland) by personal interviews because that suits best with the OOHB question format.
5.1 Questionnaire The questionnaire contained of six sections. The first section included the screening questions in order to check whether the person contacted was included in the target population (an adult tourist in selected location). The second section revealed the recreational profile of the respondent, and in the third section the attitudes about coastal problems and seawater quality were asked. The fourth section was about travel expenditure and the fifth about willingness to pay for seawater quality improvements. In the last section the socio-demographic information was collected.
In the willingness to pay section the respondents were first asked whether they ever had experienced problems with HABs, i.e. high biomass non-toxic blooms (HBNT), high biomass toxic blooms (HBT) or seafood toxic blooms (STB). In the Finnish version of the questionnaire, the current algal situation in the European Union was briefly described by giving some examples about the potential harmful effects of HABs in the Baltic Sea, the Atlantic Ocean and the Mediterranean.
27
In all the areas, HBNTs may cause foam and mucilage on the coast, lime fishing nets and spoil the taste of fish. The HBTs in the Baltic Sea may cause skin and eye irritation to swimmers and kill fish, sea birds, and domestic animals. The STBs in the Atlantic Ocean and the Mediterranean may cause the accumulation of potent toxins in bivalve mollusks such as mussels. Consumption of contaminated seafood may cause diarrhea, nausea, vomiting, abdominal cramps, paralysis or even death. The geographical differences in the effects of HBT blooms were not described in other versions of the questionnaire and the respondents were only informed about the effects of STBs.
The focus of the ECOHARM project was on mussel aquaculture and tourism. However, mussels do not belong to the Finnish eating habits to the same extent than in Italy, France and Ireland. In the questionnaire design, the difference between the countries was taken into account. In the Finnish version of the questionnaire, no questions concerning mussel-eating habits of the respondent were asked before the willingness to pay question. This was made in order to avoid irritation. However, after the WTP question there was a question about potential mussel-eating in the future and then in order to reduce the confusion of the respondents, they were asked whether they usually eat mussels.
The willingness to pay question started by describing the following scenario:
The European Union is thinking of reducing the amount of pollutants in the seawater improving the treatment of wastewaters flowing into rivers as well as the sea. The European Union also wants to reduce the risk of intoxication from eating mussels increasing monitoring activities of toxic algal blooms by sampling water and mussels more often than it does right now. The objective of this policy is to achieve a 25% reduction in algal blooms by the beginning of the next summer season.
To finance this policy, the European Union is thinking of introducing a mandatory contribution specific for coastal zone management that would be paid by everybody. This mandatory contribution would be paid once a year by each resident of the European Union and by tourists who do not reside in the European Union but visit European Union coastal zones. This mandatory contribution would be paid also by
28
firms in the industrial, agricultural and fisheries sector based on the type and size of business.
For European Union residents and tourists who do not reside in the European Union but visit European Union coastal zones, this mandatory contribution would be in the range of ___ and ___ Euro per person per year. If you knew that this money would be used to achieve a 25% reduction in algal blooms by the beginning of the next summer season, and a 50% immediate reduction in the risk of intoxication from eating mussels, would you be willing to pay ___ Euro per person per year?
In order to elicit people’s willingness to pay, the one-and-one-half-bounded (OOHB) dichotomous choice format was applied. Cooper et al (2002) introduced the OOHB format as an alternative for single-bounded (SB) format presented in Bishop and Heberlein (1979) and double-bounded (DB) format presented in Hanemann, Loomis and Kanninen (1991) trying to capture the advantages of both. In the SB format, only one amount of money (bid, later Bi), is proposed to the respondent, and asked whether s/he is willing to pay that amount for a given improvement. In the DB format, the first bid (B0) is followed by the second bid (BU or BD) in consistency with the answer to the first bid (B0). In the case of a yesanswer, a higher bid (BU) is proposed, and in the case of a no-answer a lower bid (BD) is proposed. (Cooper et al 2002, 742)
In order to avoid the surprise effect caused by the second bid offer in the DB format, in the OOHB format, the respondent is first informed about two bids (B- and B+). This is done by telling that the costs of the policy described in the scenario will be somewhere between the range of two bids, and then either the lower bid (B-) or the higher bid (B+) is proposed for the respondent. If it makes sense, i.e. a yes-answer to B- or a no-answer to B+, the alternative bid is asked as well. (Cooper et al 2002, 742)
The DB format has the advantage of higher efficiency in welfare benefit estimates than the SB format, but criticism has arisen against the DB format that some of the responses to the second bid may be inconsistent with the responses to the first bid.
29
The OOHB format tries to maintain much of the efficiency gains of the DB approach and to reduce the potential for response bias on the follow-up bid in DB format. (Cooper et al 2002, 742)
5.2 Descriptive statistics The data were collected by personal interviews during July 1-29, 2003 in Hanko. We contacted 325 persons and received 212 interviews. The participation rate was 65.2 %, and reasons for non-participation were for example ‘don’t speak English’, ‘don’t have opinion’, ‘don’t want to participate’ and ‘too busy with children’. Sixteen interviews had to be discarded due to respondent’s misunderstanding (n = 7), nontrueness of the answers (n = 5) or missing answer to the WTP question (n = 4). So, in the analysis we will have 196 observations and the response rate is 60.3 %.
The interviews took place in the coastline of the city of Hanko. The main places were Silversand Camping, Eastern Harbor, and three beaches (Regatta, Plagen and Bellevue). Three interviews were completed outside the city area. In Silversand Camping the respondents were interviewed on the beach, in the café and in the camping area, and in the Eastern Harbor in cafés, terraces and on the harbor bridge. On Plagen, Bellevue and Regatta beaches, the interviews took place on the beach due to the lack of beach houses and cafés (except Café Plagen).
Table 5-1. The percentual description of the data (N = 196).
Domestic
89.3 % 10.7 %
Foreign
Regular visitor
55.4 % 44.6 %
Not regular visitor
Male
50.0 % 50.0 %
Female
Lives on coast
55.6 % 44.4 %
Lives in inland
Experience on HABs
55.6 % 44.4 %
No experience on HABs
WTP > 0
67.4 % 32.6 %
WTP = 0
30
Table 5-2. The average description of the data (N = 196).
Length of stay (days)
6.58
Vacation expenditure in Hanko (Euro per capita per day)
33.36
Vacation expenditure in Hanko (Euro per capita)
145.69
Age (years)
42.91
Education (years)
13.43
Monthly net income (Euro per capita)
1276.87
From tables 5-1 and 5-2 the data can be described as follows.
Tourist profile
Large majority, 175 (89.3 %) respondents were domestic visitors and 21 (10.7 %) were foreigners. According to Statistics Finland (2003a, 52), the percentage of foreign tourists in Hanko was 9.4 % in 2002, so the survey sample corresponds well enough to the target population. Foreign tourists were from 12 countries: Sweden (n = 4), Germany and the Netherlands (n = 3), the United States and France (n = 2), and Austria, the Republic of South Africa, Norway, Denmark, Switzerland, Great Britain and Estonia (n = 1).
The average length of stay was 6.58 days. Slightly more than half of the respondents (55.4 %) defined themselves as regular visitors in Hanko. The respondents defined the term regular in different ways, for example, ‘I come here once each summer’ or ‘I come here every weekend and summer holidays’.
The average expenditure during the stay in Hanko was 33.36 Euro per person per day. The average vacation budget per capita was 145.69 Euro. This information was only collected from 158 respondents because it was difficult for some respondents to report the exact amount of expenditure.
31
The average age of the respondents was 42.91 years and 50.0 % of the sample was males. In Finnish population the percentage of males is 48.2 % (Statistics Finland, 2003b). The number of children in the households ranged from zero to five and was on average one child. The average length of education was 13.43 years. The monthly net income per capita was 1276.87 Euro.
Slightly more than half of the respondents (55.6 %) lived on the coastal cities of Finland (plus the Kauniainen city surrounded by the coastal Espoo city). The distance from Hanko to the place of respondent’s residence ranged from 36 to 9537 kilometers with the average of 409 kilometers. The large majority (87.6 %) came from the distance less than 400 kilometers and 60.2 % of the respondents came from the distance less than 150 kilometers.
None 15 %
Boat 18 %
Hotel 9%
Own property 9%
B&B 1%
Friend’s property 12 % Camping 31 %
Rental property 5%
Figure 5-3. Accommodation of respondents.
From figure 5-3 we see that less than one sixth (15 %) of the respondents were daytrippers. The rest owned a flat in the city or the free time house outside the city area (9 %), stayed in the camping area (31 %) or in his or her own boat (18 %). About one
32
fourth (27 %) were accommodated either in hotel, pension or with friends or relatives. The stay in pension can be either Bed & Breakfast or a room can be rented for a longer period.
According to Statistics Finland (2003a, 69), 61.4 % of the visitors in Hanko in 2002 were accommodated in room or cottage, 14.9 % in caravan or camper and 23.9 % in tent or other. These percentages, however, do not contain boaters, day-trippers or the visitors owning freetime house or flat.
Problem experience and opinions
The respondents living on the coast (55.6 %) were not exactly the same people who had experienced problems (55.6 %) with either high biomass non-toxic (HBNT) or high biomass toxic (HBT) algae or seafood toxic blooms (STB). The HBNTs had caused more problems, namely, 52.6 % of the respondents had experienced problems with HBNTs while the corresponding percentage for HBTs was 13.8 %. Only 85.2 % of the respondents were able to rank four coastal problems6 based on the importance. Out of these 156 respondents, 45.5 % ranked ‘algal blooms in the sea’ as the most important coastal problem followed by ‘sometimes noise or smell’ (31.4 %), ‘traffic and not enough parking spaces’ (16.0 %) and ‘not enough recreational activities’ (7.1 %). Moreover, 11 respondents were not able to rank all the problems but one or two. Taking these into analysis, 48.2 % of the respondents who were to some extent able to answer the question stated algal blooms as the most important problem. This was 40.8 % of the whole sample.
On average, 74.3 % of the respondents was either very satisfied or satisfied with the seawater in Hanko during the interview period. The proportions of ‘very satisfied’ people during the four interview weeks were 0.27, 0.19, 0.29 and 0.17, respectively. The development of algal blooms in the sea is reflected to the peak during the third week. During the first interview week (1-4 July) the temperature of seawater was
6
The Finnish versions of coastal problems differed slightly from the problems in other countries. The problems were chosen in co-operation with Virpi Mikkanen, Environmental Secretary, the City of Hanko. June 30, 2003.
33
approximately 15 Celsius degrees, and still too cold for algal blooms. During the second week (9-13 July) the blooms started to form in the open sea. During the third week (14-21 July) the threat of blue green algae blooms was already present, but it had not yet reached the beaches of Hanko except Silversand Camping where the warning about blue green algae was put on the beach in the beginning of the third week. The seawater quality worsened during the fourth interview week (22-29 July) due to warm weather.
34
6. Results The motives for willingness and reluctance to pay are discussed in section 6.1, and the parametric probit model is estimated and benefit estimates are calculated in section 6.2. The parametric model is discussed in section 6.3. Finally in section 6.4, the results of the study are compared to other studies.
6.1 Motivations for willingness and reluctance to pay Out of 196 respondents, 132 (67.4 %) stated positive WTP and 64 (32.6 %) stated a zero WTP. In order to discuss the motives for willingness and reluctance to pay, the respondents are divided into four classes in table 6-1 dependent on whether they 1) had experienced any problems with HABs and 2) stated positive or zero WTP.
