Regional-scale non-market benefits of improved lakes and rivers when perceived and monitored ecological status diverge

Abstract

There is increased call to demonstrate the benefits of EU Member States’ river basin management plans, whose implementation has been delayed largely due to insufficient funding. This paper applies a contingent valuation study to investigate the benefit value of improved ecological status in river basins and the discrepancy between the quality of waters as perceived by river basin residents and as monitored under the Water Framework Directive. Respondents often reported worse quality for their focal water body than the monitored status indicated, a tendency established in a GIS analysis. The likelihood of such divergence increased most with degree of perceived deterioration of surface waters. Observed deterioration in waters, official status of one’s focal water body and divergence between two quality measures had implications for welfare estimates. Describing water quality according to ecological criteria and as uniformly as possible would facilitate the use of valuation results in future benefit transfers.

Keywords:

• water quality perceptions
• perceived and monitored ecological status
• economic valuation
• stated preferences
• environmental benefits
• EU water framework directive

1. Introduction

The environmental objective of the EU Water Framework Directive (WFD 2000/60/EC) is to achieve good ecological status (GES), defined in the Directive, in European freshwater bodies by 2027 at the latest. Thus far, the ecological status of surface waters has shown visible improvement in only 10% of water bodies (European Environment Agency 2018). For example, in Finland, monitoring of impacts on surface waters for the period from 2008 to 2020 indicates that diffuse pollution from agriculture and forestry continues to pose a significant risk where achieving GES is concerned (Vilmi et al. 2021).¹

Economic valuation provides a tool to assess the benefits to society of implementing the WFD’s river basin management plans (RBMP). The scientific literature features extensive research on the economic valuation of water quality improvements (see, e.g. the reviews by Boeuf and Fritsch (2016) as well as the meta-analyses, such as those by Newbold and Johnston (2020) and Brouwer and Pinto (2021). In the WFD context, monetary benefits have been estimated using stated preference methods (Hanley and Black 2006; Brouwer 2008; Del Saz-Salazar, Hernandez-Sancho, and Sala-Garrido 2009; Martin-Ortega 2012; Metcalfe et al. 2012; Ramajo-Hernández and del Saz-Salazar 2012; Soliño, Joyce, and Farizo 2013; Buckley et al. 2016; Pinto et al. 2016; Lazaridou and Michailidis 2020). This research notwithstanding, no common understanding exists about the factors underlying water quality preferences (Artell, Ahtiainen, and Pouta 2013; Jeon et al. 2005; Soliño, Joyce, and Farizo 2013; Whitehead 2006). What is more, we lack an adequate understanding of the divergence between perceived water quality and “actual” (monitored) quality, the latter being a classification based on ecological criteria. While the literature offers several economic valuation studies applying usability criteria to study such divergence (Poor et al. 2001; Jeon et al. 2005; Artell, Ahtiainen, and Pouta 2013; Ahtiainen, Pouta, and Artell 2015), one finds very few contributions that document differences between quality as perceived by residents and status as monitored under the WFD and go on to analyse the factors explaining the discrepancies (Artell and Huhtala 2017).

A basic assumption in stated preference studies is that the survey respondents are valuing the objective quality improvement presented in the survey. This may be not the case when respondents have varying levels of prior information – and thus subjective perceptions – about the change in quality (Whitehead 2006). Hence, little is known about the extent to which individuals have accurate perceptions of water quality (Jeon et al. 2005) or which factors related to the respondent, or the study site, cause perceptions to deviate from objective measures (Artell, Ahtiainen, and Pouta 2013). More information is needed not only on possible biases in quality perceptions but also on welfare changes in average water quality from low or high baseline levels (Newbold and Johnston, 2020). Subjective water quality perceptions have been shown to be correlated pairwise with physical water quality measures (Jeon et al. 2005; Artell, Ahtiainen, and Pouta 2013; Artell and Huhtala 2017), and especially with those measuring water clarity (Jeon et al. 2005). In revealed preference studies on recreation, individuals’ site choice decisions have depended on a variety of water quality measures (physical quality, quality index and quality perception) (Jeon et al. 2005). As noted by Deely, Hynes, and Curtis (2019), the literature seems to favour models based on subjective data, and the inclusion of subjective data in revealed preference studies has improved the fit in models explaining choice of recreation site (Adamowicz et al. 1997; Jeon et al. 2005; Deely, Hynes, and Curtis 2019).

