Residents’ interest in landscape value trade related to wind energy: application of the attitude–behavior framework to willingness to pay

ABSTRACT

Reducing the undesirable outcomes of wind power (WP) in the vicinity of homes with forest management practices can increase the acceptance of WP. If forest owners would avoid clear felling and use light forest management close to homes towards wind turbines, residents might be interested in paying for such a ‘landscape shield’ in the payment for ecosystem services (PES) context. The majority (83.7%) of the survey sample from Finland were interested in participating in the PES mechanism. On average, they were willing to pay €80.9 per hectare annually to participate in landscape value trade arranging a landscape shield against wind turbines. We applied the attitude–behavior framework to understand the factors and structures underlying residents’ willingness to pay (WTP). The analysis emphasized the importance of intentions, attitudes, and subjective norms over socio-demographic variables in explaining WTP. WTP and the intention to contribute to the cost of the landscape shield were determined by the intention to discuss landscape protection with forest owners and further by the attitudes towards the landscape shield and the interest of neighbors it. This result strongly emphasizes the importance of communication, both with the providers and other consumers of the service.

EDITED BY:

• Ram Pandit

KEYWORDS:

• Payments for ecosystem services
• harmful effects of wind turbines
• forest landscapes
• landscape value trade
• attitude–behavior framework
• willingness to pay

1. Introduction

Wind farms are commonly regarded as a sustainable way to produce energy, as they reduce the release of CO₂ into the atmosphere per MWh generated and accordingly mitigate climate change. More generally, the environmental benefits of wind power (WP) generation come from eliminating the negative externalities of energy production based on fossil fuels, which reduce ecosystem services by locally, nationally, or globally polluting the air (Kennedy 2005). Regardless of the climate benefits and the reduction of air pollution, wind farms may cause local harm perceived by citizens (Warren et al. 2005; Groothuis et al. 2008; Krekel and Zerrahn 2017). This is due to the visual landscape disturbance and noise pollution by wind turbines, concerns over health effects, or even effects on biodiversity (Zerrahn 2017). If the perceived local harm were minimized, the acceptability of WP among the residents of potential target areas could increase and decisions over the location of new plants would be easier. Here, we focus on the opportunities to relieve the perceived harmful visual effects of WP by changing forest management practices, as well as the interest of citizens in the ecosystem services provided by forests.

Because of local and regional conflicts regarding WP construction, the magnitude and significance of the observed local nuisance has been analyzed in many studies (Wolsink 2007, 2015). Using data sets from various countries and different contexts, several studies have analyzed attitudes towards WP construction and local and regional effects perceived by citizens (Krohn and Damborg 1999; Ek 2005; Warren et al. 2005; Wolsink 2007; Ladenburg 2008; Bidwell 2013; Lindén et al. 2015; Zerrahn 2017). Beyond public attitudes and acceptability, several studies have investigated the monetary value of perceived harmful effects. Bartczak et al. (2021), Drechsler et al. (2011), Mariel et al. (2015), Meyerhoff (2013), Meyerhoff et al. (2010), and Vecchiato (2014), for example, assessed the perceived local effects of wind turbines and measured monetary willingness to pay (WTP) to avoid them. In addition, Dimitropoulos and Kontoleon (2009) assessed residents’ willingness to accept compensation (WTA) for wind energy production. These studies explained the monetary measures of preferences for the effects of WP with different socio-economic and attitude-based variables, but they did not focus on the underlying theoretical framework or the interlinkages of the independent variables in the empirical models.

A potential theoretical framework for better understanding preferences and to theoretically structure the model of perceived WP externalities, support for wind energy, and related WTP could be the attitude–behavior framework (Ajzen and Fishbein 1977; Ajzen 1991). This framework suggests that behavioral intention arises from attitudes towards the behavior, subjective norms, and optionally also perceived behavioral control associated with the behavior (Ajzen 1991). However, in the analysis of WTP for consuming wind energy or for avoiding the externalities of wind turbines, the framework has only been applied in a few studies (Bang et al. 2000; Lei et al. 2011; Liu et al. 2019). This is surprising, as studies dealing with the antecedents of WTP are suggested to have been inspired by the attitude behavior–framework (Oerlemans et al. 2016). Although the published literature applying the framework in the case of wind energy production and the perceived effects of wind turbines is rather limited, we argue that the use of the framework may provide important insights into the phenomenon.

In this paper, we focus on how the undesirable impacts of WP in the vicinity of permanent residences or vacation homes could be reduced through silvicultural practices. By avoiding clear cutting or other intensive regeneration measures and using selection felling or extended rotation in forests close to homes towards wind turbines, forest owners could provide a forest ‘shield’ to reduce the harmful visual effects (Tyrväinen et al. 2021; Mäntymaa et al. 2021). Then, the belt of standing mature trees near houses, here referred to as a landscape shield, would provide various ecosystem services, particularly landscape services, by concealing the turbines from sight and preventing them from spoiling the scenery from homes. If landowners were prepared to refrain from felling mature trees between wind turbines and nearby homes, the undesirable effects of the turbines could be mitigated or at best completely avoided. Monetary compensation, i.e. a payment for ecosystem services (PES) (Wunder 2007; Smith et al. 2013), in this case landscape value trade (LVT), could encourage landowners to protect the landscape and to reduce the scenic externalities of wind energy parks. The idea of applying LVT to minimize the negative externalities of wind turbines with forest management practices is appealing but has only been investigated from the supply side. On the supply side, Mäntymaa et al. (2021) demonstrated the interest of forest owners in providing a landscape shield close to housing areas towards wind turbines by delaying the cutting of a stand from the economically optimal time point and consequently bearing income losses if compensated in the LVT mechanism.

