Going to the Woods Is Going Home: Recreational Benefits of a Larger Urban Forest Site — A Travel Cost Analysis for Berlin, Germany

Ecological Economics 132 (2017) 255–263

Contents lists available at ScienceDirect

Ecological Economics journal homepage: www.elsevier.com/locate/ecolecon

ANALYSIS

Going to the Woods Is Going Home: Recreational Benefits of a Larger Urban Forest Site — A Travel Cost Analysis for Berlin, Germany Christine Bertram a,1, Neele Larondelle b,c,⁎,1 a b c

Kiel Institute for the World Economy, Kiellinie 66, 24105 Kiel, Germany Institute of Geography, Humboldt University Berlin, Unter den Linden 6, 10099 Berlin, Germany Potsdam Institute for Climate Impact Research, Telegraphenberg A 31, 14473 Potsdam, Germany

a r t i c l e

i n f o

Article history: Received 17 August 2015 Received in revised form 2 August 2016 Accepted 27 October 2016 Available online xxxx

a b s t r a c t We present an application of the travel cost method to a large urban forest site in Berlin, Germany. The analysis is based on a large onsite survey and the same survey administered online. Although such applications are rare in an urban context, applying a seasonal demand model to the case of Grunewald is possible because the distances travelled are relatively large, the majority of the respondents use motorized or public transport, and Grunewald is a large and unique urban forest site with very few substitutes. The main results are the following: (1) The demand for visits to Grunewald is less elastic if only Berlin residents are taken into account compared to when residents from the entire larger urban area of Berlin are considered. (2) Estimated consumer surpluses are therefore greater if only Berlin residents are taken into consideration. (3) In addition, demand is more elastic for the internet sample than for the on-site sample. (4) Results suggest a lower bound overall consumer surplus of 14.95 € per visit. The results indicate that despite its inherent limitations, non-market economic valuation through the travel cost method can provide administrations with a powerful tool to monetize the benefits of urban forest recreation to increase public funding and redirect resources to address intensified use. © 2016 Elsevier B.V. All rights reserved.

1. Introduction In an increasingly urban world (UN, 2014), the maintenance of urban green space plays an important role in safeguarding human well-being in cities (Elmqvist et al., 2013). Large forest areas, which are an important type of green space in cities, for example, offer numerous benefits to society (Dobbs et al., 2011; Ninan and Inoue, 2013; Wang and Fu, 2013). Improved air and water quality (Baumgardner et al., 2012; Larondelle et al., 2014; McPherson et al., 1997; Paoletti, 2009), the elimination of pollutants and carbon sequestration (Brack, 2002; Jansson and Nohrstedt, 2001) are just a few of examples of the regulating ecosystem services or indirect benefits that urban dwellers derive from the presence of large forest areas in and around cities. The most important direct effect of urban forest sites probably is cultural ecosystem service provisioning, and, more specifically, the benefits provided from their recreational values (Hörnsten and Fredman, 2000; Jim and Chen, 2009; Zandersen and Tol, 2009). The recreational benefits that urban forests and other green spaces provide are closely connected to their positive effect on the mental and physical health of citizens (Bratman et al., 2015; Clement and Cheng, 2011; Velarde et al., 2007). ⁎ Corresponding author at: Institute of Geography, Humboldt University Berlin, Unter den Linden 6, 10099 Berlin, Germany. E-mail address: [email protected] (N. Larondelle). 1 Both authors contributed equally to the preparation of the manuscript.

http://dx.doi.org/10.1016/j.ecolecon.2016.10.017 0921-8009/© 2016 Elsevier B.V. All rights reserved.

Even though forest areas in cities can provide multiple services, they are also threatened by urbanization and increasing population density (EASAC, 2009; Loewenstein and Loewenstein, 2005; U.S. EPA, 2000). Indicators are needed to assess the demand and supply of this valuable urban ecosystem in order to safeguard its existence. A range of biophysical and socio-ecological indicators are commonly used to assess the benefits that humans can derive from urban forest ecosystems (Dobbs et al., 2011). Additionally, several economic valuation methods have been developed to express the value that people attach to the flow of benefits from natural ecosystems in monetary terms. This can be useful information if managers have to weigh the costs and benefits of policy measures that affect environmental goods and services for which no market exists. Economic valuation of urban green in general and urban forest sites in particular has so far mostly been carried out using the contingent valuation method (e.g., Tyrväinen, 2001; Vesely, 2007) or the hedonic pricing method (Sander et al., 2010). See Gómez-Baggethun and Barton (2013) for a recent overview of the approaches used for the valuation of urban ecosystem services. Based on the ground-breaking research conducted by Clawson and Knetsch (1966) and subsequent methodological developments, the travel cost method (TCM) has commonly been used to place a value on the benefits of forests in rural areas. Examples of research where the random utility framework has been applied to select forests for recreation in rural surrounding areas Bujosa Bestard and Riera Font (2009) for forests in Mallorca, Spain, or