Ecological status is one of the two main components of surface water status, the other being chemical status. In the European Union, countries’ surface waters are currently classified, using primal biological factors, as high, good, moderate, poor or bad ecologically. The classification describes how human activity affects the health of the aquatic environment; that is, biological organisms and their functioning (Heiskanen et al. 2004). By contrast, people’s perception of water quality is often based on how they use waters, and can thus be more positive or negative than the scientific classification (Artell, Ahtiainen, and Pouta 2013; Kataria et al. 2012). Most of the earlier water-related valuation studies have described water quality primarily in terms of usability (Van Houtven, Powers, and Pattanayak 2007), and stated preference surveys have used vague ecological descriptors in characterising ecosystem change (Johnston et al. 2012).

We designed and conducted a contingent valuation (CV) study to estimate the aggregate welfare gains to be realised by the residents of a river basin district from achieving GES as defined in the WFD. Specifically, we analyse the factors underlying water quality preferences in the Vuoksi river basin district, where the quality of surface waters differs along the length of the water course. The study addresses the following questions:

1. Do the residents of the Vuoksi river basin district perceive surface water quality differently to the classification for WFD implementation (i.e. ecological status)? If they do, what are the determinants of this divergence?

2. What are the non-market benefits of achieving good ecological status in surface waters in the Vuoksi river basin district and which factors, particularly those related to water quality, are associated with the benefits?

Section 2 proceeds to detail the contingent valuation (CV) method and its application, the study area and econometric models. Section 3 describes our main results and goes on to discuss them in the light of our research questions. Section 4 concludes the paper with a brief consideration of the applicability of our approaches in broader contexts.

2. Methods and data

Over the past 40 years, the contingent valuation (CV) method has been the most widely applied stated preference approach for assigning a monetary value to changes in environmental quality or quantity. Environmental change is a non-market good, that is, one whose value is not captured in the market system. Stated preference methods allow estimation of both use and non-use values of the outcome of a certain programme or policy. They differ from other economic valuation approaches (i.e. revealed preference methods), which utilise market information indirectly to elicit the use value of environmental change (Bateman et al. 2002; Perman et al. 2011; Freeman, Herriges, and Kling 2014). CV applications rely on a carefully structured survey questionnaire that presents respondents with a future scenario related to a programme or policy. Using either direct questions or choice options, the questionnaire elicits the amount each respondent would be willing to pay (WTP) or, if more appropriate, the amount of compensation he/she would find acceptable (WTA) for a given change. The theoretical underpinnings of the CV method lie in microeconomic theory.

2.1. Study area

The basin of the Vuoksi River (hereinafter “Vuoksi”), encompassing 1,182 lake and 328 river water bodies, is the largest of the eight river basin districts in Finland and the one with the most freshwater bodies. It has a total land area of 58,200 km² and a population of 512,600 people (2016). Some 84% of the land area in Vuoksi is used for forestry; approximately 8% is arable land; and about 4% is built. Despite significant pressures from agriculture, forestry, sparsely populated areas and several point sources, most of the water bodies (84% of the lakes and 70% of the rivers) are in good ecological status (Manninen and Kotanen 2016). While the majority of surface waters in both southern Vuoksi (located in the provinces of Southern Savonia and South-Eastern Finland) and northern Vuoksi (located in the provinces of Northern Savonia and Northern Karelia) are in high or good ecological status, the proportion of water bodies in high status is larger in southern than in northern Vuoksi (37% and 18%, respectively) (Figure 1). It is estimated in the river basin management plan that, despite the measures taken, some 65 lakes and 28 rivers in Vuoksi will still not have reached good ecological status in 2021. Plans called for restoration of a total of around 130 surface water bodies between 2016 and 2021 (Manninen and Kotanen, 2016).

Figure 1. Vuoksi River Basin District and the ecological classification of its freshwater water bodies (n = 1,530). The border between southern and northern Vuoksi is shown on the map with a bold black line.

2.2. Application

Before the data collection proper, the survey design and valuation scenario, as well as the electronic questionnaire, were pre-tested over the course of a month using interviews and reviews by several residents, water authorities and valuation experts. The cover letter, signed by the Director General of the Centre for Economic Development, Transport and the Environment, explained that the results would be utilised in the development of the Vuoksi River Basin Management Plan for 2016–2021, to be submitted to the Government for consideration in 2015. The statements at the end of the survey were used to assess consequentiality, that is, whether the respondents accepted the opportunity to contribute to water management and whether they found the description of the hypothetical Vuoksi Water Management Foundation credible. Consequentiality is important in encouraging truthful preference revelation, and thus to obtain more reliable results the respondents were informed how the survey results would be used and disseminated (Vossler and Watson 2013; Johnston et al. 2017). At the beginning of the survey, the respondents stated the water body of highest importance for them in Vuoksi, provided a postal code for its location and indicated its approximate distance in kilometres from their residence. Before the respondents were given objective information on the ecological status, they indicated their perception of the water body most important for their use. The question was framed as follows: “How would you describe the status of this body of water (i.e. the lake, river, or stream that is the most important to you or your household)?”. The response options were: Excellent, Good, Moderate, Poor, Bad and Can’t say. After a general description of Vuoksi, the respondents were asked how they used surface waters in their area and whether they had noticed any changes in these waters over the last ten years.