The aim of this paper is to illuminate the demand side of LVT in the situation of landscape externalities of wind turbines. Using survey data, we investigate the interest of residents in a landscape shield. We assess the monetary value of landscape ecosystem services in the PES framework. Furthermore, we apply the attitude–behavior framework to understand the factors and behavioral structures behind citizens’ WTP. As far as we are aware, there have been no published studies combining these three aspects: WP, LVT with PES, and the attitude–behavior framework. In particular, the role of socio-demographic and attitudinal variables is analyzed. To facilitate evaluation of the feasibility of LVT in the case of WP and to enable a comparison between landowner WTA, we provide information on WTP for minimizing the externalities of WP.

2. Theoretical background

A conventional theoretical underpinning of the economic valuation of non-marketed goods or services, e.g. ecosystem services, lies in consumer theory and the theory of individual welfare. In the latter, the problem is how much individual welfare changes if the price or the quality/quantity of a commodity (i.e. a good or service) increases or decreases (see, e.g. Varian 2010). In the calculation of gains and losses from the changes, the changes in quality/quantity typically target ecosystem services. For the practical valuation of the changes in an ecosystem service, several methods, e.g. contingent valuation (CV), travel cost, and choice experiments, have been developed (see e.g. Perez-Verdin et al. 2016). CV, the method we use in this study, creates a hypothetical market by which respondents are asked to state their maximum WTP or minimum WTA for a change in the quality/quantity of an ecosystem service (Mitchell and Carson 1989). In the case of WTP, the question is usually worded as follows: How much at most are you willing to pay for the fact that the quality/quantity of the service will improve, or the quality/quantity of the service will not deteriorate? (Mitchell and Carson 1989). The latter option is appropriate in the case of this study.

Here, we are especially interested in explaining the stated WTP in a valuation setting with the attitude–behavior framework. This framework covers theories that predict individual behavior based on attitudes and other individual perceptions. At present, the prevailing theory is the theory of planned behavior (TPB), a social psychological model applied to understand and predict individual behavior. TPB can be seen as a series of associated variables. It assumes a link between behavior and behavioral intentions (Bi), and secondly, a link between behavioral intentions and a weighted combination of attitude (Att), a subjective norm (Sn), and perceived behavioral control (PBC). The perceived behavioral control differentiates TPB from TRA (the theory of reasoned action) which was widely used in literature before PBC was introduced by Ajzen (1985).

In both theories, attitude is defined as overall evaluations, i.e. an attitude towards any concept is simply a person’s general feeling of the favorableness or unfavorableness of the concept (Ajzen and Fishbein 1980). The theory includes an additive model (Equation 1) in which attitude is formed as a summative belief index that is composed of n salient beliefs concerning the outcomes of specific behavior (b_i) and the evaluations of these outcomes (e_i),

1 $Att={\sum }_{i=1}^n{b}_i{e}_i$1

A subjective norm is defined as the influence of the social environment on intention and behavior. It refers to individuals’ perceptions of whether people who are important to them think they should or should not perform the action in question.

PBC refers to an actor’s evaluation of the perceived ease or difficulty of performing the specific action, reflecting past experience and anticipated impediments and obstacles based on second-hand information. If PBC is omitted from the model, TPB reduces to its predecessor, the theory of reasoned action (TRA) (Ajzen and Fishbein 1977).

When applying either TRA or TPB in a valuation context, the stated preferences can be considered as behavioral intentions (Heberlein and Bishop 1986) preceding actual behavior. Furthermore, in the case of valuation, the closer the intention is in time and context to the actual behavior, the more precise it is in predicting it. The application of TPB in CV is illustrated in Figure 1. Ajzen and Peterson (1988) discussed how WTP could be assessed using the attitude–behavior framework. They pointed out that a whole range of attitudes – from the attitude towards the public good to the attitude towards the policy dealing with the public good, and, finally, to the attitude towards paying for the public good – are related in valuation. The value placed on the good itself may differ from the value of the policy designed to provide the good, and furthermore, the value of the policy may again differ from the WTP for the good. Here, we start from TPB and apply it to explain the willingness of respondents to participate in LVT and to pay for a landscape shield in the case of WP. Whether TPB or TRA is supported by our data is an empirical question that is clarified in the following analysis.

Figure 1. The theory of planned behavior in the case of landscape value trade, the starting point for this study.

3. Previous literature: TPB and PES participation in the case of WP

Attitudes toward the production and consumption of green energy, including wind electricity, have been examined from several perspectives (Wolsink 2015; Rand and Hoen 2017). Many of these studies have used models based on social psychology, such as TRA or TPB. Based on the TRA approach, Lei et al. (2011) found that information distribution, pro-environment values, and the tendency to emulate a particular behavior from the surrounding neighborhood play key roles in the intention to consume wind or solar energy in China. Halder et al. (2016) applied TPB in explaining the intention of high school students to use bioenergy in two culturally different contexts, i.e. in Finland and India. Regarding opposition to planned and existing wind farms in Australia, Read et al. (2013) found that, among the factors of TPB, only a subjective norm in the form of social pressure from family, friends, and neighbors anticipated intentions to oppose the wind farms. Applying TPB, Johansson and Laike (2007) analyzed how visual perceptions and attitudes related to wind turbines affect people’s tendency to resist turbines locally in Sweden. They observed that the most important factors behind this resistance were the impression of landscape unity, personal attitudes towards the effects of wind turbines on landscape aesthetics and recreation, and the general attitude towards wind electricity production.

Many studies have combined social psychological models and monetary valuation. López-Mosquera et al. (2014), for example, specified the influence of TPB components on visitors’ WTP for urban park conservation, Bernath and Roschewitz (2008) for the recreational benefits of urban forests, and Pouta and Rekola (2010) for the reduction of forest regeneration to improve recreational possibilities and the quality of the landscape, as well as protecting biodiversity.