256

C. Bertram, N. Larondelle / Ecological Economics 132 (2017) 255–263

Termansen et al. (2013) for forests in Denmark. Examples of research where the single site TCM has been applied to value the benefits of forests in rural areas include Elsasser (1996), Garrod and Willis (1992) and Ovaskainen and Kniivil (2005). For a detailed meta-analysis, see Zandersen and Tol (2009). The TCM has, however, rarely been used for the economic valuation of urban forest sites. We are aware of only a few examples. Chaudhry and Tewari (2006) use the zonal TCM for the valuation of recreational benefits of an urban forest in India. Other examples include the application of the TCM to urban parks. Dwyer et al. (1983) for example use the TCM to estimate the willingness to pay for hypothetical entry fees to parks in Chicago, and Liu et al. (2014) use the TCM to evaluate the benefit of an urban park in Taiwan. As early as 1983 Dwyer et al. already stress that “the technique is well known among economists and recreation planners and will hopefully be applied more widely to urban forest sites” (Dwyer et al., 1983:184). The main objections against and consequently limited use of the application of the TCM in urban contexts are related to the supposedly large number of substitute sites, various means of transport, and consequently, the low costs associated with visiting recreational sites within cities (Gómez-Baggethun and Barton, 2013; Tyrväinen et al., 2005). If a lot of visitors reach the site by free means of transport, their valuation of the recreation site will not be captured by the TCM, such that the calculated consumer surpluses may be underestimated (Zandersen et al., 2012). However, it has also been mentioned that “the method is useful in a setting where large urban forests within city limits are scarce and people have to travel further to reach the areas” (Tyrväinen et al., 2005). This study demonstrates the applicability of the TCM in an urban context by applying it to a large urban forest site in Berlin, Germany. We argue that this is particularly interesting because Berlin citizens travel on average as much as 11.5 km per single distance within the vicinity of the city to the forest site, which is quite a distance compared to most cities where the distances covered are smaller. The distance that Italians travel on average to a forest site in the country amounts to 32 km and is thus not that much higher, considering that this distance refers to journeys within a whole country (Tyrväinen et al., 2005). The reason for the large distances covered by Berlin citizens is that Berlin can be regarded as a big city from an area point of view, covering an area of 891.8 km2 (by comparison: New York City: 789.4 km2, Paris: 105.4 km2) with a comparably low population density of 3872 inhabitants per km2 (by comparison: New York City: 10,560 inhabitants/ km2, Paris: 21,258 inhabitants/km2) (Amt für Statistik Berlin-Brandenburg, 2013; Institut national de la staistique et des études économiques, 2014; United States Census Bureau, 2010). Within Germany, Berlin is by far the biggest and widest city, followed by Dresden with only a third of its area (European Commission, 2011). This makes travelling long distances in the city normal for the citizens of Berlin, which is usually not the case for other German cities. Responding to the concerns about the means of transport, the majority of forest visitors in Berlin arrive by motorized vehicles or public transport. Multiple answers were allowed for this question, meaning that answers add up to N 100%. In the analyzed sample, 66% of the respondents use a car or motorbike to travel to Grunewald, and 13% use public transport as one of their means of transport; 35% of the respondents also report cycling. Cycling also incurs costs related to depreciation and maintenance even though there are no fuel costs. The share of people that report walking as their only means of transport amounts to 9.6%. Furthermore, we argue that the urban context is especially interesting for a TCM application. Urban areas are most complex socioecological systems (McPhearson et al., 2014), where the demand and supply of ecological services is often poorly balanced and underlying population and development pressures make it difficult but all the more important to value the intangible benefits of green areas (Gómez-Baggethun and Barton, 2013). Moreover, the different lifestyles and resulting modes of recreation, transportation and appreciation of green are often immense in urban settings, which make the