Table 6-1. Willingness to pay and problem experience on HABs.
No problem experience
Problem experience
Total (%)
WTP = 0
15.3
17.3
32.6
WTP > 0
29.1
38.3
67.4
Total (%)
44.4
55.6
100
In table 6-1, the percents in the upper row show that 15.3 % of the respondents stated zero WTP without any problem experience on HABs and 17.3 % of the respondents stated zero WTP but had experienced problems with HABs. In the lower row, 29.1 % stated positive WTP without problem experience with HABs and the biggest share of the respondents, 38.3 % stated positive WTP and had experienced problems with HABs.
35
To be willing to pay (WTP > 0) makes sense if the respondent has had problems with HABs, as 38.3 % of the respondents. However, 29.1 % of the respondents was willing to contribute to the algal reduction even without any problem experience. There are some possible explanations for this behavior. They valued the seawater quality improvement in the European Union level based on the information given during the interview or they were willing to pay for the problem because of other people (altruism) or future generations or because they got nice feeling from charity (‘warm glow’ effect).
The zero bids (WTP = 0), are further divided into four classes in table 6-2 dependent on 1) the problem experience on HABs and 2) whether the answers are identified as true zero bidders or protesters. The respondents who reasoned their reluctance to pay as ‘the problem is not worth paying’ or ‘I can’t afford that’ are considered true zero bidders.
Table 6-2. Protest behavior and problem experience.
No problem experience
Problem experience
Total (%)
True zero
21.9
9.4
31.3
‘Protest’
25.0
43.7
68.7
Total (%)
46.9
53.1
100
From table 6-2 we see that 68.7 % of the zero WTPs were ‘protests’. It makes sense to be reluctant to pay if the respondent has not experienced problems with algal blooms (21.9 %) as well as if the respondent has experience problems with HABs but either does not consider the algal blooms an important problem enough to be contributed to or s/he cannot afford that (9.4 %). If the respondent has not experienced problems with HABs, s/he may protest for example for still being asked to pay an environmental tax (25.0 %). The respondent may protest also if s/he considers HABs as a problem but does not think it is his / her responsibility to pay for it (43.7 %).
36
Out of 64 zero WTPs, 20 were true zeros and 44 protesters. The term protester, however, is misleading because in addition to the protest against some feature in the questionnaire – for example the payment vehicle or other component in the scenario – these answers may reflect strategic behavior as well (Bateman et al 2002, 178). The percentage of protesters in the whole sample was 22.4 %. The protest motivations are divided in five groups in table 6-3.
Table 6-3. Protest motivations.
Motivation
Number of respondents 18
Polluters should pay
Proportion 40.9 %
No more taxes or payments for individuals
12
27.3 %
Reduction described in the scenario is impossible
6
13.6 %
Protest against EU or doubts about bureaucracy
4
9.1 %
Government or other’s responsibility
4
9.1 %
From table 6-3 we see that two fifths (40.9 %) of protesters stated ‘polluters should pay’ and 9.1 % of protesters ‘government or others should pay’ as the motivation for their zero WTP. Slightly more than one fourth (27.3 %) were against additional payments, 9.1 % protested against the EU or bureaucracy and 13.6 % doubted the reduction described in the scenario. The typical protester7 was a male who had experienced problems with toxic algae blooms but did not rank ‘algae blooms in the sea’ as the most important coastal problem in Hanko. It makes sense to protest against the collection of a common tax if the respondent is familiar with the problems caused by HABs but does not consider the occurrence of HABs as the most important problem to be solved, and thinks that it is not his duty to contribute.
7
Tendency to protest behavior was analyzed by comparing the means of the groups of protesters and non-protesters and using the t-test (continuous normally distributed variables) or by the chi-square test with cross-tabs (dummy variables).
37
Protest bids are included in the analysis because otherwise the amount of observations falls dangerously low due to small sample size and the coefficients are not significantly different from zero. One characteristic of protest behavior – not to consider algal blooms as the most important coastal problem – actually reflects more true zero WTP (‘the problem is not worth paying’). This indicates that the respondents got irritated when they were asked to contribute to something they do not feel being worth it. Protesting, however, does not mean that they do not put a positive value for seawater quality improvement.
The justification of including protests in the analysis lies in the assumption that the true WTP of the protesters is similar to the WTP of the respondents with comparable characteristics (Bateman et al 2002, 178). As the particular characteristics for the protest behavior were found, excluding the protests would have systematically biased the WTP analysis. By including protest answers we have more zero values in the analysis and thus, we get a conservative WTP estimate as recommended by the NOAA Panel on Contingent Valuation (NOAA 1993, 4608).
The location of Finland in the border of the European Union affected some people’s willingness to pay a EU wide environmental tax. The nutrient flow from Russia into the Gulf of Finland or from other non-EU countries8 into the Baltic Sea is a wellknown feature among the Finns (Kiirikki et al, 2003). In our questionnaire no information about the sources of nutrients were given, and it was not said that the payments collected from Finnish people would be used in the Gulf of Finland. Out of 196 respondents, 6.1 % reasoned their reluctance to pay with the contribution of Russia (especially the city of St. Petersburg) or other non-EU countries to seawater quality. Additionally, 6.6 % of 196 respondents mentioned Russia or St. Petersburg during the interview, and stated the positive WTP with the condition for their own contribution that ‘I pay only if those polluters do their own share’.
8
During the time of interviews in July 2003, Finland was the only EU member state of the three countries around the Gulf of Finland. In the beginning of May 2004, Estonia joined the EU together with Lettland, Lithuania and Poland.
38
6.2 Parametric willingness to pay estimation As mentioned in section 4.2, defining the parametric model is not necessary to get the estimation of WTP. The estimation can be done non-parametrically based on the proportions of the people who accepted the proposed bid. In non-parametric estimation, a monotonically decreasing survivor curve H(A) as a function of the offered bid A is defined, and the area under the survivor curve represents the mean and median WTP of the respondents. (See for example Hanemann & Kanninen 1999, 394-396)
The parametric model explores the factors behind the willingness to pay and their size and magnitude. Checking the signs of the covariates, for example income, tourist profile or environmental attitude, can test the validity of the CV answers. The parametric model helps with benefits transfer and, moreover, in the case the sample is not representative for the target population, the distortion can be corrected. The disadvantage of the parametric model is that it has to be correctly specified; otherwise the covariate effects and hypothesis tests are not valid. (Haab & McConnell 2002, 24)
In the model, the dependent dichotomous variable CHOICE reflects the probability of the respondent to state a positive WTP, i.e. to accept the offered bid.
39
Table 6-4. The results of the probit model (N = 139).9
Variable
Description and coding of variable
Coefficient
St.
P-value
Mean
Error Constant
1.2307***
0.6406
0.0547
Bid
First bid proposed to respondent
-0.0150*
0.0051
0.0029
23.9137
Bound
1 = first proposed bid is higher
0.4687***
0.2450
0.0557
0.51799
0.1383
0.1732
0.4247
1.77698
-0.3180**
0.1290
0.0137
1.84173
bound of estimated cost range 0 = lower bound Inccap
1 = monthly net income per capita 0 – 1000 Euro 2 = 1000 – 2000 Euro 3 = 2000 – 3000 Euro
Coastal
1 = algae blooms ranked as the most important coastal problem 2 = the 2nd most important problem 3 = the 3rd most important problem 4 = the least important problem
Expend
Vacation budget per capita
0.0012***
0.0007
0.0751
144.130
Satisf
1 = Very satisfied with water quality
-0.3107***
0.1665
0.0621
2.12230
2 = Satisfied 3 = Not very satisfied 4 = Not satisfied at all Log likelihood (UR) –81.98967
Significance level 0.0007
Log likelihood (R) –93.70841 Chi squared 23.43749
*), **), ***) statistically significant at 99 %, 95%
Pseudo R squared 0.125
and 90% confidence levels
From table 6-4 we see that the probability that the respondent accepts the proposed bid increases as the bid decreases, the higher bound of the possible cost range is proposed, income increases, the importance of algal blooms as the coastal problem increases, the vacation budget increases and the satisfaction of the respondent with the seawater quality increases.
9
The choice of the variables for the parametric model and alternative model specifications are presented in Appendix 2. The model in table 6-4 is the model D in table A2-2.
40
The likelihood ratio test performed to the model shows that the coefficients of the model are different from zero since the chi square value for the model, 23.44, exceeds the 95 % confidence level chi square value with seven degrees of freedom, 14.07. The Pseudo R squared value is a measure for the goodness of fit of the model and may take the value between zero and one. In this model, the value 0.125 is rather low since high values are preferred, but there is no commonly accepted value that denotes a well-specified model (Hanemann & Kanninen 1999, 344).
The coefficient BID is statistically significant at 99 % confidence level, the coefficient COASTAL at 95 % confidence level and the coefficients BOUND, EXPEND and SATISF are significant at 90 % confidence level. The signs of the coefficients BID and INCCAP are consistent with the theoretic expectations: the bid has negative effect indicating that the higher the bid the lower the probability to accept it, while income has positive effect. However, the effect of income is not statistically significant that is discussed in section 6.3. The signs of coefficients COASTAL and EXPEND make sense as well, while the negative sign of SATISF was unexpected.
The interpretation of the positive sign of BOUND is ambiguous. The respondents were told about the range of the estimated cost of the algal reduction, for example ‘the tax would be in the range of ___ and ___ Euro per person per year’. The positive sign of BOUND indicates that when the upper bound of the cost range is asked first, the probability that the respondent is willing to contribute increases. Thus, the positive sign may imply that the bids were too low to capture the maximum WTP of the respondents. Half of the respondents accepted the highest bid, 100 Euro, but the amount of respondents who faced this bid was low, only 10.
On the other hand, the positive effect of BOUND is a good sign because it indicates the absence of strategic behavior. The respondents knew that the tax could be set lower than the bid asked. They also were aware that their answers affect the decision making, so they could have refused the higher bid in order to obtain the lower tax level. However, the existence of the incentive to strategic behavior is essential for the economic interpretation of the WTP answers, because otherwise any WTP response
41
would be as good as any other, which is not similar to the market situation (Carson et al 2001, 189).
The positive sign of the coefficient COASTAL indicates that the higher rank the respondent gives to algal blooms in the sea – the more important s/he considers the algal problem – the more likely s/he accepts the offered bid. Instead of varying from 1 to 4, the coefficient COASTAL could be a dummy that takes the value of 1 when the respondent considers algal blooms in the sea as the most important coastal problem, and the value of 0 otherwise. The strict assumption behind a dummy is that the respondent would only contribute to the most important coastal problem if s/he were willing to contribute at all. We relax this assumption and use the varying variable instead of a dummy.
The positive sign of the coefficient EXPEND is consistent with the expectation. The higher the vacation expenditure per capita, the more likely the respondent is willing to contribute to algal problem. The variable EXPEND captures both the length of stay in Hanko and the travel costs from the place of residence, and thus, it reflects the importance of Hanko as vacation location. The more binding relationship to Hanko the respondent has, the more willing s/he is to contribute to algal bloom reduction.
The negative sign of the coefficient SATISF was unexpected. By intuition, the less satisfied the respondent is with seawater – the worse the quality of seawater is in his / her opinion – the more likely s/he is willing to pay for the improvement. Our result shows the opposite, the probability to state a positive WTP increases with the level of satisfaction.