On the other hand, attitudes and WTP related to the perceived externalities of renewable energy sources have been examined in a wide variety of studies (Stigka et al. 2014; Oerlemans et al. 2016). Assessing welfare effects with WTP, Bartczak et al. (2021) investigated whether preferences and their heterogeneity related to wind energy development close to peoples’ homes are influenced by individual beliefs about the negative effects of wind turbines in Poland. Roe et al. (2001) determined that several consumer groups in the US are willing to pay significantly more for electricity when emission reductions are made by increasing the use of renewable fuels. Several studies have also examined the perceived landscape externalities of wind turbines. Using an approach based on WTA, Dimitropoulos and Kontoleon (2009) examined factors affecting the local acceptance of WP in Greece. Moreover, Mäntymaa et al. (2021) assessed the WTA that forest owners would potentially require for preventing wind turbines from being visible from peoples’ places of residence or vacation homes.

However, studies explaining WTP for preventing WP externalities with the TPB/TRA framework are rare. One exception is a study by Bang et al. (2000), who used TRA as a theoretical framework and examined the relationships between consumer concern for the environment, consumer knowledge and beliefs about renewable energy, and consumers’ increased WTP for using renewable energy in the US. They found significant positive relationships between beliefs, concern, knowledge, and WTP.

There have already been a multitude of empirical studies related to WTP for ecosystem services in the PES framework (Mäntymaa et al. 2009; Bhandari et al. 2016; Nielsen-Pincus et al. 2017; da Motta and Ortiz 2018; Ren et al. 2020; Tyrväinen et al. 2021). However, as far as we are aware, no studies to date have assessed WTP for ecosystem services in the PES framework with either TPB or TRA. In the following, we fill this gap.

4. Case study area, data, and methods

4.1. Case study area: two counties in southwestern Finland

We focus on two counties in southwestern Finland, i.e. Varsinais-Suomi and Satakunta (Figure 2), where wind electricity production has been predicted to expand (Huttunen 2017). Wind farms are and will be located mainly on private land, because in these counties, most (76%) of the forestry land is owned by private individuals, followed by the state and municipalities (9%), foundations, churches, and other private entities (8%), and businesses (7%) (Finnish Forest Centre 2022). The regional land-use plans have listed and mapped the sites suitable for the construction of wind electricity parks (Regional Council of Southwest Finland 2011; Regional Council of Satakunta 2014). These sites are mostly located in rural areas, having dispersed settlements but not being uninhabited. Land-use legislation in Finland prescribes that before the detailed planning and construction of a WP park, the area must be indicated in the regional land-use plan ratified by the regional council of a county (Land Use and Building Act 132/1999). Local municipalities have authority over the preparation and decision making related to detailed land-use planning within their borders. After the decision making, private electricity companies apply for permission for construction. The same or other private companies own, manage, and operate the constructed turbines, selling the produced electricity to the owners of the companies or to the markets on the Nordic electricity exchange. WP companies lease land from landowners for the construction of turbines.

Figure 2. Case study area: the counties of Satakunta and Varsinais-Suomi. Legend: red points = existing wind parks, green points = plafigurenned wind parks.

4.2. Questionnaire, data collection, and sample representativeness

Via an Internet survey, we collected data on citizens’ interest in participating in an LVT initiative and their WTP for purchasing a landscape shield to minimize the landscape degradation caused by wind turbines. The questionnaire of the survey had four sections. The first section asked about the respondents’ attitudes regarding anthropogenic changes in the landscape, their attitudes towards wind turbines, and beliefs regarding the impacts of WP on energy production, the landscape, and nature. This section also included a question about respondents’ attitudes towards compensation for the externalities of wind turbines.

The questionnaire contained a map (Figure 2) showing the locations of existing and planned WP parks in the study area. The respondents were told: ‘The provincial land use plans of Satakunta and Varsinais-Suomi have reserves for wind farms. Some of them have already been implemented and some are still unrealized. The following map shows the locations of wind farm reservations. Moving to map view can take a few seconds’. The respondents were able to zoom in and out of the map, which increased the spatial preciseness of the survey. They were asked to locate their permanent residences, as well as possibly owned vacation homes and forest lots on the map in relation to the turbines.

Section 2 briefly explained the concept of a landscape shield. The respondents were informed that ‘The effects of wind farms could be reduced with forest management. This would mean that the landowner avoids forest logging between residential areas and wind farms or would apply lighter forest management near residential areas towards wind farms, for example continuous cover forestry. This would allow the vast majority of trees to be preserved, obscuring wind turbines from sight. These forest areas are called landscape shields in the following’. In the survey, the idea of a landscape shield was illustrated with a schematic diagram (Figure 3) showing respondents the location of a narrow belt of mature standing trees that would be preserved between their homes and the turbines. To be able to hide the turbines of 250 m to 300 m in height from the visual landscape of housing areas, the forest belt would need to be located near the housing areas. This implies that even rather small areas of forest are enough to prevent landscape damage.

Figure 3. Illustration presented in the survey: ‘Above, you can see a schematic diagram of the principle of operation of the landscape shield. The protective effect of the trees is best at a distance of two to three kilometers from wind turbines’.

Thereafter, the section described a scenario for LVT and a possibility to pay forest owners for providing a landscape shield. Here, the respondents were asked to ‘Think of having a group of wind turbines, i.e. a WP park, built in your vicinity. Forest owners who have property near a wind turbine would have the opportunity to reduce the negative impact of the wind park. Landowners would avoid logging the forest in the area between wind parks and the settlement; instead, a forest strip, or landscape shield, that would hide the turbines would be left in the area’. Then, the respondents were asked about their beliefs regarding the feasibility and effectiveness of a landscape shield for minimizing the harmful visual effects of turbines and for conserving the benefits of ecosystem services. Moreover, the section asked about the respondents’ own intention to pay for a shield and their perception of the interest of neighbors in doing the same. The latter was important for the organization of LVT, because the interest of the neighbors in participating was crucial to be able to collect a sufficiently large group of payers for the trade of a particular landscape shield. Here, we asked respondents to imagine that a possible landscape shield could prevent the visibility of wind turbines from their permanent residence or vacation home. We wanted to create a scenario that would be applicable to all the respondents, allowing them to picture a hypothetical condition in which a turbine near their own residence or vacation home would be hidden by a standing forest belt. There was also a question concerning their intention to participate in discussing the preservation of a landscape shield with neighboring forest owners.