To illustrate and specify changes in water status in Vuoksi, the written description was supplemented by drawings of differences between various states of a water body. Respondents were presented with a series of images of a typical lake and river in three different ecological states and a colour map of the ecological status of the river basins in Vuoksi. This combination of materials was inspired by the “water quality ladder” in Hime et al. (2009). Images were tested as part of testing the electronic survey. Overall, the image series was considered well done and illustrative. Text related to the drawings clarified how the bottom biota, fish, vegetation, algae abundance and water clarity vary in different ecological conditions. For example, the respondents were given the information that surface waters in high ecological status are clear but may be darkened by humus, and that restoration of a lake in moderate ecological status does not necessarily change the colour of the water or remove all aquatic plants (Figure 2). In addition, a colour map of the current ecological status for Vuoksi was provided. Respondents then stated whether they agreed with the given information about the ecological surface water status in general in Vuoksi, choosing from among the following possible responses: “much better than I thought”, “somewhat better than I thought”, “as I thought”, “somewhat worse than I thought”, “much worse than I thought” and “difficult to evaluate”. Respondents then indicated which measures they considered the most important in Vuoksi; examples being a decrease in diffuse loading from agriculture and protection of the natural migration of brown trout. Information was provided on anticipated changes in water management as described in the river basin management plans at the time of the survey. Respondents were also asked what their reactions would be if no water management measures were implemented in the river basins.

Figure 2. Sample description used on the questionnaire.

WTP estimates were determined in two stages (Figure 3): a screening question was presented to determine which respondents were willing to pay a positive water management fee, followed by a second item in which those willing to pay indicated the size of the payment. Individual WTP for an improvement of surface waters through implementation of the river basin management plan was elicited using a “non-referendum” variation of the multiple bounded discrete choice (MBDC) format (Welsh and Bishop 1993; Welsh and Poe 1998; Alberini, Boyle, and Welsh 2003; Broberg and Brännlund 2008). Respondents then stated some amount from a range of payments² to be made to a hypothetical Vuoksi Water Management Foundation to achieve the status of “good” in all waters in the area. Following Welsh and Poe (1998), a preference uncertainty matrix was constructed presenting the payments and the following categories of certainty: “I would certainly pay”, “I’d almost certainly pay”, “I’m not sure I would pay”, “I’m pretty sure I wouldn’t pay” and “I certainly wouldn’t pay”. The payment vehicle chosen was a monthly water management fee paid over the Foundation’s first six-year term, the duration of the planning cycle. Following Johnston et al. (2017) such a non-binding payment vehicle – although the most realistic and familiar – might prompt respondents to adopt strategic behaviour such as free riding (Johnston et al. 2017), and we acknowledge that the vehicle chosen here might result in underestimation of values.³ Therefore, respondents were told of the need for additional funding as follows: “The state would contribute up to 40% of the costs of water management; private beneficiaries and operators would have to provide the remaining 60%, contributing about 30% each. Beneficiaries would include local residents, businesses, communities and fishing districts.” After the valuation questions, follow-up questions were presented that made it possible to distinguish protest responses from legitimate zero responses.⁴ Finally, respondents were asked for socio-demographic and other background information.

Figure 3. Willingness to pay questions.