One possible explanation is that people are rather willing to preserve the present good seawater quality than to contribute to the improvement of seawater quality when seawater is already spoiled and they are disappointed in it. People may purchase moral satisfaction, but there is no moral satisfaction anymore if someone else has polluted the sea. Satisfied people may want to contribute to ensure that seawater stays clean and one way to influence is to pay an environmental tax. When valuing an environmental good, people do not necessarily put the value on the good itself but the idea of environmental protection. 42
The respondents may have valued the HAB problem in the European Union scale due to the information given before the WTP question. They may have stated high WTPs because harmful algal blooms are a big problem in Europe. However, revealing this altruistic component in the WTP requires information on the people’s attitudes and motives to state positive WTP that we did not collect.
Studying the influences of the variables SATISF and COASTAL together, another explanation for the negative sign of SATISF is found. Since the tourists in Hanko were in general satisfied with the seawater (the average satisfaction level was 2.07 with the median of 2 implying ‘satisfied’) and despite, 40.8 % of the respondents ranked algal blooms as the most important coastal problem. This indicates that there are no very important coastal problems in Hanko. The comparison with the results from other case studies supports this as reported in section 6.4.2.
The aim of the study was to estimate the benefit for tourism sector in Hanko from a 25 % reduction in algae blooms by the beginning of the next summer season and a 50 % immediate reduction in the risk of getting shellfish poisoning. The estimation of the mean and the median WTP is 24.90 Euro per person per year. The mean WTP is the expected willingness to pay of the representative individual in the target population that is tourists in Hanko. The median WTP is the amount of money that half of the tourists are willing to pay for reduction, or if there were election about introduction of an environmental tax level of 24.90 Euro per person per year, half of the tourists would vote for it.
Given that the amount of tourists staying in accommodations in Hanko during the whole year is around 47 000 (Statistics Finland 2003a, 52), and thus during three summer months around 11 750, the aggregate yearly benefit for tourism sector from a 25 % reduction in algal blooms and a 50 % reduction in risk of getting shellfish poisoning is approximately 293 thousands Euro per year.10 10
This is likely underestimation because the amount of tourists reported in Statistics Finland (2003a) does not include day-trippers that are an important proportion of the visitors in Hanko (in our sample the proportion was 15 %). Moreover, the sea around the city of Hanko is iced on wintertime, and most of tourism takes place during summer months June, July and August. However, we have no better estimates available about the amount of tourists in Hanko in summertime.
43
6.3 Discussion The use of the variables COASTAL and EXPEND reduces the amount of observations from 196 to 139 due to lack of answers. Ranking the problems was difficult for some respondents probably because of the different nature of the problems ‘algal blooms in the sea’, ‘traffic and not enough parking spaces’, ‘not enough recreational activities’ and ‘sometimes noise or smell’, although they all affect the enjoyment from vacation. For some respondents it was difficult to estimate their vacation expenditure. The variable EXPEND has good explanatory power even though the number of observations is low. Probably people who are used to think in monetary terms are also more able to estimate the value of environmental improvement in monetary terms.
The negative sign of the coefficient SATISF reveals the difficulty of the CVM to reveal the monetary valuations of all people in the sample. Neither being satisfied with seawater at the moment of the interview nor being reluctant to pay means necessarily that the respondent does not place the positive value on seawater quality improvement. The presence of protest answers in the analysis may have the link with the sign of the coefficient SATISF as can be seen in table 6-5. The proportion of protests among the respondents in satisfaction categories increases when the level of satisfaction decreases.
Table 6-5. Protest behavior and the level of satisfaction with seawater quality (N = 139).
SATISF
Number of
Number of
Percentage of
respondents
protests
protests
1 = very satisfied
26
4
15.4 %
2 = satisfied
77
19
24.7 %
3 = not very satisfied
29
8
27.6 %
4 = not satisfied at all
7
3
42.9 %
139
34
Total
44
Besides the satisfaction with seawater quality of the interview day, the variable SATISF captures the respondent’s opinion on seawater quality in general, because s/he likely compares seawater quality of the day to his / her previous experience. Thus, comparing different people’s opinions about the quality of the day is difficult because we did not collect information on the respondents’ opinions on the seawater quality in general or on environmental attitudes of the respondents.
The expectation from economic theory that income has positive effect on willingness to pay is not statistically significant (p-value = 0.4247) even when income per capita is expressed in three income categories 0 – 999 €, 1000 – 1999 € and 2000 – 2999 €. This may be due to lack of sufficient number of household income categories in the questionnaire. From table 6-6 below we see that 11.7 % of the respondents of the sample (N = 196) were in the highest income category where net monthly household income was 5000 Euro or more and 4.1 % in the second highest income category with net monthly household income of 4500 – 4999 Euro. Thus, the long right tail of household income distribution is not revealed. The mean household income of the sample is 3161 Euro and the median household income is 3000 Euro. The distortion of the right tail of household income distribution, however, likely reduces when we use income per capita as the variable and divide household income by the number of household members.
45
Table 6-6. Personal and household willingness to pay per household income categories (N = 196).
Net
Number
Proportion
Average
Personal
Household
Personal
monthly
of obs. in
of total
number of
mean
mean
median
income
amount of
household
WTP
WTP
WTP
category
obs.
members
DK / REF
4
2.0
NA
NA
NA
NA
Less than
3
1.5
1.3
17.67
22.97
13.00
500 – 999
2
1.0
3.0
15.25
45.75
15.25
1000–1499
15
7.7
2.0
15.70
31.40
9.00
1500–1999
12
6.1
2.3
4.25
9.78
0.00
2000–2499
23
11.7
2.5
10.15
25.38
9.00
2500–2999
41
20.9
2.7
21.72
58.64
9.00
3000–3499
22
11.3
3.2
25.32
81.02
21.50
3500–3999
25
12.8
3.1
15.10
46.81
3.50
4000–4499
18
9.2
3.4
7.64
25.98
2.25
4500–4999
8
4.1
3.4
18.13
61.64
21.50
More than
23
11.7
3.3
17.41
57.45
9.00
household income (in Euro)
500
5000
Table 6-6 also presents the mean and the median WTPs calculated by household income categories. This is different from the view we have in the parametric model where income variable is per capita income. Alternative view is discussed here due to uncertainty of the budget constraint the respondents faced during the interview.
Our payment vehicle was a personal tax that would be paid by each resident in the European Union. During the interviews, it was noticed from the comments of the respondents when answering the WTP question that they often made the decision based on the total tax for the household that depends on the number of household members. In the parametric analysis, we studied only the individual budget constraint
46
and the personal WTP and ignored the household budget constraint because we have no information about how the respondents made the choice.
The household budget constraint view gets some support from table 6-6. The average number of household members (column four) increases slightly with the monthly household income. The higher the household income, the more members there are in the household. This indicates that spending vacation in Hanko is a kind of extra commodity that the individual can afford only if s/he has a certain level of per capita income available.
The mean WTP of respondents (column five), however, does not increase with household income but ranges from 4.25 to 25.35 Euro across household income categories. When we multiply the personal mean WTPs by the average number of household members, the ‘household WTP’ (column six) should increase with household income. However, due to the large WTP range the potential existence of the increase cannot be seen.
The large differences in the mean WTPs between the categories are likely due to small amount of observations. The median WTPs reported in the last column in table 6-6 support this. For example, in income category of 1500 – 1999 Euro, the mean WTP is 4,25 Euro and the median WTP is zero. In the small sample, zero WTPs influence strongly the average WTP calculation.
In addition to individual budget constraint and household budget constraint, there is one more possibly relevant budget constraint. The vacation budget can be the budget constraint faces by the individual in the utility maximization process instead of his / her disposable income. However, this approach would make more sense if our payment vehicle was an entrance fee instead of a common tax.
When the payment for the seawater quality improvement is included in the vacation budget the respondent considers seawater quality as the quality component in his / her vacation enjoyment, and s/he only values seawater quality improvement if s/he spends vacation in Hanko. This feature called weak complementarity is the link between the value of environmental improvement and behavior of the respondent in 47
revealed preference techniques such as travel cost method (see for example Haab & McConnell 2002, 10-11).
As mentioned in section 1.2, the use of revealed preference methods restricts the values to be estimated to direct use values, although our WTP estimate may contain for example indirect use value (marine ecosystem benefits), altruistic value (other people’s benefits), existence value and bequest value (benefits for future generations). However, we can test the convergent validity of our results by checking the sign of the coefficient EXPEND that reflects the vacation expenditure of the respondent. The positive sign supports the convergent validity of the results though we did not make a complete travel cost analysis. (see Carson et al 2001, 194)
The parametric model specification presented here was chosen based on the best statistical significance of the coefficients although the number of observations falls rather low (N = 139). In another model specification (see Model E in Appendix 2), the variable EXPEND is replaced by a dummy PROPERTY. From this model, owning a free time house in Hanko or a boat – having a relatively binding relationship to vacation location Hanko – has significant positive effect on WTP.
6.4 Comparison of results In section 6.4.1, the WTP estimates of our study and the previous studies reviewed in chapter 2 are discussed by comparing the components of the valuation scenarios. Since it is more appropriate to compare the WTP estimates of the Finnish case study to the WTP estimates of other case studies of the ECOHARM project, this is done in section 6.4.2.
6.4.1 Previous studies
Comparing the results from different CV studies is more than difficult because of different nature of the problem, target populations, hypothetical goods being valued,
48
payment vehicles, time scales and other components in valuation scenarios as well as different model specifications.
The study of Nunes and van den Bergh (2002) estimated the willingness to pay an annual pollution tax for two years. Due to the unspecified length of payment period in our study we can estimate the benefit for one year. The mean WTP in the Dutch study ranged from 58.2 to 58.8 Euro per person per year, while our mean WTP was lower, 24.90 Euro. Our scenario was a 25 % reduction in algal blooms by the beginning of the next summer season and a 50 % immediate reduction in the risk of getting shellfish poisoning, while the Dutch scenario was a 90 % reduction in risk of a biological pollution event.
Gren et al (1997) had the same payment vehicle, an environmental tax, as we did, and a specified timescale, 20 years. The mean WTP for total effect of nutrient reduction was 110 Euro per person per year. The amount is far higher than our 24.90 Euro per person per year for reduction in harmful algal blooms. The value of 110 Euro is Baltic Sea wide, while 24.90 Euro is the value of reduction in algal blooms and in risk of getting shellfish poisoning in the EU. The EU wide tax may have negative effect on the WTP in our study because not necessarily the money collected from Finnish people would be invested in the Gulf of Finland. Moreover, Finland is situated on the border of the EU and according to Kiirikki et al (2003), Finnish people are well aware that nutrients do not respect country borders.
Söderqvist and Scharin (2000) estimated the local WTP in Stockholm archipelago. The mean WTP of adult resident in the counties of Stockholm and Uppsala was from 48 to 79 Euro per year for ten years in the increased prices of agricultural products and tap water. The estimate is higher than ours, probably due to different payment vehicle.
Kiirikki et al (2003) studied the harmful effects of eutrophication in the Gulf of Finland. Although the results of a non-CV and a CV study cannot be directly compared, the answers in both studies reflect the same kind of attitudes and experience on HABs. Concerning financing of an environmental policy, 70 % of the respondents stated that society and polluters should pay for nutrient reduction. In our 49
study, 9.7 % of the respondents stated this as the motive for reluctance to pay, and 6.6 % demanded others to contribute to algal reduction as well.