study of socio-ecological interactions challenging and meaningful (Coles and Bussey, 2000). With this point in mind, this study analyzes the data of a large socioecological survey on Grunewald, a large forest site in urban Berlin, using the individual TCM. The results strongly favor the applicability of the TCM within urban areas. The study shows that travel costs have a significantly negative effect on the number of visits to Grunewald, which translates into non-negligible, positive recreational values of this site. The remainder of the paper is structured as follows: Section 2 presents the case study, the data, the empirical methodology, and the uncertainties related to the analysis. Section 3 presents the results including descriptive statistics of the survey results, the regression results, and the calculations of consumer surplus per visit to Grunewald. Section 4 discusses the results, and Section 5 draws some conclusions. 2. Methods 2.1. Case Study Berlin is the capital and the biggest city in Germany. Its population is projected to increase by 245,000 to 3.75 million over the next 15 years (Senatsverwaltung für Stadtentwicklung und Umwelt, 2012). Berlin has three big forest sites within the city boundaries which serve as important recreational sites for a large proportion of its citizens. The urban forest site Grunewald covers an area of 3000 ha and is located in the south-western part of Berlin (see Fig. 1). It has been protected by a permanent contract since 1915 in order to maintain it as a recreational area and to sustain the quality of life of the citizens in a city that was already strongly growing at the beginning of the 20th century. Its territory covers an area of about 750 ha, where dogs are allowed off their leads. This is the largest area of its kind in a European city (Senatsverwaltung für Stadtentwicklung und Umwelt Berlin, 2015) and is an extraordinary pull-factor for people travelling to Grunewald. Another pull factor is the existence of several small, clear lakes ideal for swimming. Other forest areas include the forest administration TreptowKöpenick in the south-eastern part of the city, surrounding the greater Müggelsee. In comparison to Grunewald, this area is characterized by less public transport stops and no areas allowing dogs off their leads. The forest administration Tegel, which extends along the northern part of the river Havel in the north-west, is even less accessible via public transport. Both substitutes do include water access but lack smaller lakes, where for example the supervision of kids is often much easier due to the smaller area. Today, Grunewald - as many urban forest sites - has to serve many purposes and deal with conflicting interests such as providing habitat for species, producing timber, hosting environmental education programs, and, above all, providing the citizens of Berlin with one of the biggest and most varied recreational areas for multiple kinds of users. Grunewald is the perfect case study to apply the TCM in an urban context due to its sheer size, its ability to provide multiple services to people, and its unique urban character. 2.2. Survey The survey was conducted over 20 days in August 2014 (on workdays, at weekends, and during school holidays) in the field at 3 different times and at 9 different locations in the urban forest Grunewald. The date of the survey was chosen to capture summer holidays half of the time as well as regular workdays during the other half of the time. The locations were chosen in close cooperation with the local forest administration and identified beforehand as locations with a good distribution of all possible forest visitor groups. Every visitor encountered at these locations willing to take part in the survey, was asked. Additionally, the survey was available online between April 2014 and March 2015, officially distributed through the website of the forest

C. Bertram, N. Larondelle / Ecological Economics 132 (2017) 255–263

257

Fig. 1. Location of the urban Forest Grunewald, ESRI basemap: ESRI, HERE, DeLorme, MapmyIndia, OpenStreetMap contributor and the GIS user community.

administration and circulated around local forest user networks (e.g. regional sports clubs, senior clubs, school networks, social media forums). Moreover, paper notes with the online survey link were given to those claiming not to have any time to participate in the survey. The survey was explicitly addressed with an entry text to people who know and visit Grunewald, nevertheless we cannot exclude the possibility that people answered the questionnaire, who never visit Grunewald and ticked “less than once a year” as their visit frequency. Furthermore, double-counting cannot be completely excluded, even though we argue that the possibility of someone who took the 25 min-survey in the field, taking the survey again online is unlikely. This was done in order to minimize the bias of applying either only on-site sampling or only internet sampling. The content and distribution of the survey were prepared in close cooperation with the local forestry administration. All in all, 1294 survey questionnaires were collected; 839 in the field and 454 online. Due to non-responses to individual questions of the survey or invalid postal codes, we analyzed 567 responses from the field-survey and 316 online responses, resulting in a final sample size of 883 valid answers in total. Blocks of questions were asked [1] about the purpose of the visit, including how often people visit the forest, what they do, why they choose to do that in this forest and where they come from. Another block [2] included questions about awareness of climate change and forest measurements followed by [3] standard demographic questions. In this study, we mainly used blocks [1] and [3] to estimate the individual travel cost model for the urban forest site of Grunewald. The survey was originally designed to evaluate the connection between user activities, frequencies and awareness of local climate change topics. Even though it was not originally intended to cover a travel cost analysis, the case and the data showed great potential to be used for a first exploratory TCM in the complex urban surrounding of the city of Berlin. Therefore questions about income, substitute sites and multi-purpose trips were