2.3. Sampling and protest responses

A random sample of 1,495 residents from across Vuoksi received the questionnaire, with addresses obtained from the Finnish Population Register Centre in 2014. First, all the residents in the sample were sent a signed postcard with an invitation to reply to an online questionnaire via an internet link or a QR code. After ten days, respondents received a second postcard with a reminder and the link to the questionnaire; and finally each non-respondent received a signed cover letter and a 12-page mail questionnaire in booklet form. Respondents were given the option of replying by traditional post if they preferred not to answer via the internet; 40% chose to do so. Ultimately, 333 individuals responded in sufficient detail to be included in the initial analysis (response rate 22.3%). A total of 91 of the 333 responses (27%) received were excluded as zero-protest responses, yielding a final sample of 242 respondents. By way of comparison, the proportion of zero-protest responses among those not willing to pay (69/175, 39%) falls toward the lower end of the results in the meta-analysis by Meyerhoff, Morkbak, and Olsen (2014). In keeping with common practice, follow-up questions were used to elicit motivations for zero bids and to distinguish valid zero responses from zero protest bids. Among the identified protests, the most common reasons for a reluctance to pay were the following: “I think surface water polluters should pay the costs if the causes of the harm are known” (68/91), “I think society should be able to finance the costs of surface water management” (43/91), and “The tax funds I have already paid should be directed to surface water management” (37/91). Hence, and as in many CV studies, a significant proportion of respondents were not willing to pay, although they might assign a positive value to improved surface water bodies (Hanley, Wright, and Alvarez-Farizo 2006). The group of protest respondents had a higher proportion of men and people over 60 years of age (Table 1). An outlier analysis of possible “excessively high” WTP values showed that none of the respondents’ annual WTP exceeded 3% of their annual income. According to an ex-post mail survey sent to 300 non-respondents, no non-response bias was found between the non-respondents and actual data in terms of gender, age and use of water bodies. The ex-post survey contained seven questions, six of which were identical to those in the survey proper (response rate was about 25%). Residents mostly justified their non-response in terms such as the following: i) “I don’t usually answer surveys”, ii) “The survey seemed too tedious to answer”, and iii) “I didn’t have time to answer”. Furthermore, only 12% of those who responded to the non-respondent check justified their non-response by saying that they felt that improving the condition of surface waters was unimportant. The sample is a rather good representation of the population in terms of gender distribution and income, but it has a smaller proportion of people under 40 years of age than the population at large (Table 1). In an additional finding, southern and northern Vuoksi did not differ from each other statistically significantly for any other characteristic than individual income when analysed using independent sample t-tests. Respondents in southern Vuoksi had a higher individual income compared to respondents in northern Vuoksi (t = 1.655, p = 0.099).

Table 1. Sample, zero-protest responses, and population characteristics.

Variables	Sample	Zero-protest	River basin district^a
N	242	91	507,913
Age distribution
18–39 years	17.4%	13.2%	29.4%
40–59 years	38.4%	33.0%	21.8%
≥60 years	44.2%	53.8%	48.8%
Gender distribution (women)	49.6%	33.0%	51.0%
Household size (head count)	1.8	1.9
Households with children <18 years old	21.1%	12.1%
Monthly individual income (€)	3,170	3,145	3,264

Note: ^aStatistics Finland, last accessed 11 October 2021.

2.4. Statistical analyses

A respondent’s mean WTP estimate was expected to lie at the midpoint of the interval between the highest payment chosen and the next higher value on the payment card.⁵ These estimates were calculated using the non-parametric maximum likelihood (ML) Turnbull (1976) estimator (Haab and McConnell 1997) and non-parametric Kriström (1990) partly linear welfare measure. These estimators are robust against misspecification of the response probability distribution and are well suited for our data because it is rare for respondents to state a higher level of certainty for a higher payment (Vossler (2003), see the matrix in Figure 3). The Turnbull estimator for interval-censored data is calculated as follows: (1) $$E{(\mathit{\text{WTP}})}_T=\sum _{b=1}^B{l}_b(Y_b-Y_{b+1}),$$(1) where B is the total number of bids, l_b is the level of bid b and Y is the proportion of respondents who chose bid level l_b. Kriström’s (1990) welfare measure is based on Ayer et al.’s (1955) theorem and estimated as follows: (2) $$\mathit{\text{Mean}}\hspace{0.17em}\mathrm{}\mathit{\text{WTP}}=\sum _{j=0}^{B}0.5\left(b_j+b_{j+1}\right)\left[\mathrm{}F\left(b_j\right)-F\left(b_{j+1}\right)\right],$$(2) where B is the total number of bids, b_j is the given bid at j in the bid vector, and F(b_j) is the value of the probability density function at bid j. The mean WTP estimates were calculated treating “certainly yes” and “almost certainly yes” responses as “yes” (the highest WTP chosen by the respondent) and all other responses as “no” (zero WTP).

An ordered probit model was used to analyse which variables were associated with a higher WTP, using NLogit 6.0 software. Regarding independent variables, the model included those variables that were individually correlated with the response to the WTP question and not highly correlated with each other. The econometric analysis is supplemented with a comparison of the survey data on perceived water quality and monitored data on spatial ecological status using an additional ex-post Geographical Information System (GIS) analysis of the total of 200 focal water bodies mentioned by the respondents. The monitored, or “real”, ecological status of each focal water body was then retrieved from the environmental database using the name of the water body and the postal code given by the respondent. Chi-square and t-tests were performed using SPSS 27 to estimate WTP differences between respondents in southern and northern Vuoksi.