The respondents of Kiirikki et al (2003) classified toxic algal blooms very harmful and fouling of beaches and turbidity of water harmful or very harmful. In our study, 13.8 % of the respondents had experienced problems with toxic algal blooms (HBT), and 52.6 % with high biomass non-toxic algae (HBNT), but we did not ask the degree of harmfulness. According to Gren et al (1997) in Sweden, 30 % of the respondents had personal experience on the effects of eutrophication (including HABs).
6.4.2 Case studies of ECOHARM project
Due to differences of the studies compared above, it is safer to compare the case studies from Finland, France, Italy and Ireland. The numbers in parentheses in the following text refer to the question numbers in the annotated questionnaire in Appendix 1. The differences between the countries are to the certain extent similar to the results of the European Union opinion poll Eurobarometer 58.0 (2002) that studied the attitudes of the Europeans towards the environment.
The recreational profile of the tourists differed across the countries. In Ireland, the tourists were not visiting only Galway but other cities as well (question B.2.), they came from longer distance (B.2.1.), and did not come just for beach activities such as swimming (B.6. and B.7.). In Finland, the proportion of day-trippers was high (B.3. and B.5.) as well as the amount of boaters (B.5.). In Italy, the respondents stayed mostly in hotel (B.5.), the stay in Riccione was longest compared to other countries (B.3. and B.11.) as well as the amount of regular visitors (B.9.). In Hyères Le Pradet Conquieranne in France, the amount of regular visitors was almost as high as in Italy (B.9.), and the number of trips to the location during the typical summer was highest (B.10.). In brief, the tourists in Galway were more of sporadic nature than in other locations. Comparing Italy and Finland, the Italians stayed in hotel while the Finns stayed in own property such as a boat or a free time house.
50
Concerning the coastal problems, the ranking of ‘algal blooms in the sea’ as the most important coastal problem was highest in Finland followed by Italy, Ireland and France (C.0.). This was even though the other coastal problems were slightly heavier in Finland (see C.0. in Appendix 1 for the Finnish versions of coastal problems) indicating that Hanko is in general a less problematic vacation location compared to other case study locations. The Finns were the most satisfied with the seawater quality, while the Italians were the least satisfied (C.2.). The Italians were also the only ones who stated that they would not come back to Riccione next summer due to seawater quality (B.16).
According to Eurobarometer 58.0 (2002), the Finns have the most positive associations with the word environment compared to the Irish, the Italians and the French. When hearing the word environment, the Finns think first about ‘green and pleasant landscapes’, ‘protecting nature’, ‘the state of the environment our children will inherit’ and ‘the quality of life where I live’, while the Irish, the French and the Italians think first about ‘pollution in towns and cities’ and ‘protecting nature’. (Eurobarometer 58.0 2002, 5)
According to Eurobarometer 58.0 (2002), the Italians and French are most worried about pollution of the seas and coasts, since the percent of people stating they are ‘very worried’ is 53 % in Italy, 49 % in France, 33 % in Finland and 30 % in Ireland. The average of the 15 EU countries is 42 %. In general, the southern countries are more worried about the environment as a whole than the northern countries. The Finns feel being the most aware of sea pollution compared to other EU citizens. The percents of people feeling either ‘very well or fairly well informed’ are 76 in Finland, 61 in Italy, 42 in Ireland and 38 in France. The EU average is 48. (Eurobarometer 58.0 2002; 9-11, 16)
Our study shows that the Finns and the Italians had more experience on high biomass nontoxic algae (HBNT) than the Frenchmen and the Irish (E.2.). The Finnish had most experience on toxic algal blooms (E.3.). This is likely due to the scale of the question E.3. The Finnish version included the experience on high biomass toxic blooms (HBT) in addition to seafood toxic blooms (STB), while in Italy, France and Ireland the question E.3. concerned only seafood toxic blooms (STB). 51
The mean willingness to pay of our study (24.90 Euro) was calculated from the standard probit model. However, when comparing the results of case studies we use the WTP estimates calculated from the bivariate probit model without covariates that was used in other countries (see Scatasta et al 2003). From that model specification, the mean WTP was highest in Finland, 34.11 Euro per person per year for one year, followed by Ireland (29.21 Euro), Italy (25.63 Euro) and France (7.97 Euro) (E.10.). The high WTP in Finland probably reflects the higher income per capita in Finland (F.6.), the importance of the ‘algal blooms’ as the coastal problem (C.0.) and the larger scale of the problem.
The high WTP for algal reduction in Finland may also indicate the belief for the effect of own actions on the environmental state in general. In Eurobarometer 58.0 (2002), the Europeans were asked about opinions concerning the ability to affect the environment by one’s own actions. While 66 % of the Finns are confident about their actions do make differences, the Italians, the French and the Irish have a more negative view. More than half (58 %) of the Italians and half (50 %) of the French are of opinion that environmental state is beyond their control. The Irish are in between, 45 % do believe that their actions make difference and 37 % believe that the environmental state is beyond their control. (Eurobarometer 58.0. 2002, 23)
In ECOHARM case study of Italy, the most common reasons for reluctance to pay (E.11.) were ‘it is not worth paying’ and ‘we pay enough taxes’. The same reasons were the two most popular in other countries as well. The percent of ‘I do not believe in the scenario’ was considerably high in Finland likely due to awareness of the Finns about the nature of the algal problem in the Gulf of Finland.
Concerning the trust in the EU in the environmental issues, the Italians are most trusting, as 33 % of them trust the EU. The percents in other countries are 11 % in Ireland, 10 % in Finland and 9 % in France. The EU average is 13 %. What comes to solving the environmental problems, the attitudes of people all of the four countries have the same direction. The percentages of people stating ‘taxing those who cause environmental problems’ ranged from 31 % (Finland) to 49 % (France), and ‘taxes for everyone’ ranged from 4 % (Italy) to 7 % (Ireland) while the percentages of other 52
two ECOHARM countries are somewhere in between. (Eurobarometer 58.0 2002; 27, 32)
53
7. Methodology discussion and conclusions In this chapter some critical points against the CVM as well as the counter-arguments are presented and the conclusions are drawn.
7.1 Critical view In literature, the criticism against contingent valuation method has arisen from the valuation of nonuse values, the absence of preferences of respondents or the difficulty or inability to put the preferences into order.
According to Diamond and Hausman (1994), the reliance on CV surveys on policy decision making is misguided. The CV surveys remind opinion polls and thus, they are not accurate basis for the government actions. The respondents hardly are responding to the question the interviewer is trying to ask, and although the non-true answers can be discarded from the analysis, it is not so that the not obviously wrong answer is the accurate answer to the interviewer’s question. Moreover, the researchers sometimes arbitrarily discard the high WTP responses as outliers without perfect knowledge about the real preferences of the respondent as well as the zero bidders because they are identified as protest zeros based on the questions elsewhere in the questionnaire. (Diamond & Hausman 1994, 46-47)
The ‘embedding effect’ analyzed by Kahneman and Knetsch (1992) means that the respondents are not valuing the good in question but rather embedding the good with the similar goods in the larger scale. Andreoni (1989) states that people get a ‘warm glow’ from supporting a good purpose, and Kahneman and Knetsch (1992) call this the purchase of moral satisfaction. People may also not be able to find their own valuation for the good in question, but rather they might think which value the others may place on the environmental good. The presence of many possible misleading features in a CV study leads to that very likely the amount of people whose answers can be treated as accurate reduces since different problems face different people. It is
54
unclear whether the amount of answers is sufficient enough to make any useful conclusions about the best policy implementation. (Diamond & Hausman 1994, 4648)
Serious problems arise in the welfare analysis. First, an altruism component – i.e. people state the value including their concern for other people’s benefits from environmental amenity – in the stated willingness to pay leads to double counting in cost-benefit analysis, when the WTPs of people are added up in order to get the aggregate social benefits of the policy in the question. Second, even if the stated WTP was the best guess the respondent can make, it may still be a poor guess, because the link between the good to be evaluated and the utility from the good is hard to define when maximizing utility. Also the ‘warm glow’ effect may harm the analysis of the CV data by leading to inaccurate responses. (Diamond & Hausman 1994, 55-56)
Diamond and Hausman (1994) root their criticism to the so-called Washington fallacy, the argument that ‘some number is better than no number at all’. Consistently, it is better to make the cost-benefit analysis based on biased and inaccurate WTP numbers than with zero values which are adjusted somewhere else in the analysis. The question is whether including the inaccurate CV survey results about the nonuse value of environmental amenity will improve the determination of the government policy. (Diamond & Hausman 1994, 58-59)
The referendum format used in CV surveys is criticized because different bases for the voting behavior of people in the referendum imply different appropriate uses of the responses. In addition, people do not necessarily vote in the same way in binding and nonbinding referendum. The time available for the interview may be too short for the respondents to digest the information given. Moreover, the legitimacy of using the referendum in obtaining economic values can be questioned because of the skepticism of the extent to which the referendum represents informed decision making. (Diamond & Hausman 1994, 59-60)
In 1993, the NOAA Panel was set to discuss the controversies concerning the CVM and as a result the guidelines for an ideal CV survey were reported. Diamond and 55
Hausman (1994, 62) argue that while the Panel states that the survey that does not follow their guidelines may be biased, this does not necessarily mean that the survey meeting the guidelines is not biased. The Panel does not motivate the conclusion that the guidelines produce estimates ‘reliable enough’ (NOAA 1993, 4610), and it does not call for testing the reliability of the survey made based on the guidelines. As Mäler (1993) puts it, the main problem in CVM is that there is no error correction mechanism that would automatically point out the errors and biases in the CV study.
The conclusion of Diamond and Hausman (1994) is that the contingent valuation methodology used in measuring the nonuse values does not estimate what the proponents of the methodology claim it to estimate. The hypothetical market restricts the ability to judge the quality of the CV responses and the ability to calibrate the answers to usable numbers. The internal consistency problems are due to the absence of preferences of the individuals for the particular environmental sites, and because the responses across the surveys are inconsistent, the survey responses are not accurate bases for the environmental policies. (Diamond & Hausman 1994, 62-63)
7.2 Counter-arguments Carson, Flores and Meade (2001) point out that the criticism against the CVM has led to the development of the theoretical framework of the contingent valuation and the approaches for assessing the quality of CV results. Traditionally, economists have favored the revealed preference techniques, but the growing concern over the environment has shown that for many commodities there is no direct link to behavior to be used in the valuation. The answer to the criticism that passive use is not an economic value is that whatever gives utility to the individual has an economic value whatever the source is. The reliable aggregation of the CV results is not a problem of the CV method itself but rather of the use of the results. (Carson et al 2001, 173-179)
As the counter-argument to criticism about the absence of meaningful responses, Carson et al (2001) present some real market situations. As the new products enter the market, people have no prior experience but still they make rational choices.
