Tourist Viewshed Externalities and Wind Energy Production

Abstract

This study uses an experiment where ferry passengers are sold hotel room “views” to evaluate the impact of wind turbines views on tourists’ vacation experience. Participants purchase a chance for a weekend hotel stay. Information about the hotel rooms was limited to the quality of the hotel and its distance from a large wind turbine, as well as whether or not a particular room would have a view of the turbine. While there was generally a negative effect of turbine views, this did not hold across all participants, and did not seem to be effected by distance or hotel quality.

While wind energy is a promising source of renewable energy, it continues to make up a small fraction of the world's energy supply (Hau 2005, Kumar et al. 2016). Placement of wind turbines has prompted public concerns related to their effects on wildlife and scenic views (U.S. Department of Energy (U.S. DOE) 2015, for a review, see Ladenburg 2009). This has been especially relevant in coastal areas, where the proposed development of large offshore wind energy facilities in states including Massachusetts, New Jersey, Delaware, Virginia, and Oregon have been seen as a potential threat to the tourism industry. The response from the tourism industry has delayed or even blocked development of offshore wind facilities in some of these states. Literature surrounding economic effects on residents of wind turbines has mostly focused on residential property values through hedonic value studies focusing on the relationship between property price and turbine density (Sterzinger, Beck, and Kostiuk 2003, Poletti 2005, Heintzelman and Tuttle 2012, Lang, Opaluch, and Sfinarolakis 2014, Gibbons 2015). The values of beach community residents, however, may be different from the values of tourists visiting those communities. Other research has used stated-preference methods to measure preferences for reduction in the effects of offshore wind turbines (Gee 2010, Krueger, Parsons, and Firestone 2011, Landry et al. 2012, Parsons et al. 2016), or preferences for offshore turbines relative to other renewables (Ladenburg 2014). Recently, researchers have started to use choice experiments to access how wind farms influence individual preferences (Lutzeyer, Phaneuf, and Taylor 2016). This study uses an economic field experiment using actual beach visitors as research participants.

Participants were offered an opportunity to bid on chances to win hotel rooms in a beach resort town. Participants bid on rooms of differing quality and location. We worked with private hotel owners to offer rooms in our experiment as either having wind turbine views or not. This allowed us to obtain a measure of beach visitors’ preferences for otherwise identical rooms, varying only the existence of wind turbine views from the room.

We find a negative effect of turbine views on beach visitors’ willingness to pay for hotel rooms, however this effect does not hold for all visitors. Indeed, a small portion of the sample were willing to pay a premium for turbine views. We were unable to find any evidence that these values varied either with distance or hotel quality.

Methods

This research uses a within-subject field experiment offering participants the opportunity to purchase a lottery for a weekend stay at one of several hotels in Lewes, Delaware, a popular coastal vacation community. Lewes has a large wind turbine that is visible from nearby hotels. This experiment estimated willingness to pay (WTP) for hotel rooms as a function of distance from the wind turbine and hotel quality measured by a common star rating. Specifically, WTP for the view of the wind turbine was isolated by offering rooms that were distinguished only by the presence or absence of a wind turbine view from the room. The question of interest is whether there is a significant WTP a premium or not for rooms without turbine views, and, for rooms with turbine views, if WTP is affected by the quality of the hotel or the distance from the turbine. Table 1 describes the full set of hypotheses.

Table 1. Hypotheses and Results

This experiment used the BDM mechanism (Becker, DeGroot, and Marschak 1964) to offer lotteries for six different hotel rooms. This mechanism is a type of auction commonly used in economic experiments to obtain a theoretically consistent measurement of participant value for an item by soliciting bids then randomly selecting a price. Bids below the price do not purchase the item; bids above the price purchase the item at the randomly selected price. The experiment was conducted during the summer on a ferry on the Lewes-Cape May Ferry traveling between the coastal vacation towns of Lewes, Delaware, and Cape May, New Jersey. A one-way ferry trip lasted about 85 minutes. This sample population was selected for two reasons. First, ferry riders, at least on this particular ferry during the busiest summer weekends, are largely vacationers to either the Delaware beaches or southern New Jersey shores. Second, this group offers a captive sample for a period long enough to run a fairly complicated design. In other locations, finding participants who would be willing to give up an hour or more of their vacation time is challenging.