The third section represented a hypothetical opportunity to enter into an agreement to create a landscape shield for a defined period of time and asked for the highest sum of money per hectare that respondents would be willing to pay for this type of agreement. The intention to pay, i.e. the contingent valuation WTP question, was asked for all respondents, regardless of whether they had shown an interest in the shield earlier in the survey. To urge the respondents to reveal their maximum WTP, we reminded them here that if offers were too small, they might not lead to agreements with forest owners. A payment card CV method was used to reveal WTP (Boyle and Bishop 1988; Ryan and Watson 2009). The payment card included a set of progressive monetary sums and asked the respondent to specify the highest WTP. The bid vector was €0, €5, €7, €10, €15, €20, €25, €35, €50, €75, €100, €140, €200, €300, €400, €550, €750, €1,000, and over €1,000 per hectare per year. This means that WTPs were revealed in 18 interval classes. Following the payment card, the exact WTP was located somewhere between the bid a respondent had chosen and the next larger bid in the payment card. If, for example, the choice was €10/ha/year, the exact WTP was somewhere between €10 and €15/ha/year (Hackl and Pruckner 1999). Before the principal survey, we carried out a test survey, which confirmed that the bid vector operated well. Although the landscape shields are a relatively narrow belt of trees, they have a depth direction in addition to the width direction. For this reason, we used the area unit ‘per hectare’ as a measure in this study. In addition, the section had questions on attitudes towards the governance of landscape issues in general and towards the LVT initiative in particular. The final section collected background information on the respondents.

After testing the questionnaire with a pilot survey of 100 respondents in January 2019, we made final clarifications to the questionnaire. The principal survey was conducted in February 2019. The data were collected by a commercial survey company, Taloustutkimus Oy, from a representative panel of people selected from the population of the counties by the company. The survey was conducted online by sending out a call and an Internet link to the survey in e-mail messages aiming to reach 1,400 respondents, the targeted number of respondents of the study. The survey company has an extensive list of people who have volunteered to participate in surveys. However, since not everyone responds to every survey conducted by the company, the original sample needed to be supplemented with new people. After two reminders and the supplements of non-responses with new recruitments, we received 1,271 responses, representing a response rate of 26% calculated as the ratio between those who responded to the survey and those who were invited.

We evaluated the representativeness of our data in relation to statistical data on the demographic structure from the same area published in Official Statistics of Finland (2020) (Table 1). In our data, the relative shares of genders among respondents were similar to those in the general population (one-sample t-test, p = 0.256). However, compared to the population, the respondents of the survey were older (chi-squared test of consistency, p = 0.000) and they more often lived in towns and cities than in the countryside (p = 0.000). One may expect that those people who were more concerned than average about the harmful effects of wind turbines would be more likely to respond to the survey. This must be considered when interpretating the results.

Table 1. Socio-demographic features of the respondents in the counties of Satakunta and Varsinais-Suomi in the wind turbine study and in the Official Statistics of Finland (2020).

		Satakunta and Varsinais-Suomi in the wind turbine study	Demographic structure of Satakunta and Varsinais-Suomi^a
Sample size^b		1,271
Gender (%)	Female	48.2	50.0
	Male	51.8	50.0
Age (years, %)	15–24	4.3	14.1
	25–34	8.6	15.3
	35–44	13.1	15.4
	45–54	19.9	17.5
	55–64	24.1	16.9
	65–74	24.8	16.8
	75–79	5.3	4.1
Population	15–79		59,3831
Type of living environment (%)	Countryside	19.8	27.1
	Town or city	80.2	72.9

^aSource: Official Statistics of Finland (2020).

^bThe number of observations varies between questions.

Those who answered ‘zero’ or ‘don’t know’ to the WTP question were asked in a follow-up question to explain the answer (Table 2). Because we allowed respondents to choose the top five reasons, there is considerable overlap in the numbers in the table. Overall, the frequency of choices of different justifications ranged from 19 to 261, i.e. from 1.5% to 20.5% of the total number of respondents. The justifications ‘The WP company’s obligation is to pay for the landscape shield’ and ‘The task of the WP company is to minimize the effects on the landscape’, mostly chosen by the same respondents, received the most mentions, i.e. 261 (20.5%) and 252 (19.8%), respectively. This indicated that a clear majority of the respondents accepted the WTP-type questions and only these were included in the data set used in the modeling.

Table 2. Rationales for ‘zero’ or ‘don’t know’ responses for WTP related to a possible landscape shield.

	Number of respondents	%
The obligation of the wind power company is to pay for the landscape shield.	261	20.5
The task of wind power companies is to minimize the effects on the landscape.	252	19.8
Money and landscape cannot be compared.	152	12.0
I couldn’t afford to pay.	140	11.0
Wind power in the landscape doesn’t cause harm.	122	9.6
Other problems are more important.	100	7.9
I don’t think a landscape shield would work.	96	7.6
I haven’t received enough information.	80	6.3
Most likely, the landowner would not log the area anyway.	69	5.4
I don’t care about the scenery.	19	1.5
All respondents	1271

4.3. Statistical analysis

We applied structural equation models (SEMs) in the statistical analysis. An SEM has several advantages compared to other regression models. It allows researchers to control for measurement error when using latent constructs, to investigate modeled path coefficients simultaneously, to test for the overall consistency between the data and the hypothesized model, and to test for mediating relationships between variables in a more straightforward manner than traditional methods (Blanthorne et al. 2006).