56
Moreover, they do not often use too much time to make decisions about the purchases. Taking the value people put on culturally valuable topics as an example, people certainly are able to value to something they do not have personal experience. (Carson et al 2001, 178-179)
The sensitivity of the WTP estimates to the components of the hypothetical market can be the advantage of the CVM, because the influences of the way an environmental good is provided can be estimated. The context in which the decision is made is under the researcher’s control, and due to this, the researcher must know for example the effects of different elicitation formats on WTP. A number of studies have been conducted concerning theory about predictions of differences between WTP estimates and empirical studies have supported theory. The differences are described for example in Carson, Groves and Machina (1999). When creating a market in a CV study, the researcher should make a decision context as real as possible, and take the possibility of strategic behavior seriously. (Carson et al 2001, 180-189)
Performing tests for reliability and validity are essential to show the accuracy of the CV results. Construct validity of the CV results is about how well WTP is predicted by the factors that were expected to be predictive. The consistency with economic theory can be tested by checking that the proportion of people willing to pay decreases as the bid increases. The scope tests can be performed to check that people are willing to pay more for larger amount of the good. In addition, for example direct users of the good and environmentalists are expected to be willing to pay more. Perception variables concerning for example the payment vehicle or the expected successfulness of the project are as well expected to be good predictors of WTP. (Carson et al 2001, 180-194)
Convergent validity is tested by comparing the CV estimates with the corresponding estimates from travel cost or hedonic pricing studies. The ways of comparison are the ratios, the differences and the correlations between estimates. The reliability of estimates means the reproducibility and stability of WTP measure. Stability is essential when the CV results are used for policy recommendations. The reliability is
57
checked by comparing estimates from different samples or from the same respondents at different times. (Carson et al 2001, 194-195)
As final conclusion, Carson et al (2001) remind that the criticism arisen due to some particular low-budget or otherwise non-accurate CV studies does not imply that all CV estimates are unvalid. The CVM is not an easy nor cheap method to conduct, and the challenge for future is to reduce the cost of CVM without losing the quality obtained in the CVM field. Concerning the use of CV estimates in policy making, the estimates should not be taken as a self-evident basis, but without the CVM some of the aspects people care about are missing in the CBA making it incomplete or misleading. (Carson et al 2001, 196-197)
In addition to the components of the counterfactual market, the functional and distributional assumptions in the analysis step affect the accuracy of WTP estimates as discussed in section 4.2. Moreover, different methodologies work better in different cases. The recent development of the CVM has elicited the preferences in alternative ways, for example by combining the formats of contingent choice, contingent ranking or scaling with the referendum contingent valuation format, or used them instead of the referendum format. (Hanemann & Kanninen 1999, 407) In Finland, Rekola (2003) used the model of lexicographic preferences to analyze incommensurability between the private and the public good. Pouta (2003) combined the theory of planned behavior with the CVM and studied the role of respondents’ attitudes and beliefs in WTP responses.
7.3 Conclusions We applied the contingent valuation method to estimate the benefit from seawater quality improvement for tourism sector. The survey conducted in Hanko was one of the case studies of the European Union project ECOHARM.
The benefit from a 25 % reduction in algal blooms by the beginning of the next summer season and a 50 % reduction in the risk of getting shellfish poisoning was
58
estimated by calculating people’s mean willingness to pay. The estimated willingness to pay was 24.90 Euro per person per year for one year. Since the sample was as representative as possible to the tourist population in Hanko, the benefit estimate can be expanded to the aggregate yearly benefit for tourism sector as 293 thousands Euro per year for one year.
The estimate for willingness to pay was calculated from the standard probit model. The parameters of the model show that the willingness to pay for algal bloom reduction increases significantly with vacation expenditure, importance of algal blooms compared to other coastal problems and satisfaction with seawater quality. The bid has negative effect on the willingness to pay. The signs of the coefficients bid and income did meet theoretical expectations. In another model specification, owning a free time house in Hanko or a boat has significant positive effect on the willingness to pay for algal bloom reduction.
The study revealed information on Finnish people’s experience on harmful algal blooms in the Gulf of Finland and attitudes towards coastal problems in Hanko. More than half (55,6 %) of people had experienced problems with harmful algal blooms. About two fifths (40,8 %) of people ranked algal blooms in the sea as the most important coastal problem in Hanko, but it must be noted that only 85,2 % of the respondents were able to answer this question.
The estimated willingness to pay amounts vary across the case studies of ECOHARM. The differences in attitudes towards the environment, characteristics of tourists and local nature of algal problem explain the variation between Finland, Italy, France and Ireland.
The willingness to pay for the described algal bloom reduction scenario was highest in Finland as well as income per capita. The high willingness to pay may also indicate that the Finns valuated the algal problem in the European Union scale while the Italians, the French and the Irish valuated the problem in local scale. Moreover, compared to the Italians, the French and the Irish, the Finns are the most confident about that their contribution has effect on the situation.
59
Our study focused on the benefit from algal bloom reduction for tourism. Environmental problems such as algal blooms in the sea affect the welfare of residents of the coastal cities and municipalities as well. Expanding the research to study willingness to pay of coastal residents and people who never use the coastline for recreational purposes could reveal the variation in willingness to pay due to different values of the coastline. The estimates of a locally conducted study could be the basis for the policy making at the local level.
60
References Andreoni, J. 1989. Giving with Impure Altruism: Applications to Charity and Ricardian Equivalence. Journal of Political Economy 97, 1447-1458.
Bateman, I.J., R.T. Carson, B. Day, M. Hanemann, N. Hanley, T. Hett, M. JonesLee, G. Loomes, S. Mourato, E. Özdemiroglu, D.W. Pearce, R. Sugden and J. Swanson. 2002. Economic Valuation with Stated Preference Techniques: A Manual. Edward Elgar Publishing.
Bishop, R. and T. Heberlein. 1979. Measuring Values of Extra-market Goods: Are Indirect Measures Biased? American Journal of Agricultural Economics 61(4), 926-930.
Cameron, T.A. 1988. A New Paradigm for Valuing Non-Market Goods Using Referendum Data: Maximum Likelihood Estimation by Censored Logistic Regression. Journal of Environmental Economics and Management 15: 355379.
Carson, R.R., N.E. Flores and N.F. Meade. 2001. Contingent Valuation: Controversies and Evidence. Environmental and Resource Economics 19: 173-210.
Carson, R.T., T. Groves and M. Machina. 1999. Incentive and Informational Properties of Preferences Questions. Planery Address. European Association of Environmental and Resource Economists. Oslo, Norway.
Cooper, J.C., M. Hanemann and G. Signorello. 2002. One-and-one-half-bound Dichotomous Choice Contingent Valuation. The Review of Economics and Statistics 84(4): 742-750.
61
Diamond, P.A. and J.A. Hausman. 1994. Contingent Valuation: Is Some Number better than No Number? The Journal of Economic Perspectives, Vol. 8, No 4. Pp. 45-64.
Eurobarometer 58.0. 2002. The attitudes of Europeans towards the environment. The European
Opinion
Research
Group
(EORG).
Available
in
web:
http://europa.eu.int/comm/environment/barometer/barometer_2003_en.pdf. August 22, 2003.
Gren, I.M., T. Söderqvist and F. Wulff. 1997. Nutrient reductions to the Baltic Sea: Ecology, Costs and Benefits. Journal of Environmental Management 51: 123143.
HAEDAT 1998. The Harmful Algal Blooms Events Database. Maintained by the IOC. Available in web: http://ioc.unesco.org/hab/data33.htm. May 8, 2003.
Haab, T.C. and K.E. McConnell. 2002. Valuing Environmental and Natural Resources: The Econometrics of Non-market Valuation. New Horizons in Environmental Economics. Edward Elgar Publishing.
Hanemann, W.M. 1984. Welfare Evaluations in Contingent Valuation Experiments with Discrete Responses. American Journal of Agricultural Economics 66: 332-341.
Hanemann, W.M. 1999. Neo-classical Economic Theory and Contingent Valuation. In: Bateman I.J. and K.G.Willis: Valuing Environmental Preferences. Theory and Practice of the Contingent Valuation Method in the US, EU, Developing countries. Oxford University Press. Pp. 42-96.
Hanemann, W.M. and B. Kanninen. 1999. Statistical Analysis of Discrete-response CV data. In: Bateman I.J. and K.G.Willis: Valuing Environmental Preferences. Theory and Practice of the Contingent Valuation Method in the US, EU, Developing countries. Oxford University Press. Pp. 302-441.
62
Hanemann, W.M., J. Loomis and B. Kanninen, 1991. Statistical Efficiency of Double-Bounded Dichotomous Choice Contingent Valuation. American Journal of Agricultural Economics 73(4): 1255-1263.
Hanko 2003. Website of the city of Hanko. Available in web: http://www.hanko.fi. May 6, 2003.
Kahneman, D. and J.L. Knetsch. 1992. Valuing Public Goods: The Purchase of Moral Satisfaction. Journal of Environmental Economics and Management 22: 57-70.
Kiirikki, M., P. Rantanen, R. Varjopuro, A. Leppänen, M. Hiltunen, H. Pitkänen, P. Ekholm, E. Moukhametshina, A. Inkala, H. Kuosa and J. Sarkkula. 2003. Cost effective water protection in the Gulf of Finland: Focus on St. Petersburg. The Finnish Environment 632.
Markandya, A., P. Harou, L.G. Bellù and V. Cistulli. 2002. Environmental Economics for Sustainable Growth. A Handbook for Practitioners. Edward Elgar Publishing.
Markowska, A. and T. Zylicz. 1996. Costing an International Public Good: the case of the Baltic Sea. Mimeo, Warsaw Ecological Economics Center, Warsaw University.
McConnell, K.E. 1990. Models for Referendum Data: The Structure of Discrete Choice Models for Contingent Valuation. Journal of Environmental Economics and Management, 18: 19-35.
Mäler, K.-G. 1971. A Method of Estimating Social Benefits from Pollution Control. Swedish Journal of Economics 73: 121-133.
Mäler, K.-G. 1974. Environmental Economics: A Theoretical Inquiry. Resources for the Future. Johns Hopkins University Press, Baltimore.
63
Mäler, K.-G. 1993. Contingent Valuation – does it prove any useful information? Comment on the Report of the NOAA Panel on Contingent Valuation. In: Kuyvenhoven, A., Ruijs, A., Schipper, R.A. and J. Wesseler (Eds.). 2003. Cost Benefit Analysis and Environmental Valuation. Wageningen University, Social Sciences.
NOAA. 1993. Report of the NOAA (National Oceanic and Atmospheric Administration) Panel on Contingent Valuation. Federal Register, Vol. 58, No. 10, January 15, 1993. Pp. 4601-4614.
Nunes, P.A.L.D and J.C.J.M. van den Bergh. 2002. Measuring the Economic Value of a Marine Protection Program against the Introduction of Non-Indigenous Species in the Netherlands. Tinbergen Institute Discussion Paper.
Pitkänen, H., J. Lehtoranta and A. Räike. 2001. Internal Fluxes Counteract Decreases in External Load: The Case of the Esturial Eastern Gulf of Finland, Baltic Sea. In: Ambio Vol 4-5 August 2001. Royal Swedish Academy of Sciences.
Pouta, E. 2004. Attitude and Belief Questions as a Source of Context Effect in a Contingent Valuation Survey. Journal of Economic Pychology 25(2): 229242.
Rekola, M. 2003. Lexicographic Preferences in Contingent Valuation: A Theoretical Framework with Illustrations. Land Economics 79(2):277-291.
Scatasta, S., E. Granèli, W. Stolte, H.-P. Weikard and E. van Ierland. 2003. The Report
for
the
Project
ECOHARM.
Available
in
web:
www.bom.hik.se/ECOHARM. January 8, 2004.