not asked. The time of travel could nevertheless be derived as an output of the network analysis. 2.3. Network Analysis, Cost Assumptions, and Frequency of Visits 2.3.1. Network Analysis We conducted a network analysis with ArcGIS 10.2 to calculate the distances that the respondents have to travel to visit Grunewald. The center points of the postal code areas in which the respondents live were set as origin points and the 17 most commonly used entry points to Grunewald that were identified were set as destination points. Accordingly, distance was calculated from each origin (n = 343) to each destination (n = 17), which resulted in a total of 5831 routes. A route length was assigned to each respondent according to the postal code of her residence and the preferred entry point. In cases where respondents reported more than one entry point, the mean distance was calculated. Distance was multiplied by two for each respondent to account for the journey to Grunewald and back home. 2.3.2. Cost Assumptions We calculated the travel costs to reach Grunewald based on the distances derived from the network analysis. Direct travel costs should include fuel costs and depreciation, depending on the means of transport chosen (Parsons, 2013). We used a rate of 0.30 € per km for journeys by car and motorbike. These costs are based on ADAC (2013) and the German tax law, which allows tax reimbursements for an amount of 0.30 € per kilometer for motorized travel to work. For public transport, a cost of 5.20 € per round trip was assumed based on the price of two single journey tickets for the use of public transport in Berlin. For cycling, we assumed a cost of 0.06 € per km. Even though there are no fuel costs involved with cycling, it is costly due to depreciation and maintenance. Costs for cycling are based on Brühbach (2009), who

258

C. Bertram, N. Larondelle / Ecological Economics 132 (2017) 255–263

calculates costs between 0.03 € and 0.12 € per kilometer based on depreciation and annual maintenance costs depending on the annual distance travelled. Walking is also assumed not to incur any costs. If respondents stated that they used more than one means of transport, we weighed the costs of each equally to calculate travel costs. There is no entrance fee charged for accessing Grunewald. Furthermore, we do not have data on the income of the respondents, so we cannot include opportunity costs of time. The calculated travel cost in our sample ranges from zero to 49.07 € per round trip, with a mean of 4.84 € and a standard deviation of 4.58. The route length was also calculated for two alternative inner-city forest sites (Forest Spandau and Forest Müggelsee), which constitute the closest substitutes to Grunewald. These forests are located in the north-west and south-east of Berlin, respectively, and are also used for a range of recreational activities. In travel cost analysis, it is important to include the distance or travel costs to potential substitute sites in the regressions. One basic assumption of the TCM is that the value of a certain site increases with increasing distance. Thus, the further away people live from the site to be valued, the higher the travel costs, and the higher the “price” of the visit. But the further away people live from the site, the greater the probability that a substitute site is available near their home. This tends to reduce the value of the actual site being valued. Excluding distance to substitute sites thus gives rise to concerns about an upward bias in the estimated value per visit (Smith and Kaoru, 1990). We did not observe the means of transport that the respondents of our survey would have chosen if they had visited either of the substitute sites. Consequently, individual travel costs to these sites cannot be calculated. However, we directly include distance to the two substitute sites as control variables in our regressions to take substitution possibilities into account. 2.3.3. Assumptions About Frequency of Visits The annual frequency of visits was deduced from the answers to the survey. The answer options for frequency provided in the survey were given in a rank order with non-equal intervals. The frequency of visits per year thus had to be approximated from the ordinal data. According to von Grünigen and Montanari (2014), frequencies were translated as displayed in Table 1. Minima, maxima and average values were calculated in order to minimize the error and provide information for a sensitivity analysis of the results (see Section 3.2.2).