Table 2 presents summary statistics of sample demographics and average bids by treatment. Household income was approximately $74,000. This is relatively high for Delaware, which averages $58,000 per year. However, it is important to keep in mind that the respondents were tourists who likely had higher than average incomes. In fact, the average household income observed in this study is only slightly above the broader mid-Atlantic region, as Maryland and New Jersey both have household incomes averaging $70,000. Participants also tended to be older (mean 44.7) and had, on average, 16 years of education. The majority were on multi-day trips, and average spending on the vacation, across both day and overnight trips, was $1,563.

Table 2. Summary Statistics

Participants were recruited through open announcements, at a waiting area where pedestrian traffic lined up to board the ferry, and in an outside viewing deck after the ferry departed. The experiment took approximately 60 minutes, and the total possible earning was $45 – $20 for participating, and an additional $25 in bidding money to be used during the experiment.

Single sessions with groups of participants were run per ferry trip. The sessions took place in a cabin that had a small vending area and a large room full of tables and booths. This room was quiet and always had sufficient open space for participants to spread out, ensuring their privacy. Sessions consisted of low-stakes practice rounds which were done in a group, followed by an individual high-stakes part. The low-stakes part familiarized participants with the BDM mechanism and the randomization procedures used in the experiment. Participants first bid on a pen. Next, they bid on two lottery tickets to win either a pencil or a pen. For each bid, participants received an initial endowment of one dollar. During the practice round, the low-stakes part of the experiment, participants bid on low-value items while sitting at a central table. Each participant was allowed to ask questions publicly of the administrators. For the low-stakes part, participants received a follow-up explanation and had the opportunity to ask questions. For the second, high-stakes portion of the experiment, bidding on a weekend stay, participants were separated, seated at different tables to ensure privacy. Communication between participants was not allowed.

From the initial endowment of $23 that participants received in the second part, they placed independent bids on lotteries for a weekend stay in six different hotel rooms. Participants could make a weekend reservation for the hotel any time in the following six months. The hotel rooms were identical except that arrangements had been made with the hotels so that one had a view of a 2-megawatt wind turbine, and the other did not. The hotels varied by quality and their distance from the wind turbine. Installed in 2010, the turbine is approximately 400 feet tall and 300 feet wide (diameter across rotors), and visible from all three hotels.

Participants were provided with the hotel's name along with its quality (star) rating and distance from the turbine. The Virden Center is a one-star hotel, located 0.23 miles from the turbine. The Beacon Motel is a one-star hotel, located 1.58 miles away. Hotel Blue is a three-star hotel, 1.53 miles away. ¹ Participants were provided with a table explaining the amenities associated with these ratings and examples of comparable chain hotels with which they were likely familiar. Each participant bid on rooms with and without turbine views for each hotel. To illustrate the visual scale of the turbine at these distances, participants were provided with a computer-generated graphic showing the turbine at each distance compared to a three-story office building (Figure 1). The participants were not given information on the hotels’ rates, but for reference, a standard room in the Virden Center during the summer is typically $100 per night, the Beacon is approximately $120 per night, and Hotel Blue is around $250 per night.

Figure 1. Images of a Turbine at Various Distances

The graphic presented in Figure 1 was generated based on measurements of the existing wind turbine in Lewes, Delaware, scaled in a CAD program to ensure proper perspective. This visualization, while scaled accurately, certainly is limited in that it presented the view in an abstract setting instead of in the context of an actual view from a hotel room. This was done because we wanted to experimentally isolate the impact of the turbine height and distance of the view. The reality was that the views from the hotel rooms differed by more than just the height and distance from the turbine. Other research has shown that the presentation format of visuals of turbines can influence participants’ perception of them (see for instance Hevia-Koch and Ladenburg 2016). Because of this format, what is measured in this experiment can be interpreted as a tourist's expectation of the appearance of wind turbines before a hotel stay, as might influence decisions when booking, instead of the ex post value that one would have after staying in the hotel room.

Prices were randomly determined by dropping a marker onto a piece of paper with a grid of uniform random numbers between 0.01 and 23.00. If a participant's bid was above the determined price, he or she received the initial endowment minus the determined price and received the lottery ticket. If the bid was below the price, then he or she kept the initial endowment and did not receive the lottery ticket. Because only one randomly selected lottery ticket bid was binding, participants were encouraged to make each bid as if they had the full $23 to bid on that ticket. For each selected ticket, the participant had a 1-in-10 chance of winning. If the participant had purchased the ticket, they would win the item if they rolled a seven on a ten-sided die.