Statistics Finland 2003a. Tourism statistics. Transport and Tourism 2003:11. Pages 52 and 69. Statistics Finland.
Statistics Finland 2003b. Website of Statistics Finland. Available in web: http://www.tilastokeskus.fi/index_en.html. August 15, 2003. 64
Sugden, R. 1999. Public Goods and Contingent Valuation. In: Bateman I.J. and K.G.Willis: Valuing Environmental Preferences. Theory and Practice of the Contingent Valuation Method in the US, EU, Developing countries. Oxford University Press. Pp. 131-151.
Söderqvist, T. and H. Scharin. 2000. ”The regional willingness to pay for a reduced eutrophication in the Stockholm archipelago”, Beijer Discussion Paper Series No. 128, The Beijer Institute
65
Appendix 1. Annotated questionnaire
Note: the results from Italy, France and Ireland are from Sara Scatasta, Wageningen University Note: the information in gray boxes are discussed in section 6.4.2.
The locations: Hanko Finland Riccione Italy Hyères, Le pradet, Conquieranne France Galway Ireland
Section A: Introduction and screening
Good morning/afternoon. Would you like to participate to a study for the European Union on tourism and seawater quality? We ask questions like how many times you came to [location name], what is your favorite recreational activity, what you think about the quality of seawater and similar. It will take at most 15 minutes. [If NO reply “ Please your opinions are very important to our study”. If the respondent still refuses go to section G of the questionnaire] [If YES continue] Thank you.
A.1. Are you a resident of [location name] or are you just visiting? By resident I mean that in a typical year you live in [location name] for at least 6 months. 1. Resident 2. Visitor
A.2. Are you older than 18 years of age? 1. YES 2. NO
66
Population: Tourists of selected location older than 18 years of age. Response rate: Number of usable interviews / Number of contacted members of the population 196 / 325 = 60.3%
SECTION B: Recreational profile
B.1. How many members of your household are traveling with you including yourself? By members of your household I mean members of your family who live with you in your house. B.1 Group size
Mean
Median
St. deviation
Min / Max
Finland N=196
2.45
2
1.20
1/6
Italy N=203
2.69
2
1.39
1/8
France N=193
2.76
2
1.34
1/7
Ireland N=199
2.45
2
1.22
1/6
B.2. Is visiting [Location name] the main purpose of your visit to this region or are you visiting other cities in this area? B.2. Unipurpose (%) 1 = Main purpose
0 = Not main
99 = DK/REF
purpose Finland N=196
62.76
67.73
0.51
Italy N=203
89.16
10.84
0
France N=193
87.05
8.29
4.66
Ireland N=199
54.77
44.22
1.01
B.2.1. Where are you from? B.2.1 Distance
Mean
Median
St. deviation
Min / Max
Finland N=196
408.90
140
1102.81
36 / 9537
Italy N=203
357.52
296
374.18
3.5 / 2820
France N=190
420.13
345.00
438.34
2.5 / 2339
Ireland N=198
2714.89
1564
3774.51
34 / 18600
67
B.3. How long are you going to stay in [Location name]? B.3. Current stay
Mean
Median
St. deviation
Min / Max
Finland N=196
6.58
3
12.53
1 / 90
Italy N=203
17.71
13
20.31
1 / 92
France N=189
10.87
7
13.05
0.13 / 92
Ireland N=198
13.7
3.5
23.44
0.08 / 92
B.4. What means of transportation did you use? 1. Car
5. Mixed (car, train, plane, bus)
2. Train
6. Recreational vehicle
3. Bus
7. Motorbike, bike
4. Plane
8. Boat 99. DK / REF
B.4. Transportation 1
2
3
4
5
6
7
8
99
(%) Finland N=196
57.14 3.57
1.02
0.51
1.02
17.86 2.04
16.84 0
Italy N=203
84.24 8.37
2.46
3.45
1.48
0
0
0
0
France N=193
88.08 5.7
0
1.04
3.11
1.04
0
0
1.04
Ireland N=199
31.66 7.54
28.14 6.53
0
1.51
1.51
23.12 0
B.5. What type of accommodation did your household choose? 0. None
5. Friend’s property
1. Hotel
6. Own property
2. B&B
7. Hostel
3. Camping
8. Boat
4. Rental property
99. DK/REF
B.5.
0
1
2
3
4
1.02
31.12 5.10
5
6
7
8
99
0
17.86 0
Accommodation (%) Finland N=196
14.80 9.18
Italy N=203
0
57.64 0.49
8.87
France N=193
4.15
8.29
13.47 21.24 6.74
Ireland N=199
4.02
13.57 23.62 1.51
0.52
68
12.24 8.67
14.78 6.4
10.84 0
0
0.99
11.92 0
0
37.82
11.06 13.07 8.54
21.61 1.51
1.51
B.6. When you are in [Location name] in the summer, what is your favourite recreational activity? By summer I mean the months of June, July and August. B.6. Swimming (%)
1 = If swimming
0 = If other
99 = DK/REF
Finland N=196
45.41
53.06
1.53
Italy N=203
19.21
79.8
0.99
France N=193
43.01
56.48
0.52
Ireland N=199
8.54
88.94
2.51
B.7. When you are in [Location name] how often do you go to the beach, go swimming, go fishing, go boating, or do saltwater sports? 1. Every day 2. Every other day 3. Once a week 4. Other: B.7 Beach *)
Mean
Median
St. deviation
Min / Max
Finland N=191
0.88
1
0.28
0/1
Italy N=203
0.97
1
0.11
0.11 / 1
France N=192
0.85
1
0.27
0.14 / 1
Ireland N=187
0.25
0.14
0.33
0/1
*) Note: Days on the beach / Days at the location; Every day = 1, Every other day = 0.5, Once a week = 1/7 etc. B.8 How many vacation days can you get in a typical summer? 1.Whole summer 2. 1 month 3. 2 weeks 4. 1 week 5. Other: B.8. Vacation days *) Mean
Median
St. deviation
Min / Max
Finland N=195
38.69
31.00
23.15
0 / 92
Italy N=201
39.43
21.00
32.15
0 / 92
France N=188
50.40
31.00
34.99
0 / 92
Ireland N=185
38.14
21.00
33.01
0 / 92
*) Note: The whole summer = 92, 1 month = 31, 2 weeks = 14, 1 week =7, etc.
69
B.9. Do you come to [Location name] regularly? 1. Yes
[Go to B.10.]
2. No
[Go to B.12.]
B.9. Regular (%)
1 = Yes
0 = No
99 = DK/REF
Finland N=196
55.10
44.90
0
Italy N=203
67.49
32.51
0
France N=193
65.80
32.12
2.07
Ireland N=199
36.18
63.32
0.50
B.10. In the typical summer, how many times do you come to [Location name]? 1. Every weekend and the summer holidays 2. Every weekend but not the summer holidays 3. Only for the summer holidays 4. Other: B.10. Reg Trips *)
Mean
Median
St. deviation
Min / Max
Finland N=103
3.06
1
3.19
1/9
Italy N=137
3.00
1.00
7.49
1 / 60
France N=115
12.00
3.00
16.81
1 / 92
Ireland N=65
2.98
1.00
3.81
1 / 12
*) Note: Every weekend and summer holidays = 9, Every weekend but not the summer holidays = 12 – (holidays / 7), Only for the summer holidays = 1, etc. B.11. On average, how many days does a typical visit to [Location name] last? [Skip B.12., B.13. and B.14.] B.11. Reg Days
Mean
Median
St. deviation
Min / Max
Finland N=108
9.34
4
15.33
1 / 92
Italy N=137
21.17
14.00
24.50
1 / 92
France N=75
6.55
1.00
16.02
0.5 / 92
Ireland N=66
9.48
3.75
17.50
0.5 / 92
B.12. Last summer how many times did you visit [Location name]? B.12. Nonreg Past Trips
Mean
Median
St. deviation
Min / Max
Finland N=88
0.37
0
0.90
0/5
Italy N=66
0.23
0
0.49
0/2
France N=42
1.11
0
4.87
0 / 30
Ireland N=127
0.15
0
0.42
0/2
70
B.13. This summer how many times will you be visiting [Location name]? B.13. Nonreg Current Trips
Mean
Median
St. deviation Min / Max
Finland N=88
1.34
1
1.04
1 / 10
Italy N=66
1.04
1
0.27
1/3
France N=43
2.38
1
4.89
1 /30
Ireland N=125
1.10
1
0.32
1/3
B.14. On average, how many days does a typical visit to [Location name] last? B.14. Nonreg Days Mean
Median
St. deviation
Min / Max
Finland N=88
2.62
2
3.36
1 / 31
Italy N=66
6.09
5
5.08
0.5 / 31
France N=43
11.58
14
7.31
1 / 31
Ireland N=124
11.89
3
20.92
1 / 92
B.15. Will you come back to [Location name] next summer? If Yes, go to the question B.17. If No, ask the next question. B.15. Comeback (%) 1 = Yes
0 = No
0.5 = DK / Maybe
Finland N=196
71.94
9.18
18.88
Italy N=203
71.43
13.3
15.27
France N=193
59.07
18.65
22.28
Ireland N=199
62.31
26.63
11.06
B.16. May I ask you why? B.16. Reasonnotback (%)
1 = Seawater quality
0 = Other reason
99 = DK/REF
Finland N=18
0
94.44
5.56
Italy N=27
1.48
98.52
0
France N=36
0
100
0
Ireland N=53
0
100
0
71
B.17. Last summer, did you go somewhere else for vacation? [Add “beside [Location name]” if necessary] B.17. Additional Site (%)
1 = Yes
0 = No
99 = DK / REF
Finland N=196
86.67 (72.45) *)
12.82 (27.04)
0.51 (0.51)
Italy N=203
60.10
37.44
2.46
France N=193
56.99
43.01
0
Ireland N=199
83.92
15.08
1.01
*) Note: the percentages in parentheses correspond to ‘additional site’ other than Hanko, Gulf of Finland, Archipelago Sea or Åland (this is done due to the occurrence of HABs widely in the sea)
B.18. Where did you go? (Not available)
SECTION C. Coastal problems and seawater quality
C.0 Now I am going to read you 4 types of problems that might be important in [Location name]. Based on you experience and opinion, I would like you to order them from 1 to 4, 1 for the most important and 4 for the least important. The problems are: (Finnish versions of coastal problems) •
Traffic and (not enough) parking spaces__________
•
Algal blooms in the sea __________
•
Not enough recreational activities__________
•
Too much noise (Sometimes noise or smell)__________
C.0. Algae (%) *) 1
2
3
4
Mean Median
St. dev
Min/Max
1.84
2
0.96
1/4
Finland N=167
47.90 26.95 17.96 7.19
Italy N=202
32.51 18.72 6.40
41.87 2.58
2
1.32
1/4
France N=161
15.54 2.59
58.03 3.29
4
1.18
1/4
Ireland N=176
12.56 13.07 24.12 38.69 3.00
3
1.08
1/4
7.25
*) Note: Algae varies from 1 to 4
72
C.2 How would you say you feel about the quality of the seawater today? Would you say: you feel 1. Very satisfied 2. Satisfied 3. Not very satisfied 4. Not satisfied at all C.2. Seawater (%) 1
2
3
4
Mean Median
St. dev
Min/Max
Finland N=191
23.56 50.78 20.42 5.24 2.07
2
0.80
1/4
Italy N=202
10.34 51.23 29.06 8.37 2.36
2
0.78
1/4
France N=193
13.99 61.14 19.17 5.70 2.16
2
0.73
1/4
Ireland N=162
9.55
2
0.69
1/4
48.74 20.10 3.02 2.20
C.3 Could you tell me why? (Not available)
SECTION D: Travel costs and recreational expenditure
D.1. How much has your household planned on spending for this visit to [Location name] including travel and accommodation costs? D.1. Holiday budget
Mean
Median
St. deviation
Min / Max
Finland N=88
335.43
160.00
509.31
10 / 3000
Italy N=199
1549.00
1250.00
1269.00
12 / 9000
France N=138
602.60
176.22
798.03
0 / 4000
Ireland N=182
583.94
275.00
892.35
20 / 6000
D.3. How much did your household spend for accommodation? D.3. Accommodation costs Mean
Median
St. deviation
Min / Max
Finland N = 24
97.71
38.00
139.06
0 / 540
Italy N = 41
1246.00
832.00
1022.00
0 / 5500
France N = 9
680.56
575.00
804.26
0 / 2500
Ireland N = 11
576.32
450.00
473.21
0 / 1380
D.4. Did your household purchase or rent any recreational equipment to be used during this visit in [Location name] for beach or water related activities such as boats, wind surfs, and similar? (Not available.)