relate the number of visits that a person makes to a certain site over a certain time period to the travel costs incurred when visiting this site. The basic idea is that the costs that people face when they want to visit a recreational site constitute the “price” that they are willing to pay for the recreational use of that site. Based on the estimated demand relationship it is possible to calculate individual measures of consumer surplus derived from one visit. The basic assumption of the TCM is that the travel costs of an individual to a certain recreational site depend on several variables, i.e., the costs of driving, entrance fees to the site, and potentially the opportunity costs of the time spent travelling to the site. The individual number of visits (yi) is then a function of travel costs to site j (tcij), but also costs of travel to other substitute sites (tcis), and individual characteristics of respondent i (zi):
 yi ¼ f tcij þ tcis þ zi

ð1Þ

The estimation of this functional form is carried out using a count data model. The basic model would be the Poisson model. One basic assumption of the Poisson model is equidispersion, i.e., the equality of the mean and the variance of the number of visits. This assumption is often violated in reality. In the likely case of overdispersion, a negative binomial model can be estimated instead (see, e.g., Cameron and Trivedi, 2009). In the negative binomial model, the probability of making y visits to a site per season is given as

PrðY ¼ y; μ; α Þ ¼


  α −1  y Γ α −1 þ y α −1 μ −1 −1 −1 μ þα Γ ðα ÞΓ ðy þ 1Þ α þ μ

ð2Þ

where y is the number of visits observed, Γ(.) denotes the gamma integral, α is the variance parameter of the gamma distribution, and μ is the expected or mean number of visits made: Eðy; μ; α Þ ¼ μ

ð3Þ

In the negative binomial model, the variance of the number of visits is not equal to its mean but is modeled more generally as: Varðy; μ; α Þ ¼ μ ð1 þ αμ Þ

ð4Þ

2.4. Econometric Approach Several economic methods have been developed to value the benefits of natural sites such as forests and parks. This includes stated preference methods such as contingent valuation (CV) and choice experiments (CE) as well as revealed preference methods such as hedonic pricing (HP) and the TCM we are using in this study. See Brander and Koetse (2011) for a meta-analysis of international valuation studies that focus on different types of urban green space using in particular CV and HP. Based on revealed preference data, i.e., the actual behavior of the respondents, the TCM is used in particular for the valuation of recreational benefits. There are different types of travel cost models. Seasonal demand models, which go back to Clawson and Knetsch (1966),

Table 1 Frequency translation. Ranking

Frequency

Visits per year (max)

Visits per year (min)

Visits per year (mean)

1 2 3 4 5

Every day Several times per week Several times per month Not regularly Less than once a year

365 208 48 24 1

365 104 24 12 0

365 156 36 18 1

The constant parameter α thus reflects the degree of overdispersion in the model. In estimation, the expected number of visits is modeled as the exponential of the underlying influencing factors, i.e., μ = exp (x ′ β), where the vector x contains the explanatory variables and β represents the parameters of the model. Given this functional form, surplus values per visit are given as: ^ cc si ¼ 1=−β tc

ð5Þ

^ is an estimate of the cost parameter in the model. where β tc Seasonal demand models such as the one used in this paper are particularly helpful if the aim is to calculate surplus values per visit. In addition, it is best applicable if policies focus on a single site or only on a few sites that could serve as substitutes for one another (Parsons, 2013). The TCM assumes weak complementarity between the environmental asset and consumption expenditures (Freemann, 2003). This implies that if no visits are made to a site and travel costs are zero, then the marginal utility of the site would also be zero. The TCM therefore only captures use values, typically related to recreational use, but it does not capture non-use values or option values. The TCM also assumes that the demand for visiting a certain site for recreational purposes is independent of the demand for other leisure activities or alternative marketed non-leisure goods (Hanley and Barbier, 2009).