3. Results and discussion

3.1. Perceived water quality vs ecological status of the main body of water

According to the descriptive statistics (Table 2), the respondents were familiar with the surface water resources, as nearly everyone (99%) used the region’s freshwater bodies for recreation. Respondents’ homes were located quite near their focal body of water, as the average distance between the two was 16.7 kilometres (DIST), and about 45% of the respondents lived on a shore. Recreational uses include swimming, admiring the water landscape, boating, outdoor activities, fishing, sailing or spending time in one’s holiday home. Almost all respondents (93%) were able to provide an assessment of the current state of their focal lake/river (on average between moderate and good; QUALITY). The perceived water quality of this body of water was often lower than the ecological status when monitored by experts (on average between good and high; STATUS, Table 2, Figure 4). According to the GIS analysis, the monitored status of respondents’ focal body of water was better in southern than in northern Vuoksi, confirming the visual difference shown in Figure 1. Of the 200 respondents, 119 perceived the water quality as being worse than the classified status indicated. Only a small proportion of respondents in southern Vuoksi (4%) found their main body of water to be in excellent status, although a total of 64% of the water bodies in the area were in high ecological status at the time. The corresponding percentages for northern Vuoksi are 3% and 16%. Hence, the perceived water quality corresponded with the monitored ecological status for only 32% of the respondents, confirming a previously observed discrepancy between public observations and monitored ecological status in the medium-sized or large clear-water lakes of Eastern Finland (Vuori and Korjonen-Kuusipuro, 2018).

Figure 4. The monitored ecological status classification (an ex-post GIS analysis, darker bars) and the perceived water quality (from the survey results, lighter bars) of the respondents’ focal body of water (n = 200).

Table 2. Descriptive statistics.

Variable descriptions		Mean (Stand. Dev.)	Expected effect on WTP
INCOME	Ten classes (1= EUR 800, …10= EUR 8,300 per month)	4.06 (2.59)	+
SOUTHERN	Lives in southern Vuoksi (yes = 1, no = 0)	0.30 (0.46)	?
USE OF WATERS	Number of different water uses 0–10	5.04 (2.28)	+
DETE	Has observed changes in the waters of the region for the worse over the last ten years; (yes = 1, no = 0)	0.40 (0.49)	+
HUMUS	Opinion that humus, peat, darkness, or colour affect the current state of focal water body (yes = 1, no = 0)	0.07 (0.26)	?
EUTRO	Opinion that eutrophication (e.g. algal blooms) affects the current state of focal water body (yes = 1, no = 0)	0.15 (0.36)	?
DIST	Distance between the respondent’s residence and focal water body: 0–30 m (1); 50–200 m (2); 300 m–1 km (3); 1.5–23 km (4); 25–700 km (5)	3.11 (1.26)	−
SCENARIO	Degree of accepting the statement ‘The Vuoksi Water Management Foundation was plausible’: no (1); partly (2); totally (3)	2.35 (0.66)	+
QUALITY	Opinion on the current status of focal body of water: bad (0); poor (1); moderate (2); good (3); excellent (4)	2.18 (0.96)	−
STATUS	Ecological status of focal body of water: moderate (1); good (2); high (3) (ex-post GIS analysis)	2.11 (0.70)	−
DIVERG	For the focal water body, difference between perceived water quality and monitored ecological status (yes = 1, no = 0)	0.62 (0.49)	?
ASIS	Ecological status of surface waters in Vuoksi (shown in a map) is as I thought (yes = 1, no = 0)	0.39 (0.49)	?

The majority of respondents reported that their focal water body was in a different (and most often worse) status than the official classification indicated in an ex-post GIS analysis (DIVERG, Table 2). The observed and classified status differed by slightly less than one status class, with the former often indicating worse water quality than the latter. The divergence may result from observed changes in the frequency of algal blooms, shoreline sedimentation, surface water colour or clarity (browning) and water level. For example, the following themes were repeated in the responses: “We have algal blooms on the shore”, “The brook brings nutrients and humus into the lake from the forest/mire ditches upstream.”, “The water is dark, and your fishing gear gets sticky quickly”, and “There is a large variation in the level of the water in the lake”. These phrases suggest that describing GES to respondents through drawings with texts and a map in the valuation survey was of paramount importance. However, the respondents’ perceived water quality correlated positively with the monitored status of their focal water body, a finding in keeping with studies by Jeon et al. (2005), Artell, Ahtiainen, and Pouta (2013), and Artell and Huhtala (2017).