Participants placed bids on all six hotel rooms simultaneously. To control for order effects, the order in which the rooms were listed was reversed for half of the participants. Even though participants placed bids on all six rooms, only one bid for a room was selected at random at the end of the experiment to determine the outcome and cash earnings. This procedure of selecting one random choice is common in experimental economics (Kotani, Messer, and Schulze 2008). To help participants understand how this random selection worked, the second parts of the practice section followed these procedures with the choice between a pencil and pen (see Appendix).

After all bids were collected, the implemented hotel room choice and price were randomly determined. For those who purchased a lottery, they also rolled a die in private. If a seven was rolled, the participant won the weekend stay. All participants were paid privately to ensure all decisions made during the course of the study were confidential.

Results

Only three participants (4.6 percent) were familiar with of any of the three hotels. This ameliorates concerns about participants’ pre-familiarity with the experiment items, which was particularly important because the image of the turbine was a key control used in this field experiment. ² Among the 65 participants, only 12.3 percent bid higher for hotel rooms with turbine views. On average, bids were significantly higher for the three-star hotel and significantly lower for rooms with turbine views. The lowest average WTP ($4.88) was for the weekend stay at the Virden Center (one-star hotel) in a room with turbine views. The highest average WTP ($10.10) was at Hotel Blue (three-star hotel) in a room without turbine views.

Because the weekend stay was offered in a lottery, the magnitude of point estimates depends on assumptions about risk preferences. For example, if we assume risk neutrality, a bid of $10.10 can be interpreted as a WTP of $101.00 for a weekend at Hotel Blue without turbine views. If a participant is risk averse, $101.00 captures the lower bound of the participants’ WTP. The differential becomes larger when the curvature of the utility function, measuring risk aversion, increases. Average across rooms bids compare at $5–$10 for a 1:10 lottery, implying risk-neutral equivalents of 40–50 percent of market value. This is not unreasonable, considering that the room stays were nontransferable and good only for a relatively limited time, and that the average included a fairly high number of zero bids.

Our main interest concerns how wind turbines affect tourists’ preferences. Table 3 shows estimates of a model of participant i's bid as a function of the attributes of room j, being bid on:

(1) $$\eqalign{bid_{i,j} =\ & \beta _0 + \; \beta _1Turbine_j + \beta _2Near_j + \beta _3ThreeStar_j + \beta _4Turbine*Near_j \cr & + \beta _5Turbine*ThreeStar_j + \alpha _i + \varepsilon _{i,j}}$$

Table 3. Coefficient Estimates of Tobit Estimate of Mean WTP.

Here, turbine is a dummy variable with a value of 1, if the hotel room being bid upon has a turbine view and 0 otherwise, Near has a value of 1 for the Virden Center and 0 otherwise, and ThreeStar has a value of 1 for Hotel Blue and 0 otherwise. The α term indicates subject-level effects. Because there were large numbers of bids of either $0 or $23, this was estimated as a Tobit model.

We estimate four variations on this model – two including the interaction terms, two omitting them, and two including fixed effects and two which omit them. Results across these models are quite consistent. Beach tourists have lower WTP at the 1-percent significance level for hotel rooms with turbine views or rooms that are closer to the turbines. Intuitively, tourists express a higher WTP for a three-star hotel than a one-star hotel. Additionally, all the interaction terms are found to be statistically insignificant; ³ including them does make the estimate on the turbine view effect smaller and less significant. Assuming risk neutrality, this equates to an externality cost of around $18.40 per visit for staying at the Virden Center, which was relatively closer to the turbine, and a cost of $11.50 for having turbine views, regardless of location relative to the turbine.

Conclusions

This study, to our knowledge, is the first to use a within-subject field experiment to examine the visual externalities of wind turbines on tourism in a coastal community. The results suggest that tourists prefer a weekend stay in a hotel room without turbine views. These findings are consistent with the literature that wind turbines generate negative visual externalities to local beach communities (Ladenburg et al., 2005). While WTP differs by quality, the influences of turbine views on WTP does not.

Interestingly, 12.3 percent of the tourists had a higher WTP for hotel rooms with a turbine view even when the turbines were close (0.23 miles), indicating that wind turbines could provide positive visual externalities to some tourists. This is consistent with the findings reported by Ladenburg and Lutzeyer (2012), who found three latent classes of preferences for wind farms located at two different differences, including totally positive, totally negative, and mixed based on distance. Although our sample size does not allow us to specifically analyze such behavior, it calls for future research to bring attention to studying this group of tourists’ preferences.