C. Bertram, N. Larondelle / Ecological Economics 132 (2017) 255–263

2.5. Uncertainties The assumptions on the number of visits as described in Section 2.3 can be a source of uncertainties, as well as the combination of two survey modes, an on-site survey and an online survey. Uncertainties can also arise from cost assumptions. In addition, the lowest possible answer category for the frequency of visits in both surveys was “less than once a year”. The option “never”, however, was not available so that we cannot say exactly whether our sample is truly truncated at zero or not. In Section 3.2.2, we provide robustness checks of our results with respect to these uncertainties identified. Moreover, the survey questions did not include information about whether the visit to Grunewald was the single purpose of the trip, and it did not include information about income, thus it was not possible to include opportunity costs of time in the analysis. The theoretical problem of double counting visitors using the online and on-site sample was discussed earlier (see Section 2.2). The problem of self-selection was addressed by comparing the distribution of demographic characteristics of our sample with the demographics from all over Berlin (Section 3.1). 3. Results 3.1. Descriptive Statistics of Survey Results An analysis of the demographics shows that most of the people visiting Grunewald are between 45 and 65 years old (see Table 2). Surprisingly, the group of visitors under 25 years old is very small with b5%, while people over 65 years old visit the forest more often. In terms of education, the majority of visitors held a university degree or a higher school education. The area around the forest mainly consists of upper and upper middle class settlements, thus a high share of graduates in the sample is not surprising. The distribution of household sizes reflects the trend throughout Berlin towards smaller households (Statistische Ämter des Bundes und der Länder, 2014). Compared to the general demographic pattern for Berlin, respondents in both the online and the on-site sample tend to be older and higher educated. Both characteristics, however, fit the local pattern of the neighborhoods surrounding the forest, where most of the respondents came from. We

259

therefore argue that the sample displays a good cross section of the average Grunewald visitor. Fig. 2 shows the origin of the visitors per postal code area. The map clearly shows the importance the forest site has for local recreation, as the majority of the visitors live in very close proximity to the forest; nevertheless people from all over Berlin and the larger urban area also visit the forest frequently. 159 (14.0%) respondents visit Grunewald every day, 290 (25.5%) several times per week, 263 (23.1%) several times per month, 349 (30.6%) respondents do not visit regularly, and 78 (6.8%) respondents visit Grunewald less than once a year. It is interesting to observe that N40% of the visitors come daily or several times per week. In these cases a detailed local knowledge can be assumed. Around 80% of the respondents state that they prefer Grunewald to an urban park. The majority of the respondents visit the forest for walking, dogwalking, cycling and swimming. All in all, sport-related activities such as cycling, swimming, jogging, nordic walking, and mountain biking play an important role in Grunewald. Among those answers given less often, social activities, such as meeting friends or doing something with the kids, appear relatively often. When asked for the reasons of their choice undertake these activities at the urban forest site, people answered that they find the forest visit especially relaxing and calming (63.2%), want to enjoy the beauty of the forest (54.3%), enjoy nature (36.8%), and say that it has a positive influence on their well-being (48.3%). Other reasons for visiting Grunewald entail more social motives such as meeting people, doing sports, and doing something with the kids (see Fig. 3). 3.2. Regression Results and Consumer Surplus per Visit 3.2.1. What Determines the Frequency of Visits to Grunewald? The results of the negative binomial regression for seasonal demand for recreational visits to Grunewald are presented in Table 3. The regression has been carried out using maximum likelihood estimation techniques as implemented in Stata 13. In addition to results for the full sample, which includes residents of the city of Berlin as well as respondents from the entire larger urban area of Berlin (model 1), we also report the results of the negative binomial regression for the subsample of

Table 2 Descriptive statistics and definition of main variables. The number of observations is 883 for all variables. Variable name

Definition

Mean

Gender Gender dummy; 1 if „Male“, 0 if „Female“ 0.51 Age class Age b 25 Age class dummy, 1 if “Age b 25”, 0 else 0.04 Age 25–44 Age class dummy, 1 if “Age 25–44”, 0 else 0.30 Age 45–65 Age class dummy, 1 if “Age 45–65”, 0 else 0.47 Age N 65 Age class dummy, 1 if “Age N 65”, 0 else 0.19 Education No graduation/still in school Education dummy, 1 if “No graduation/still in school”, 0 else 0.01 Normal school leaving certificate Education dummy, 1 if “normal school leaving certificate”, 0 else 0.03 Secondary school Education dummy, 1 if “secondary school”, 0 else 0.14 Highschool Education dummy, 1 if “Highschool”, 0 else 0.18 Master craftsman Education dummy, 1 if “Master craftsman”, 0 else 0.04 University Education dummy, 1 if “University”, 0 else 0.59 Education Education in 6 levels from 0 “No graduation/still in school” to 5 “University” 4.00 Household size Number of people living in the respondent's household 2.33 0.36 Internet sample Survey mode dummy, 1 if “Internet sample”, 0 if “On-site sample” Prefer Grunewald Preference for Grunewald above other green areas in the city; −1 “No”, 0 “No preference”, 1 “Yes” 0.75 Distance Grunewald Round trip distance from the postal code area of the respondent to Grunewald and back measured in km 11.59 Travel cost Travel cost to Grunewald measured in Euros 4.84 Distance Mueggelsee Distance to forest Mueggelsee measured in km 52.99 Distance Spandau Distance to forest Spandau measured in km 23.25 Transport mode Car Transport mode dummy, 1 if “Car”, 0 else 0.65 Motorbike Transport mode dummy, 1 if “Motorbike”, 0 else 0.01 Public transport Transport mode dummy, 1 if “Public transport”, 0 else 0.13 Bike Transport mode dummy, 1 if “Bike”, 0 else 0.35 Walk Transport mode dummy, 1 if “Walk”, 0 else 0.21