3.2. Factors accounting for the divergence between perceived water quality and classified ecological status

Table 3 presents the results of the probit model on a predicted probability of divergence between perceived water quality and monitored ecological status of the main water body. The likelihood of a discrepancy was associated with observed overall deterioration of the surface waters over a period of 10 years (DETE) as well as observed surface water eutrophication (EUTRO) and browning (HUMUS) in respondents’ focal water body. In addition, living in southern Vuoksi (SOUTHERN) and engaging in surface water-related activities (USE) increased the likelihood of there being a difference. The negative coefficient of the WTP variable indicated a lower WTP among those respondents who perceived the quality of their focal lake/river as different to – and most often worse than – its monitored status. However, this variable was not statistically significant in the model. Finally, and as expected, a divergence between perceived water quality and ecological status was more likely if respondents felt that the presented ecological status of all surface waters was different from their prior perception (ASIS).

Table 3. Probit model for predicted probability of divergence between perceived and classified status.

Variable		Coefficient	z	Sig.
		(Std. Error)	statistic
Constant		−0.544 (0.420)	−1.30	0.195
DETE	Has observed changes in water bodies for the worse over the last ten years; dummy (yes = 1, no = 0)	0.828 (0.228)***	3.63	0.000
SOUTHERN	Lives in southern Vuoksi (yes = 1, no = 0)	0.555 (0.232)**	2.39	0.017
HUMUS	Opinion that humus, peat, darkness, or colour affects the current state of focal water body; dummy (yes = 1, no = 0)	0.839 (0.494)*	1.70	0.089
WTP	Willingness to make a positive payment (yes = 1, no = 0)	−0.140 (0.213)	−0.66	0.510
DIST	Distance between the respondent’s residence and focal water body: 0–30 m (1); 50–200 m (2); 300 m–1 km (3); 1.5–23 km (4); 25–700 km (5)	0.012 (0.086)	0.13	0.893
EUTRO	Opinion that eutrophication (e.g. algal blooms) affects the current state of focal water body; dummy (yes = 1, no = 0)	0.676 (0.353)*	1.92	0.055
USE OF WATERS	Number of different surface water related activities 0–10	0.097 (0.052)*	1.87	0.062
ASIS	Ecological status of surface waters in Vuoksi (shown in a map) is as I thought the situation was (yes = 1, no = 0)	−0.353 (0.208)*	−1.70	0.089
N		200
Chi squared [8 d.f.]		47,108***		0.000
Restricted log likelihood		−131,248
McFadden Pseudo R-squared		0.179

Note: The dependent variable is 1 for a difference (of one or two status class sizes), and 0 for no difference between perceived and classified status.

*** p < 0.001; **p < 0.05; *p < 0.1.

Southern Vuoksi contains numerous water bodies which monitoring has classified as being in high ecological status. It may be that these surface waters are evolving towards a more eutrophic state, observed by residents but not yet captured in the water bodies’ monitored status. The results are also plausible inasmuch as surface water users are more likely to have a better view of the current state than others. According to Vilmi et al. (2021), monitoring results from the period 2008 to 2020 indicate that the ecological status of surface waters in Finland has clearly declined in forestry-dominated areas. Diffuse pollution leads to surface water eutrophication and browning, and deterioration of ecological status (Vilmi et al. 2021). Surface water browning (i.e. brownification) is a trend documented in boreal water bodies over a decade (de Wit et al. 2016; Kritzberg et al. 2020; Nieminen et al. 2021). Promisingly, research has shown that more natural management of peatlands could reduce such undesirable impacts on water quality (Härkönen et al. 2023).

3.3. Respondents’ views on the state of freshwater bodies in Vuoksi

About 40% of respondents had noticed changes for the worse (DETE) due to eutrophication, agricultural loading, changes in lake water levels or pollution from peat extraction, among other anthropogenic activities (Table 2). On the other hand, some 14% had observed improvements in surface waters due to measures such as improved wastewater treatment and water monitoring. The given information about the ecological status of freshwater bodies differed to a great extent from the respondent’s previous opinion for only some 5% of the respondents, and most often (39%) the description of ecological status reflected the respondent’s previous perception (ASIS). For about one-third (35%) of the respondents, the monitored ecological status of water bodies was better than their previous perceptions indicated. For some 17%, the monitored status was worse than they had previously thought. The map and description of the ecological status of Vuoksi most often conveyed a view that was similar to, or better than, the respondent’s own perception. In the light of previous results, it is clear that the large number of water bodies in high ecological status surprised respondents.