Our study included three different variables for intention. Intention to discuss a landscape shield with forest owners (INT_DISC) and intention to participate in the cost (INT_COST) were used in the model as measured in the survey, i.e. with an ordered three-class variable (1 = yes, 2 = maybe, 3 = no). INT_WTP was used as a continuous variable, but it was log-transformed to satisfy the linearity assumption. Due to numerous zero values for the variable INT WTP, one euro was added before transformation to avoid problems. Although an SEM can handle non-normally distributed variables, there are benefits if the data follow a multivariate normal distribution.

Attitudes towards WP (ATT_WP) and attitudes towards a landscape shield (ATT_LS) were used as single observed variables measured with statements on a five-point Likert scale from strongly disagree to strongly agree (from 1 to 5).

The subjective norm was measured regarding two reference groups: neighbors and future generations. The subjective norm related to neighbors was measured with an ordered three-class variable about the perceived interest of neighbors in a landscape shield (SN_NEIGH; 1 = yes, 2 = maybe, 3 = no). The subjective norm concerning future generations (SN_FUTURE) was constructed as a factor of six statements related to peoples’ consideration of future generations in environmental decision making (see Appendix).

The beliefs behind attitudes, i.e. beliefs about WP and about the landscape shield, were formed based on respondents’ evaluations of statements presented in the questionnaire (Table 3). Exploratory factor analysis (EFA) was used to construct the factors based on the variables related to beliefs for the analysis. The first three factors were compiled from the responses to 21 statements related to beliefs concerning the effects of wind turbines on livelihoods, the possibilities for recreation, and the scenery and landscape of the region, as well as on the importance of the turbines in future energy production and the prevention of climate change (BELF_HARM, BELF_CLEAN, BELF_DISTURB). Beliefs about the landscape shield (BELF_LS) comprised one factor compiled from the responses to five statements.

Table 3. Variables of the statistical analysis.

Variable category	Variables	Interpretation	Mean	Std dev.
Intentions	INT_WTP	Intention to pay = WTP, dependent variable, logarithm of class midpoints	2.35	2.17
	INT_COST*	Intention to participate in the cost of a landscape shield	2.61	0.56
	INT_DISC*	Intention to discuss a landscape shield with forest owners	2.07	0.76
Perceived behavioral control***	INCOME	Monthly personal income before taxes, logarithm of class midpoints	7.21	1.98
	AGRIFOR*	Occupation in agriculture or forestry; binary variable, 1 = yes, 0 = no	0.01	0.07
	AGE	Age, years	55.12	14.99
Attitudes	ATT_WP	Attitude towards wind power, factor	N/A*	N/A*
	WP_NO	Dislike of wind turbines	3.72	1.22
	WP_HA	Thinking that wind turbines are harmful	3.81	1.17
	WP_MORE*	More wind power is needed	2.22	1.17
	WP_SUPP*	Support for the construction of wind turbines	2.25	1.20
	ATT_LS*	Attitude towards landscape shields	1.74	0.72
Subjective norms	SN_NEIGH*	Interest of people living in the neighborhood in a landscape shield	1.73	0.63
	SN_FUTURE*	Consideration of future generations, factor	N/A*	N/A*
	FU_DISCUS*	Discussions of environmental topics with close relatives	2.50	1.09
	FU_NEGAT*	Consciousness of the negative environmental effects of my action	2.06	0.80
	FU_GOOD*	The surrounding environment should be devolved in a good condition	1.69	0.79
	FU_GENER*	Future generations should be considered in environmental decision making	1.56	0.75
Beliefs	BELF_CLEAN	Beliefs about the positive effects of wind power, factor	N/A**	N/A**
	CL_TECH*	Wind power is a part of future technology	2.16	1.07
	CL_DOMES*	Wind power is a good source of domestic energy	2.03	1.07
	CL_GAIN*	Future generations will gain from the development of wind power	2.01	0.96
	CL_CLIMAT*	Wind power curbs climate change	3.19	1.23
	CL_FUTURE*	Wind turbines are essential for future energy production	2.17	1.08
	CL_CLEAN*	Wind power is clean	1.99	0.97
	CL_NPROFIT	Wind power is not profitable	3.55	1.07
	BELF_HARM	Beliefs about the harmful effects of wind turbines, factor	N/A*	N/A*
	HA_IMPROV*	Wind turbines improve the quality of the landscape	3.45	1.09
	HA_REGION*	Wind turbines strengthen the originality of the region	2.96	1.07
	HA_LVALUE	Wind turbines decrease the value of land	2.98	1.05
	HA_SPOIL	Wind turbines spoil the quality of the landscape	3.28	1.23
	HA_IMAGE	Wind turbines spoil the image of the region	3.19	1.23
	BELF_DISTURB	Beliefs about the disturbing effects of wind turbines, factor	N/A*	N/A*
	DI_AGRFOR	Wind turbines make agriculture and forestry more difficult	3.54	0.96
	DI_MOVE	Wind turbines hinder people from moving freely in nature	3.61	1.09
	DI_HELTH	Wind turbines may have a health effect on people	3.10	1.04
	DI_BIRDS	Wind turbines disturb birds and other animals	2.72	1.05
	DI_HUNT	Wind turbines disturb hunting	3.45	0.98
	DI_NOISE	Wind turbines make a disturbing noise	3.13	1.06
	BELF_LS	Beliefs about landscape shields, factor	N/A*	N/A*
	LA_PREVEN	The importance of a landscape shield is small	3.21	1.09
	LA_RNOISE*	A landscape shield would decrease noise impacts	2.31	0.95
	LA_RECRE*	A landscape shield would protect recreation	2.31	0.97
	LA_NATURE*	A landscape shield would protect nature values	2.30	0.99
	LA_VALUE*	A landscape shield would maintain landscape values	2.21	0.96

 * Variable measured on an inverse scale in the analysis.