Min

Max

Std. dev.

0

1

0.50

0 0 0 0

1 1 1 1

0.20 0.46 0.50 0.39

0 0 0 0 0 0 0 1 0 -1 0.60 0 14.1 2.11

1 1 1 1 1 1 5 6 1 1 81.78 49.07 86.22 86.12

0.09 0.16 0.35 0.39 0.20 0.49 1.32 1.15 0.48 0.60 7.79 4.58 7.10 7.07

0 0 0 0 0

1 1 1 1 1

0.48 0.09 0.33 0.48 0.41

260

C. Bertram, N. Larondelle / Ecological Economics 132 (2017) 255–263

Fig. 2. Visitors per postal code of Berlin and surrounding areas.

Berlin residents only (model 2). Both samples used for these main regressions include respondents from the on-site and the online survey. Moreover, each sample includes respondents who have zero travel costs because they walk to Grunewald and do not use other means of transport. Robustness checks for estimations using different subsamples and assumptions on travel costs are provided in Section 3.2.2. A likelihood ratio test of overdispersion with H0: α = 0 carried out independently for both samples confirms that there is significant overdispersion in the data (χ2(1) = 8.8 e4,p = 0.000 for the full sample;

χ2(1) = 8.3 e4,p = 0.000 for the Berlin sample) such that the negative binomial model is preferred to the Poisson specification. Travel costs to Grunewald have a significantly negative effect on the number of visits in both model specifications, as expected. In terms of the distance to substitute sites, we observe different results for the two samples: In the full sample (model 1), distance to the substitute sites only has a slightly significant positive effect, and this only applies to the distance to Forest Spandau. In the Berlin sample (model 2), distance to both potential substitute forests has a significantly positive

Reasons for the forest visit

Activities in the forest promenading

it is calm and relaxing

dog walking

to enjoy the beauty of the forest

biking

for my well-being

swimming i like the forest wilderness

jogging/walking

to observe animals and plants

playing with kids

to seek a cool place iin hot summer days

meeting people

to do sports

picnic

it is the next green space from my home

picking mushrooms mointain biking

my dog can walk free

working

to teach kids about nature

geocaching to meet people

playground

the air is fresher

horseback riding

to do photography/art

flower picking relaxing

to play with kids

0%

20%

40%

60%

80%

0%

20%

Fig. 3. Reasons for the forest visit and activities in the forest; multiple answers allowed.

40%

60%

80%

C. Bertram, N. Larondelle / Ecological Economics 132 (2017) 255–263 Table 3 Results of the negative binomial regression; * p b 0.1, ** p b 0.05, *** p b 0.01. Standard errors in parentheses.

Number of visits Travel cost Distance forest Mueggelsee Distance forest Spandau Male Age b 25 (reference level) Age 25–44 Age 45–65 Age N 65 Household size Level of education Internet sample Prefer Grunewald Constant Ln(Alpha) Number of observations LR Chi2(11) Log likelihood

(1) Full sample

(2) Only Berlin residents

−0.0669*** (0.01) −0.0001 (0.01) 0.0100* (0.01) −0.1027 (0.07)

−0.0484*** (0.01) 0.0210** (0.01) 0.0296*** (0.01) −0.0842 (0.07)

0.0556 (0.19) 0.3639* (0.19) 0.4192** (0.20) 0.0429 (0.03) −0.0072 (0.03) −0.4840*** (0.08) 0.3431*** (0.06) 4.4595*** (0.42) 0.0375 (0.04) 883 128.61*** −5055.48