3.4. Value estimates and explanatory factors

Of all the respondents, fewer than half (42%) stated a zero willingness to pay. The majority (71%) of payers used several certainty categories when conveying their maximum WTP (i.e. “I would certainly pay”/“I’d almost certainly pay”/“I’m not sure I would pay”). The non-parametric mean WTP (€46.70–€72.80) differed between respondents in southern (€69–€107.10) and northern (€36.90–€57.50) Vuoksi, being statistically significantly higher among southern respondents (Pearson Chi-Square 13.863, p = 0.054, Figure 5). This result was unexpected since, compared to northern Vuoksi, the respondents’ focal water body in southern Vuoksi often had a better ecological status classification (Figure 1, t = 3.874, p = 0.000). Generally, WTP is assumed to be higher the greater the change in status. However, no difference was found in perceived quality of main surface waters between the two regions (see the right-hand column in Figure 3, t = 0.391, p = 0.696), and southern respondents had, on average, a higher monthly income than northern respondents, as noted in section 2.3. Furthermore, earlier studies have shown that people are willing to pay greater amounts for quality improvements in water bodies that begin at higher baseline quality levels (Alvarez and Asci 2014; Brouwer and Pinto 2021; Hynes and O’Donoghue 2020; Johnston and Thomassin 2010; Soliño, Joyce, and Farizo 2013). The mean WTP estimates are similar to earlier results by Buckley et al. (2016) and Del Saz-Salazar, Hernandez-Sancho, and Sala-Garrido (2009) but smaller compared to estimates by Brouwer (2008), Metcalfe et al. (2012) and Soliño, Joyce, and Farizo (2013).

Figure 5. Left: Classified distribution of mean WTP (2014) in southern and northern Vuoksi (n = 73 and n = 165; N = 238); Right: Non-parametric estimates for mean WTP values, standard deviation, and standard errors.

The ordinal probit model (Table 4) was applied to analyse how perceived and monitored ecological status, other respondent- and site-specific features and environmental characteristics were associated with WTP; the dependent variable is expressed in terms of four levels (0: 0€, 1: 8–34€, 2:68–136€, 3: 272–543€). As expected, based on economic theory and prior findings in the literature, the increase in WTP was associated to the highest degree with considering the Vuoksi Water Management Foundation and its work plausible (SCENARIO), perceived ability to pay for water quality changes (monthly INCOME) and observed worsening of the surface water quality in the region over the last ten years (DETE). According to the results of Pinto et al. (2016), younger residents perceived more often than older residents that the river basin had deteriorated, which could explain their higher WTP for improved ecological status. Increased WTP was also associated with the use of surface waters (USE OF WATERS) and having one’s focal water body in better ecological status (STATUS). The STATUS factor shows that respondents have, at least to some extent, assimilated the given information about the ecological status of river basins, even though the factor was evaluated with ex-post GIS analysis. Finally, for a focal water body, difference between perceived water quality and monitored ecological status decreased the mean WTP (DIVERG). The marginal effects of the explanatory factors are of a realistic magnitude. Other site-specific factors were not associated with the probability of a respondent choosing higher payments; these being living in southern as opposed to northern Vuoksi and distance between the respondent’s main residence and focal water body.

Table 4. Ordered probit model on the level (0–3) of mean WTP for improved ecological status in Vuoksi (marginal effects).

Variable descriptions		Coefficient	z	p
		(Standard Error)	statistics
Constant		−3.574 0.580)***	−6.21	0.000
INCOME	Ten classes: EUR 800 to over EUR 8,300	0.119 (0.034)***	3.51	0.000
SCENARIO	Degree of accepting the statement “The Vuoksi Water Management Foundation was plausible”: no (0); partly (1); totally (2)	0.840 (0.141)***	5.96	0.000
USE OF WATERS	Number of different surface water related activities 0–10	0.111 (0.045)**	2.48	0.031
DETE	Has observed changes in the surface waters of Vuoksi for the worse over the last ten years dummy (yes = 1, no = 0)	0.532 (0.187)***	2.84	0.005
SOUTHERN	Lives in southern Vuoksi (yes = 1, no = 0)	0.233 (0.190)	1.23	0.220
DIST	Distance between the respondent’s residence and focal water body: 0–30 m (1); 50–200 m (2); 300 m–1 km (3); 1.5–23 km (4); 25–700 km (5)	−0.075 (0.070)	−1.08	0.280
DIVERG	For the focal water body, difference between perceived water quality and monitored ecological status (yes = 1, no = 0)	−0.350 (0.210)*	−1.67	0.096
STATUS	Ecological status of focal water body: moderate (1); good (2); high (3) (ex-post GIS analysis)	0.311 (0.126)**	2.46	0.014
	N	200
	Chi squared [8 d.f.]	87.912***
	−2 Log Likelihood	−190.999
	McFadden Pseudo R-squared	0.187

Note: ***p < 0.001; **p < 0.05; *p < 0.1.