73
D.4.1. Please state the type of equipment and for each type the amount spent. (Respondents were not able to answer this in a meaningful way.)
(The questions D.5. – D.8. concerning mussel-eating habits were not asked in Finland.) D.5. In the past 12 months did you eat mussels? If you do not remember the past 12 months think simply of the last month. D.6. In a typical year how many times do you eat mussels? D.7. Do you usually eat mussels at home, at the restaurant or both? D.8. You do not eat mussels more often because of the flavor, the risk of getting shellfish poisoning, difficulty to prepare or other?
SECTION E. Willingness to pay
Now I am going to ask you your opinion about some environmental problems related to the coast of Hanko. Recent scientific measurements show the presence of algal blooms in the seawater along the coasts of the European Union in general and along the coast of Hanko in particular. Algal blooms are fast growing algae induced by particular weather conditions and the presence of pollutants (in Finnish version: nutrients) in seawater. These algal blooms may reach concentrations that can be seen with the naked eye, and cause foam and mucilage on the coast, lime fishing nets and spoil the taste of fish.
E.2. Have you ever had problems with this type of algae (HBNT) such as, for example, avoided bathing and swimming in seawater because of these algae? E.2. HBNT
1 = Yes
0 = No
99 = DK/REF
Finland N=196
52.55
47.45
0
Italy N=203
70.94
29.06
0
France N=193
17.10
82.38
0.52
Ireland N=199
26.63
73.37
0
Algal blooms may also be toxic. (In the Gulf of Finland and the Baltic Sea, these algae may cause skin and eye irritation to swimmers and kill fish, sea birds, and domestic animals. In the Atlantic and the Mediterranean Sea,) These toxic algal blooms are caused by other species and these blooms may cause the accumulation of potent toxins in bivalve mollusks such as mussels. If contaminated mollusks are eaten, they may cause diarrhea, nausea,
74
vomiting, abdominal cramps, and in more serious cases paralysis and death. In the past 10 years, there have been 1000 cases of intoxication in the European Union. Nobody died.
E.3. Did you ever have problems with this type of algae (STB) (In Finland HBT as well.)? Please specify the type of problem (Hepatitis or other) and if you contacted a doctor. E.3. STB
1 = Yes
0 = No
99 = DK/REF
Finland N=196
13.78
86.22
0
Italy N=203
9.85
89.16
0.99
France N=193
6.74
92.75
0.52
Ireland N=199
2.51
97.49
0
The European Union is thinking of reducing the amount of pollutants in the seawater improving the treatment of wastewaters flowing into rivers as well as the sea. The European Union also wants to reduce the risk of intoxication from eating mussels increasing monitoring activities of toxic algal blooms (sampling water and mussels more often than it does right now). The objective of this policy is to achieve a 25% reduction in algal blooms by the beginning of the next (the Finnish word was ‘seuraava’ instead of ‘ensi’) summer season. To finance this policy, the European Union is thinking of introducing a mandatory contribution specific for coastal zone management that would be paid by everybody. This mandatory contribution would be paid once a year by each resident of the European Union and by tourists who do not reside in the European Union but visit European Union coastal zones. This mandatory contribution would be paid also by firms in the industrial, agricultural and fishing sector based on the type and size of business.
E.9. For European Union residents and tourists who do not reside in the European Union but visit European Union coastal zones, this mandatory contribution would be in the range of ______ and ______ Euro per person per year. If you knew that this money would be used to achieve a 25% reduction in algal blooms by the beginning of the next summer season, and a 50% immediate reduction in the risk of getting shellfish poisoning when eating mussels, would you be willing to pay ______ Euro per person per year?
75
E.10. And would you be willing to pay _______ [alternative bid] Euro per person each year? Willingness to Pay – Bivariate probit without covariates E.10. WTP
Mean
Median
St. deviation
Min / Max
Finland N=185
34.11*)
34.11
4.61
25.07 / 43.14
Italy N=203
25.63
25.63
2.60
20.53 / 30.72
France N=193
7.97
7.97
3.98
0.17 / 15.77
Ireland N=199
29.21
29.21
2.88
23.57 / 34.85
*) With different model specification presented in section 6.2, the standard probit with six covariates when N = 139, the estimated Finnish WTP is 24.90 Euro. In the bivariate probit model, the responses to both (the first and the potential second bid) are taken into account in the WTP estimation.
E.11. [0 amount] Could you give me the main reason why you are not prepared to pay? Note: This was an open question and the classification of the answers was done later 1
We pay enough taxes
2
I do not trust the government (In Finland: Protests against EU and doubts about bureaucracy)
3
It is not worth this much
4
Polluters should pay
5
Others should pay
6
I do not believe in the scenario (Reduction described in the scenario is not possible)
99 DK/REF E.11. Reason0WTP (%) 1
2
3
4
5
6
99
Finland N=64
18.75
6.25
31.25
28.13
6.25
9.37
0
Italy N=68
29.58
11.27
35.21
14.08
1.41
2.82
5.63
France N=106
28.30
6.60
39.62
20.75
3.77
0
0.94
Ireland N=62
19.35
0
59.68
3.23
9.68
3.23
4.84
E.13. Contribution aside, with these improvements would you visit [Location name] more often? E.13. Extra trips (%) 1 = Yes
0 = No
99 = DK/REF
Finland N=196
15.30
77.56
7.14
Italy N=203
22.17
77.83
0
France N=193
2.59
96.89
0.52
Ireland N=199
16.08
55.78
28.14
76
Please specify how many visits you would add next summer and how many days each visit would last. (Not available.)
E.14. Contribution aside, with these improvements would you eat mussels (in Finnish version: mussels imported from the EU) more often? E.14. Extra mussels (%) 1 = Yes
0 = No
99 = DK/REF
Finland N=196
80.61
14.29
5.10
Italy N=203
31.53
68.47
0
France N=193
4.15
94.82
1.03
Ireland N=199
31.16
65.83
3.02
Mussel eating habits for the Finnish respondents who would not eat any extra mussels after water quality improvement: Mussel eating (%)
Not at all
Seldom
Yes
Often
NA
Finland N=158
45.57
5.70
5.70
1.27
43.67
Please specify how many meals containing mussels you would add in the next 12 months. (Not available.)
E.17. If instead the algal blooms continued to increase until it is not possible to swim in the first 50/100 meters of water from the shore, and these algae were there every day, would you still come visit [Location name]? (Not available.) 1. Yes and as often as I am doing now 2. Yes but less often than I am doing now 3. No, I would go somewhere else 4. No, I would stay home
E.18. Thinking of your summer vacation, what would it be the next best substitute for [Location name] for your summer vacation? (Not available.)
77
SECTION F: Socio-demographic
F.0. Gender F.0. Gender
0 = Male
1 = Female
Finland N=196
50.00
50.00
Italy N=203
33.00
67.00
France N=193
47.15
52.87
Ireland N=199
32.16
67.84
F.1. In what year were you born? F.1. Age
Mean
Median
St. deviation
Min / Max
Finland N=196
42.91
42.00
10.87
20.00 / 71.00
Italy N=202
43.07
42.00
13.56
19.00 / 76.00
France N=191
47.72
48.00
13.87
18.00 / 90.00
Ireland N=198
36.08
29.00
15.15
18.00 / 75.00
F.2. How many members does your household have including yourself? F.2. Household size
Mean
Median
St. deviation
Min / Max
Finland N=196
2.87
3.00
1.32
1.00 / 7.00
Italy N=203
3.11
3.00
1.18
1.00 / 8.00
France N=193
2.97
3.00
1.42
1.00 / 8.00
Ireland N=199
3.70
4.00
1.84
1.00 / 10.0
F.3. How many of your children are living at home? F.3. Children
Mean
Median
St. deviation
Min / Max
Finland N=196
0.99
1.00
1.15
0.00 / 5.00
Italy N=203
1.04
1.00
1.11
0.00 / 6.00
France N=193
0.87
0.00
1.13
0.00 / 5.00
Ireland N=199
0.76
0.00
1.24
0.00 / 5.00
78
F.4. What is the highest educational degree you completed or how many years did you study? F.4. Education
Mean
Median
St. deviation
Min / Max
Finland N=195
13.43
14.00
2.86
6.00 / 20.00
Italy N=203
12.35
13.00
3.44
5.00 / 19.00
France N=188
13.23
13.00
3.15
5.00 / 18.00
Ireland N=198
14.39
15.00
2.89
3.00 / 20.50
F.5. What was your major field of study? (Not available.) F.5.1. Are you self-employed or not working? (Not available.) F.6. This is the final question. I am going to show you different income categories. Could you check the category that best represents the typical net monthly income of your household? F.6. Income per capita *) Mean
Median
St. deviation
Min / Max
Finland N=196
1276.87
1125
661.72
125 / 5250
Italy N=203
836.14
687.50
581.44
50 / 5250
France N=193
929.68
875.00
526.45
35.71 / 3250
Ireland N=199
872.71
562.50
854.82
31.25 / 4750
*) Unknowns are replaced with sample mean.
INCOME CATEGORIES: What is the typical net yearly income of your household? O
Less than 500 Euro
O
500 Euro less than 1000 Euro
O
1000 Euro less than 1500 Euro
O
1500 Euro less than 2000 Euro
O
2000 Euro less than 2500 Euro
O
2500 Euro less than 3000 Euro
O
3000 Euro less than 3500 Euro
O
3500 Euro less than 4000 Euro
O
4000 Euro less than 4500 Euro
O
4500 Euro less than 5000 Euro
O
5000 or MORE
79
PART III Interview number ______________
This is the end of the questionnaire. Thank you very much for your cooperation and if you want to see the results of this study you will find them on the web site listed on the business card I gave you. Have a nice day!
SECTION G: Evaluation of questionnaire
Control questions to be answered by the interviewer (BEFORE NEXT INTERVIEW) G.1. According to you did the interview pass well? YES --- NO If No Answer G.2. G.2. Why did the interview not pass well? G.3. Which parts are not (well) understood by the respondent?