**Statistical indicators are not presented for factors, as they are standardized between [0, 1].

***Left out of the final model.

We also built a variable for PBC from those measures that related to respondents’ perceptions about their possibilities to pay, such as income, household size, and some protest responses. However, the construct was not significant at the level of α = 0.10in subsequent analysis. Because PBC was left out of the final model, the estimated model thus corresponded with TRA rather than TPB.

For the SEM analysis, all variables measured on an inverse scale, i.e. small values represented a positive view, were rescaled in line with the others for consistency and interpretability (Table 3). These transformations allowed us to use the maximum likelihood (MLM) estimation method in the SEM, which requires multivariate normality of the variables. Multivariate normality was assessed with residual plots and multivariate normality tests. Both measures used, Mardia’s based kappa (k = 0.26) and relative multivariate kurtosis (k = 1.20), together with residual plots, indicated adequate normality. However, the robust MLM estimation method was also tested due to a few ordinal variables. The results were similar, except for one path from SN_FUTURE to INT_WTP, where the statistical significance was slightly weaker (p = 0.103). This path was still retained in the model.

In the SEM, variables related to behavioral intentions, subjective norms, and attitudes (INT_COST, INT_DISC, SN_NEIGH, ATT_LS) were hypothesized to have a causal relationship with WTP.

Modification indices, such as the Lagrange multiplier test, were used to improve the fit of the SEM by omitting a few statistically non-significant relationships hypothesized to be causal. In addition, to improve the model fit, two correlations of error terms were allowed. The goodness-of-fit measures of the model, i.e. the comparative fit index (CFI), root mean square error of approximation (RMSEA), and standardized root mean square residual (SRMR), are provided in the box in Figure 5. These three known criteria were used to evaluate the goodness of fit. According to Hu and Bentler (1999), CFI≥0.90 can be considered an indicator of a reasonable fit and CFI≥0.95 a good fit, while SRMR and RMSEA≤0.08 can be considered as indicators of a reasonable fit, and RMSEA≤0.05 a good fit. The chi-squared test was also used, but it is known to be problematic with large samples (Vandenberg 2006). Statistical analyses were performed using the procedure CALIS of SAS Enterprise Guide 7.15 (SAS Institute, Inc., Cary, NC, USA).

5. Results

5.1. Descriptive results

The respondents expressed a rather high interest in participating in the LVT providing a landscape shield. With a three-step scale, 42.4% responded ‘yes’, 41.3% ‘maybe’, and 16.3% ‘no’ to the question regarding their interest in participating. If the first two categories are summed, a clear majority (83.7%) were interested in participating in the PES mechanism, despite the idea of a landscape shield and LVT most likely being new to most of the respondents.

Because we used a payment card as an elicitation technique, the survey did not provide exact monetary amounts for the provision of a landscape shield but ranges within which the individual perceptions of WTPs were located. Figure 4 presents the distribution of the bids and WTP responses across the ranges. Computed from the category centers of the bid vector of the payment card, the annual mean WTP was €80.9 per hectare (std dev. €215.6/ha/year) and the median class €10–€14.9/ha/year. In the following analysis, the scale of the dependent variable (INT_WTP) was reversed for consistency with other variables.

Figure 4. The bid vector, ranges of location (in parentheses), and the distribution of choices (%) of maximum WTP for participating in providing a landscape shield (N = 1011).

*In calculations and analyses, we used 1300 as the value of the bid “more than 1000”.

5.2. Correlations

The lower triangle of Table 4 presents the correlations between the dependent variable, i.e. intention to pay (INT_WTP), and variables directly measured in the survey, and the upper triangle the corresponding statistical significances of the correlations. The directly measured variables include variables related to intentions or interests (INT_COST, INT_DISC), and attitudes towards action (ATT_LS), the subjective norm regarding neighbors (SN_NEIGH), as well as three socio-demographic characteristics (INCOME, AGE, EDUC) that we considered to be important for theoretical and empirical reasons. INT_COST had a strong, positive correlation (r = 0.414) with INT_WTP, indicating that interest in participation in the cost of a landscape shield tended to increase WTP. Furthermore, INT_DISC, SN_NEIGH, ATT_LS, and AGE had positive and significant correlations, meaning that an increase in these variables tended to increase the intention to pay. The plus sign of INT_DISC, for example, means that the intention to discuss a landscape shield with forest owners increased the probability of being willing to pay. On the other hand, INCOME and EDUC did not correlate at all with INT_WTP.

Table 4. Correlation coefficients (lower triangle) and related statistical significances (upper triangle) (H₀: |r| = 0) of measured variables related to intentions, interests, and attitudes towards action, and socio-demographic characteristics. The scale of INT_WTP was reversed for consistency with the other variables.

	INT_WTP	INT_COST	INT_DISC	SN_NEIGH	ATT_WP	INCOME	AGE	EDUC
INT_WTP	1	0.000	0.000	0.000	0.000	0.384	0.000	0.286
INT_COST	0.415	1	0.000	0.000	0.000	0.503	0.000	0.142
INT_DISC	0.266	0.413	1	0.000	0.000	0.534	0.026	0.605
SN_NEIGH	0.143	0.305	0.356	1	0.000	0.319	0.000	0.783
ATT_LS	0.187	0.281	0.463	0.606	1	0.522	0.085	0.957
INCOME	−0.028	−0.021	0.020	0.032	0.020	1	0.000	0.000
AGE	0.116	0.164	0.070	0.138	0.054	0.219	1	0.434
EDUC	−0.034	−0.046	0.016	0.009	0.002	0.215	−0.025	1

5.3. The structural equation model

The standardized estimates of the hypothesized relationships behind WTP in landscape value trade are presented in Figure 5. To make the interpretation of the figure easier, the variables are organized to follow the TPB – TRA applied in our case in Figure 1. In other words, on the left-hand side, we have displayed factors reflecting beliefs about the positive effects of wind power (BELF_CLEAN), the harmful effects of wind turbines (BELF_HARM), the disturbing effects of wind turbines (BELF_DISTURB), and landscape shields (BELF_LS) with their measured components. In the middle are attitudes towards wind power (ATT_WP) and towards landscape shields (ATT_LS) and subjective norms, including the interest of neighbors in a landscape shield (SN_NEIGH) and consideration of future generations (SN_FUTURE). Finally, on the right-hand side are variables related to intentions to discuss a landscape shield with forest owners (INT_DISC), participate in the cost of a landscape shield (INT_COST), and pay for the landscape value trade (INT_WTP). As the value of CFI was 0.960, RMSEA 0.032, and SRMR 0.055, the model fit was found good.