Benefit-cost ratios can be calculated to find out to what extent the additional costs of WFD implementation outweigh the benefits of achieving the Directive’s objectives (i.e. a disproportionate cost analysis). The aggregated value estimate is in the range MEUR 23.9 to MEUR 37.3 annually for 512,613 (2016) adult residents in Vuoksi. Thus, improving surface water quality to good ecological status in Vuoksi would yield present non-market benefits of MEUR 134–209 over a six-year period, applying a 2.0% social discount rate.

4. Conclusion

The research contributes to the literature on documented differences between perceived water quality and monitored ecological classification as well as the effect of such discrepancies on economic valuation of water quality. Furthermore, the residents perceived the status of their nearby waters more negatively than the monitored ecological classification suggested, the magnitude of this divergence being, on average, one status category. Our results reveal that the observed eutrophication and browning of surface waters are factors underlying perceptions of water quality which differ from the monitored ecological status. Factors that naturally discourage the recreational use of water, such as natural humus content or clay turbidity, do not have a detrimental effect on the official ecological status of a water body.

Some 40% of the respondents had observed deterioration of lakes and rivers in recent years in Vuoksi, the largest river basin district in Finland. This perception, mostly due to eutrophication or surface water browning, decreased welfare estimates; in fact, of the factors related to water quality this was the one most closely associated with mean WTP. In another salient finding, the official ecological status of a resident’s focal water body was shown to affect mean WTP estimates for improved ecological status. As in previous studies, increased WTP was found for improvement of waters already having a comparatively high status.

Overall, the study highlighted that, in the majority of cases, respondents’ perceptions of their focal lake or river did not correspond to the official status and that this discrepancy resulted in lower WTP estimates for improving lakes and rivers. It is likely that the choice to use a combination of visual materials to describe the ecological status as well as an ex-post GIS analysis of the respondents’ focal water bodies brought added value to the economic valuation. This methodological insight stands to inform and enhance future freshwater-related valuation studies, in particular those relating to the WFD, in which understanding ecological status is crucial. In general, describing the initial water quality according to ecological criteria and as uniformly as possible would facilitate the use of valuation results in future benefit transfers or meta-analyses. The procedure put forward in this paper is applicable in countries where monitoring data on ecological status are available.

Acknowledgements

We would like to thank four anonymous reviewers for their valuable comments on the manuscript. We are grateful to the Ministry of the Environment (Projects YM4/5512/2016 Vearme and VN/845/2019 Koku), the Nordic Council of Ministers (Project 15228 CAPITAL) and the Strategic Research Council of the Academy of Finland (Contract No. 312650 BlueAdapt) for financial support. We thank Juho Kotanen, Pertti Manninen and Sari Väisänen for their help in compiling the questionnaire and applying for background materials. We are grateful to Marja Vierimaa for the graphical layout of the questionnaire and Richard Foley for his valuable work in correcting the language of the manuscript.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

Ministry of Environment Finland, Strategic Research Council of the Academy of Finland. Nordic Council of Ministers.

Notes

1 According to Vilmi et al. (2021), no overall temporal change in water quality and ecological status was observed between 2008 and 2020 in Finland, although improvements were observed in some rivers and lakes. It is likely that freshwater protection measures have been inadequate and targeted ineffectively, and have thus failed to bring about a nationwide improvement in ecological status (Vilmi et al. 2021).

2 Following a suggestion put forward in Rowe, Schulze, and Breffle (1996), we used an exponential response function of the form (1 + k)^n-1 to generate a set of n bids, where k > 0. The researchers’ previous CV studies in two different watersheds were used to help design the payment vector. In the testing phase, an open WTP question was used to ask respondents for the size of the payment they were willing to make.

3 In the present study, an increase in taxes would not have been a credible option for improving the state of high-quality surface waters in Vuoksi. A property-specific water fee was also not possible, as the survey also targeted owners of leisure homes, who did not permanently live in the study area (not documented in this manuscript).

4 To distinguish both protest zero and true zero responses we used the same follow-up question after the valuation question. The following statements were used for example to reveal the true zero responses: “The water bodies of the area are not very important to me”, “I can’t afford to pay for water treatment”, “I already contribute to the regional/local water management association”, and “I’d rather spend my money on something else”.

5 More specifically, and as we had an exponential payment vector, the midpoints were defined exponentially, whereby they were slightly closer to the lower (or chosen) value of the payment interval than to the upper value. That is, a respondent who was “definitely” willing to pay two euros a month, or EUR 24 a year (and not EUR 48 a year), was assumed to be willing to pay EUR 34 a year with this degree of certainty.