Water Quality checks: Algae blooms: ____________________________________ Water transparency: ______________________________ Sunny --- cloudy --- rainy
General information: End of the interview: ______________________________ Time started: ________________ Date (day/month/year): _______________ City: _______________________ Country: ___________________________ Name and address of interviewing facility:
Special notes on the interview:
80
Appendix 2. Parametric model The construction of the parametric model is described below. Five groups of potential independent variables are discussed in table A2-1, and six different model candidates are compared in table A2-2. Table A2-3 presents the WTP calculations based on three different models. The computer programs used in the analysis were SPSS 10.0 and LIMDEP 7.0.
Table A2-1. The potential independent variables influencing the choice to accept or to reject the proposed bid. (to be continued)
Variable
Comments
1. Budget constraint monthly net income per capita
- no statistically significant influence on the choice even when divided into 3 classes of 0 – 999 €, 1000 – 1999 € and 2000 – 2999 € - unknown monthly net household incomes replaced by sample mean - due to theoretic expectations should be included in the model
vacation budget per capita
- captures both the length of stay and travel costs of respondent - unfortunately
not
answered
by
all
of
196
respondents and thus, including the variable in the model reduces the amount of observations
2. Experience on HABs
81
non toxic blooms (HBNT),
- none of these variables has statistically significant
toxic blooms (HBT or STB),
influence when included in the model
either of these HAB types
3. Opinions on seawater quality coastal problems
- based on the ranking of 4 coastal problems into the importance order and thus, can be expressed by a dummy (1 = algal blooms are the most important coastal problem) or by letting the ranking vary from 1 to 4 - including a dummy in the model spoils statistical significance of constant - unfortunately
not
answered
by
all
of
196
respondents and thus, including the variable in the model reduces the amount of observations level of satisfaction with seawater of interview day
- captures respondent’s experience on HABs because while answering s/he likely compares the quality of the day to his / her previous experience on seawater quality - to be ‘very satisfied’ today does not necessarily mean that the respondent is not concerned about seawater quality
4. Recreational profile regular visitor
- no statistically significant influence on the choice, no matter whether using respondents’ definition (they understood the word regular in the different ways) or researcher’s definition based on the questions B.3., B.5., B.9. – B.14. (see Appendix 1) - the idea of regular visitor can be captured in the variable vacation budget or a dummy for owning property in Hanko
82
accommodation (freetime house / cottage owner, boater, caravan
- dummies for tourist types based on accommodation have no statistically significant influence - a dummy for owning either a freetime house in
owner, daytripper, staying in
Hanko or a boat reflects best the binding
camping area, visiting friends
relationship to Hanko as vacation location,
or relatives, staying in hotel
although some people come regularly to stay in
or pension)
pension or camping area, and boaters may visit other coastal cities as well (their vacation enjoyment depends crucially on seawater quality)
length of stay
- no statistically significant influence on the choice - the idea comes out in the variable vacation budget or a dummy for owning property in Hanko
Distance
- dummies for distance from the place of residence to Hanko less than 150 kilometres and less than 200 kilometres
have
no
statistically
significant
influence - dummies for living in the capital area (the distance from Hanko is 140 km) and on the coast have no statistically significant influence substitute sites
- dummies for spending vacation in Hanko last year and for coming back next year have no statistically significant influence
5. Socio-demographic Gender
- none of these variables has statistically significant
Age
influence on the choice when included in the
Children living at home
model
length of education
Out of these variables the best are chosen and the different model candidates including constant and combinations of the variables BID, BOUND, INCCAP,
83
COASTAL, EXPEND, SATISF and PROPERTY are compared below in table A2-2. Model D was presented in section 6.2.
Table A2-2. The model candidates.
N Log likelihood
Model
Model
Model
Model
Model
Model
A
B
C
D
E
F
196
167
139
139
165
139
-126.35
-102.83
-83.77
-81.99
-99.15
-82.31
Coefficient (p-value) 0.348
0.624
0.395
1.231
1.319
1.494
(0.231)
(0.104)
(0.379)
(0.055)
(0.017)
(0.008)
BID
-0.011
-0.015
-0.015
-0.015
-0.015
-0.015
(first bid proposed
(0.005)
(0.001)
(0.002)
(0.003)
(0.001)
(0.004)
BOUND
0.242
0.452
0.430
0.469
0.452
0.430
(1 = if first bid asked is
(0.216)
(0.037)
(0.074)
(0.056)
(0.041)
(0.072)
INCCAP
0.061
0.139
0.146
0.138
0.118
(varies 1 – 3)
(0.660)
(0.371)
(0.392)
(0.425)
(0.455)
COASTAL
-0.218
-0.231
-0.318
-0.302
-0.322
(varies 1 – 4)
(0.042)
(0.051)
(0.014)
(0.010)
(0.013)
EXPEND
0.001
0.001
0.001
(continuous)
(0.053)
(0.075)
(0.068)
constant
to respondent)
higher bound of cost range)
SATISF
-0.311
-0.297
-0.314
(varies 1 – 4)
(0.062)
(0.041)
(0.059)
PROPERTY
0.422
(1 = if owns freetime
(0.081)
house in Hanko or boat)
84
Model A is the very basic model with only constant, BID, BOUND and INCCAP as independent variables. All coefficients except the one of BID are not statistically significant.
In model B, the variable COASTAL is added and the number of observations reduces from 196 to 167. According to the likelihood ratio test
11
the parameter
COASTAL has influence on the dependent variable.
In model C, the variable EXPEND is added and the number of observations reduces from 167 to 139. According to the corresponding likelihood ratio test as performed for the parameter COASTAL, the parameter EXPEND has influence on the dependent variable.
In model D, adding the variable SATISF improves statistical significance of constant compared to model C. This is the model presented in table 6-4 (page 44).
In model E, replacing the variable EXPEND with the dummy PROPERTY leads to the increase in the number of observations from 139 to 165, and all p-values except the one of INCCAP decrease. But, the coefficient of INCCAP is not statistically significant in model D either. Both variables EXPEND and PROPERTY reflect the importance of Hanko as vacation location.
Model F makes the assumption that the budget constraint faced by the respondent when choosing to accept or to reject the proposed bid, is the vacation budget instead of the disposable income. This assumption, however, would make more sense if our payment vehicle had been an entrance fee instead of a common tax. In this model all the coefficients are statistically significantly different from zero.
In table A2-3 below, the WTP estimates are calculated from models D – F. Model D leads to most conservative WTP estimate compared to models E and F.
11
LR = -2 * [logL(Restricted) – logL(Unrestricted)] = -2 * [ -126.35 + 102.83 ] = 47.04 exceeds the 5% critical limit for the chi square distribution with df = 1 (3.84), and thus, we reject the null hypothesis that the parameter COASTAL is restricted to be zero. (Haab & McConnell 2002, 304)
85
The 95 % confidence intervals are calculated by the delta method (see Hanemann & Kanninen 1999, 335).
Table A2-3. WTPs and 95 % confidence intervals calculated from models D – F.
Model WTP (per person per year) 95 % confidence interval
Model D
Model E
Model F
24.90 €
32.37 €
40.10 €
0 – 136 €
0 – 129 €
0 – 146 €
86
Appendix 3. Interviews in Hanko in July 2003
HANKO
2.
4. 3. 1.
FINLAND
5.
Figure A3-1. Interview places in Hanko. The numbers refer to the text below. (Hanko 2003)
1. Eastern Harbor Eastern Harbor is situated on the southern coast of the Hanko city. During the summer 2002 (June, July, August) there was about 7800 boat visits, 650 of which was during one single sailing competition weekend. Each year about 6500 of boats come from Finland, and the next biggest group are Swedish, German, Dutch, Estonian, Russian, French and British boats. Every Wednesday and Friday evening during the summer there is a market on the harbor.
87
2. Silversand Camping Silversand Camping – the only camping in Hanko – is situated on the northern coast of Hanko city, about 4 kilometres from the city centre. During the summer 2002, there were about 9000 visits, and the average amount of family members was 2,1. The most crowded times are weekends. Besides Finland, people are mostly from Germany, Holland and Sweden. There are caravans (staying either for a few nights or for the whole summer), tents and rental cottages.
3. Plagen Beach Plagen beach is situated on the southern coast of Hanko. It is rather small but the only beach in Hanko with a café, and probably the most popular beach. The distance from the city centre is 900 meters and the beach is about 400 meters long.
4. Regatta Beach Regatta Beach is situated on the southern coast between the Eastern Harbor and Plagen Beach. It is a ‘sight seeing beach’: in addition to lie on the beach, people walk by, eat ice cream at the ice cream café and sit on the benches watching the sea. Regatta Beach is in the city centre and about 250 meters long.
5. Bellevue Beach Bellevue Beach is the most natural and the quietest of the beaches where interviews took place. The distance from the city centre is about 1200 meters and the beach is about 1000 meters long.
Interview calendar
The interview calendar in table A3-1. presents 19 interview days in July 2003, start and end times of interviewing, the number of completed interviews per day and the places where the interviews took place.
88
Table A3-1. Interview calendar.
DATE July 1 Tu July 2 We July 3 Th
START TIME 12.07 14.00 11.08
END TIME 13.35 18.14 19.30
July 4 Fr
11.12
16.07
7
July 9 We
11.08
18.10
14
July 10 Th
12.58
19.20
10
July 11 Fr
10.49
19.38
13
July 13 Su July 14 Mo
13.12 10.43
15.16 18.47
5 15
July 15 Tu July 16 We
9.41 10.25
19.25 21.03
19 13
July 17 Th
11.41
20.33
12
July 21 Mo July 22 Tu
16.28 14.43
17.31 20.27
2 12
July 23 We
12.22
21.50
19
July 26 Sa July 27 Su July 28 Mo
10.53 16.58 13.15
21.01 19.23 19.08
19 4 6
July 29 Tu
11.26
13.31
3
89
NUMBER 2 8 13
PLACES Harbor (2) Harbor (8) Harbor (4) Regatta (5) Silversand (3) Bellevue (1) Bellevue (3) Plagen (4) Plagen (9) Harbor (5) Harbor (7) Regatta (1) Plagen (2) Harbor (2) Silversand (4) Regatta (3) Plagen (4) Plagen (5) Plagen (11) Bellevue (4) Silversand (19) Harbor (9) Regatta (4) Plagen (10) Harbor (2) Bellevue (2) Regatta (3) Harbor (9) Silversand (11) Harbor (8) Silversand (19) Harbor (4) Plagen (5) Harbor (1) Non city beaches (3)
Appendix 4. Occurrence of harmful algal blooms in Hanko in July 2003
The Technical and Environmental Office of the Hanko city monitors the blue green algae occurrence during the summer on ten beaches in Hanko. The beaches to be monitored include Plagen, Silversand and Bellevue Beaches.
The scale for the abundance of blue green algae is: 0
No blue green algae in the water. Normal transparency.
1
Some algae in the water. Algae can be seen when taking water in the glass. There might be some thin layers of algae on the beach. Transparency is worsened.
2
Abundant algae in the water. Lots of algae in the water or small algal blooms or algae accumulations have drifted to the beach.
3
Very abundant algae in the water. Algae form large blooms or thick algae is drifted to the beach.
The algae occurrence in Hanko in July 2003 12
During July 2003, no blue green algae were observed in Bellevue or Plagen. In Silversand there was algae in the scale of 1 on July 14 and 21.
12
Personal communication. Virpi Mikkanen, Environmental Secretary, the City of Hanko. August 30, 2003.
90