Figure 5. Structural equation model of residents’ interest in participating in landscape value trade in the case of wind energy (N = 1011). The scale of INT_WTP was reversed for consistency with the other variables. Legend: correlation coefficients → figures on the double-headed arrows with dashed lines; standardized regression coefficients → figures on single-headed arrows with solid lines; variances → figures in parentheses; statistically significant parameters → black figures; variables directly measured in the survey → figures within rectangular boxes; factors constructed from variables measured in the survey → figures within ovals. Classification of variables by colors: green → beliefs; orange → attitudes; purple → subjective norms; brown → intentions; white → actual responses used as the measured components of factors. The hues of the colors vary so that the hue darkens as the decision phase proceeds. A significance level of 0.10 was used.

Our key interest, WTP in landscape value trade (INT_WTP), was at the end of a causal network of beliefs, attitudes, subjective norms, and intentions presented on the right in Figure 5. We found that direct and indirect effects explained 17% of WTP. A strong, positive, direct relation from intention to participate in the cost of a landscape shield (INT_COST) to intention to pay (INT_WTP, β = 0.401) indicated that interest in participating in the cost of a landscape shield tended to increase WTP.

The intention to participate in the cost (INT_COST) was partly explained by the intention to discuss a landscape shield (INT_DISC; β = 0.414). INT_DISC was slightly more explained by other variables (24% vs. 17%) than intention to pay (INT_WTP). The attitude towards a landscape shield (ATT_LS) had a rather strong positive relationship (β = 0.391) with INT_DISC. The attitude towards WP (ATT_WP) indicated a negative, but not as strong (β = −0.112) intention to discuss a landscape shield with forest owners.

From the measures of the attitude towards WP (ATT_WP), WP_SUPP (support for the construction of wind turbines) had the largest standardized regression coefficient (β = 0.909), whereas WP_HA (thinking that wind turbines are harmful; β = −0.883) had the highest negative coefficient.

Subjective norms that directly associated with intentions, such as the norm regarding future generations (SN_FUTURE), had a positive relationship with intention to pay (INT_WTP, β = 0.099). In addition, SN_NEIGH (interest of people living in the neighborhood in a landscape shield) associated with the intention to discuss a landscape shield with forest owners (INT_DISC), but also with the attitude towards landscape shields (ATT_LS), with a slightly higher coefficient (β = 0.505).

Of the factors regarding the beliefs about the effects of WP, BELF_CLEAN (beliefs about the positive effects of wind power) had the strongest and positive direct relationship (β = 0.625) with ATT_WP (attitude towards wind power). The factor BELF_CLEAN was formed from seven measured components. Measured with a standardized regression coefficient, CL_DOMES (wind power is a good source of domestic energy, β = 0.900) and CL_TECH (wind power is a part of future technology, β = 0.890) had the highest coefficients with the factor BELF_CLEAN. With the factor BELF_HARM (beliefs about the harmful effects of wind turbines), HA_SPOIL (wind turbines spoil the quality of the landscape) and HA_IMAGE (wind turbines spoil the image of the region) had the highest coefficients, i.e. β = 0.899 and β = 0.825, respectively. In addition, DI_NOISE (wind turbines make a disturbing noise; β = 0.813) and DI_MOVE (wind turbines hinder people from moving freely in nature; β = 0.782) had the highest coefficients with the factor BELF_DISTURB (beliefs about the disturbing effects of wind turbines).

Furthermore, with the factor BELF_LS (beliefs about the landscape shield), LA_VALUE (a landscape shield would maintain landscape values), LA_NATURE (a landscape shield would protect nature values), and LA_RECRE (a landscape shield would protect recreation) had the highest coefficients, i.e. β = 0.886, β = 0.877, and β = 0.878, respectively.

6. Discussion

To enhance WP production, the local harm of wind parks needs to be minimized and the acceptability of WP increased among the residents of potential target areas. One option is to mitigate harmful impacts by changing forest management practices close to wind parks. This study investigated the potential demand of citizens for a landscape shield, i.e. a belt of standing mature trees, concealing wind turbines and preventing them from damaging the scenery of housing areas. The monetary value of the landscape ecosystem service was examined in a PES framework assessing people’s interest in participating in the LVT mechanism to minimize the landscape effects perceived as harmful. To facilitate evaluation of the feasibility of LVT in the case of WP and to enable a comparison with the landowner WTA, we provided information on the WTP for minimizing the negative externalities. Respondents expressed rather high interest in participating in LVT, as a clear majority indicated certain or possible interest in participating in the mechanism. Citizens were on average willing to pay €80.9 per hectare annually for a landscape shield between their permanent residences or vacation homes and wind turbines, whereas the corresponding median class was €10–€14.9 per hectare per year.

As a main finding, the results provide information to evaluate the possibilities for implementing landscape value trade. The total annual WTP per hectare depends, of course, on the total number of payers per hectare. Although we do not have such detailed spatial information on WTP, we can produce rough evaluations about the trade. For the same study area and in a comparable setting, Mäntymaa et al. (2021) analyzed the supply side of the service and found that an average forest owner annually claimed €297.6 per hectare as compensation for providing a landscape shield. As the average WTA is 3.7 times higher than the average WTP revealed in this study, at least 3.7 citizens per hectare would be needed to raise enough funding to implement LVT in a region. Thus, in principle, the organization of LVT is realistic if a small area includes at least four people willing to pay for the service, but one willing person is not enough. Consequently, local conditions are crucial, and communication, discussion, and cooperation are needed to implement the mechanism.

We applied the TPB – TRA framework to understand the factors and behavioral structures underlying citizen WTP, which has not previously been done in this type of context. We found that the main aspects of the TRA also hold in our case. The magnitude of WTP was strongly influenced by the intention to contribute to the cost of a landscape shield, which in turn was significantly determined by the intention to discuss landscape protection with forest owners. In addition, the latter was most influenced by the attitude towards the landscape shield and this, in turn, by the interest of neighbors in the landscape shield. According to the study, peoples’ beliefs about the positive effects of WP had a strong positive relationship with attitudes towards WP, which increased interest in participating in the costs of a landscape shield and discussions about landscape protection, and finally, WTP for a landscape shield. This is in line with Bang et al. (2000), who found that consumers’ extra WTP for renewable energy was positively related to beliefs about the significant consequences of using renewable energy. The results also strongly emphasize the importance of communication about the service with both the providers, or forest owners, and with the other consumers, or fellow citizens.

As an additional result, our study indicated that other ecosystem services than landscape can also be supplied with a landscape shield. The landscape shield was perceived to provide nature protection values, as well as recreation opportunities. This implies that there might also be demand for landscape value trade near housing areas without the landscape disturbance caused by WP (cf. Hanley et al. 2009; Mäntymaa et al. 2018; Notaro et al. 2019).

The analysis additionally demonstrated that socio-demographic variables such as citizen’s age, occupation, or personal income did not explain WTP at all. This is a noteworthy result, indicating that an instrument designer (social planner) cannot use simple socio-demographic characteristics to find positive target audiences for this type of mechanism. In addition, the differences between regions in terms of demographics and income, for example, are relatively small in Finland. Therefore, in the study counties, it is difficult for an instrument designer to find a suitable target area on the basis of socio-economic indicators alone, and the attitudes and beliefs regarding the target need to be known. In addition, the lack of socio-demographic variables in the model, of course, makes the transfer of values from the study site to some other regions problematic.

There were also some limitations in our analysis. We started with the basic variables of TPB but noticed that our operationalization of perceived behavioral control was not adequate to include PBC in the model. The complicated setting of our study, including PES, WTP, and perceptions of WP, also limited our possibilities to include all possible normative reference groups, as well as beliefs and evaluations in the study. The study nevertheless revealed the importance of including beliefs regarding LVT, in addition to beliefs about environmental change itself.

It remains for future research to consider how to operationalize landscape value trade so that it adheres to people’s beliefs, attitudes, and normative perceptions. A future direction to continue would be the testing of a PES scheme in small-scale local cases. Here, we assumed that local residents would be responsible for paying for landscape benefits. It is also possible that some other parties, such as energy firms gaining from WP projects and local municipalities, would be willing to fund the mechanism.

It is, however, unclear who is perceived to ‘own’ the landscape and who should pay compensation for safeguarding it in different cases. In some cases, residents could be ready to pay for the landscape ecosystem services of forests. On the other hand, people may often have a sense of ownership of their local landscape (Horowitz and McConnell 2002; Rakotonarivo et al. 2018). As a tool for understanding this phenomenon, some researchers, such as Matilainen (2019), have offered the concept of psychological ownership. In such a case, the residents may be those who should be compensated for damage to the landscape resulting from the construction of wind turbines. There could also be a case where some residents want to pay for a certain part of the landscape shield because they enjoy recreation there. Along with activity and dependence, we approach the issue of the commons (Ostrom 2002; De Angelis and Harvie 2013). Then, the question is not so much about psychological ownership, but only about different effects, attempts to reduce negative effects, and finally about fairness, acceptance, and cohesion of the village community. The turbines will be in the landscape for 30 years. One day they will be gone, and the structure and mindset of the community may be different.

7. Conclusion

Our study provided evidence that a PES scheme could partly relieve the objection against WP related to local landscape impacts. Citizens could understand the PES scheme, they considered it as a solution, and they were interested in participating. Their intention to participate was strongly associated with attitudes and further beliefs about the impacts of WP. In particular, general beliefs about WP as a source of clean domestic energy together with future technology increased positive attitudes. Our data were collected before the energy crisis in Europe, but due to increasing electricity prices and strong interest in national energy solutions, we can expect that attitudes have shifted in an even more positive direction towards renewable energy and that the need and support for PES solutions will be even stronger in the near future.

Ethical approval

In the project’s research plan, a commitment was made to follow the ethical principles of the Finnish Advisory Board for Research Integrity (http://www.tenk.fi/). Participation in the survey was voluntary for the respondents. Those who participated in the survey were asked for permission to use the collected information for research purposes. The survey did not collect sensitive personal information or the names of the test subjects. The test subjects’ answers were therefore anonymous, and they were only identified by identification numbers in the data.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by the Strategic Research Council (SRC) at the Academy of Finland [PALO project, grant number 312671] and the Ministry of Agriculture and Forestry of Finland (LandUseZero project, grant number 4400T-2110).

Appendix

Six statements that were presented to respondents about considering future generations in environmental issues:

1. Future generations must be considered when deciding on environmental matters.

2. Future generations can improve their environment based on their own values and needs.

3. I feel that it is my duty to leave my environment in a good condition for the next generation.

4. The next generation will have better opportunities to take care of the environment technically.

5. I am aware of the negative environmental effects of my activities in the long term.

6. I often discuss environmental issues with my loved ones.