0.0438 (0.19) 0.3219* (0.19) 0.3725* (0.20) 0.0338 (0.03) −0.0075 (0.03) −0.4343*** (0.08) 0.3000*** (0.06) 2.9197*** (0.58) 0.0128 (0.04) 839 119.89*** −4830.62

effect on the frequency of visits. This might imply that Berlin residents actually perceive Spandau forest and Mueggelsee forest as valid substitutes that are accounted for in recreational decisions. When respondents from beyond Berlin are taken into consideration, this effect vanishes, which indicates that the two forests are not perceived as important substitutes to Grunewald for those people coming from the greater Berlin area. For the remaining variables, the effects are similar between the full model and the subsample of Berlin residents. In terms of the socioeconomic characteristics of the respondents, it can be observed that older respondents tend to visit Grunewald more often than younger respondents. In addition, respondents that prefer Grunewald to other green urban areas tend to visit the forest site more often than the other respondents. There is no statistically significant effect for gender, level of education, or household size. Moreover, the dummy that variable we included as a control for the survey mode is negatively significant. Respondents that responded to the survey via the internet thus reported lower numbers of visits to Grunewald than those respondents that were sampled on-site. 3.2.2. Consumer Surplus per Visit to Grunewald 3.2.2.1. Consumer Surplus for Main Model Specifications. Consumer surpluses per visit to Grunewald are calculated based on Eq. (5) and the results from the regressions presented in Section 3.2.1. The consumer surplus amounts to 14.95 € per visit for the full sample and to 20.66 € for the Berlin sample. For the Berlin sample, travel costs are slightly lower than for the full sample, ranging from zero to 27.66 € with a mean of 4.54 € (standard deviation = 4.07). Seemingly, the demand elasticity for visits to Grunewald differs substantially between the two samples. For the full sample, the demand for visiting Grunewald is more elastic than for the Berlin sample, which translates into smaller consumer surpluses for the full sample compared to Berlin residents only. For the subsample of Berlin residents, a larger change in travel

261

costs would thus have to occur to induce the same change in the demand for visits to Grunewald. Berlin residents can therefore be said to place a higher value on visits to Grunewald and are less willing to reduce their visitation rates even if costs increase. A slightly different interpretation would be that substitution possibilities could be more readily available to people from the more rural areas around Berlin than for the people living within city boundaries, which would also explain the different demand elasticities and resulting consumer surpluses. 3.2.2.2. Robustness Checks. As outlined in Section 2.3, we had to deduce the frequency of visits from ordinal visitation data. For robustness checks, we thus also calculated consumer surpluses for maximum and minimum assumptions on the frequency of visits (see Table 1). Changes in consumer surpluses are relatively low: For the minimum assumption, consumer surplus amounts to 13.45 € per visit for the full sample and to 19.67 € per visit for the Berlin sample. For the maximum assumption, consumer surplus amounts to 16.05 € per visit for the full sample and to 21.33 € per visit for the Berlin sample. As discussed in Section 2.5, we potentially oversample the respondents that visit Grunewald more often, which could bias surplus estimates. Using a model with endogenous stratification for the medium visit number assumption for the Berlin sample, we find that the consumer surplus per visit stays almost constant at 20.38 € per visit. Effect sizes and significance levels of the other parameters remain constant when the model with endogenous stratification is used. In addition, we assume positive costs amounting to 0.06 € per km for cycling as discussed in Section 2.3. This assumption might be criticized with the argument that maintenance costs are negligible and fuel costs are not relevant. We therefore also carried out the calculations of consumer surplus assuming that the costs for cycling are zero. In this case, the range of travel costs for the whole sample remains constant (zero to 49.07 €) but the mean slightly decreases from 4.54 € to 4.53 €. Using these cost assumptions and the medium visit number assumption, the estimated consumer surplus increases to 17.33 € per visit for the full sample and to 27.76 € per visit for the Berlin sample. In a further robustness check, we exclude all 85 respondents who have zero travel costs because they walk to Grunewald and do not use other means of transport. As for the CS per day, estimates are virtually unchanged compared to when these respondents are included in the sample. For the sample including the residents of the city of Berlin and the greater Berlin area, CS per visit amounts to 15.06 €. The CS for Berlin residents only amounts to 20.05 € per visit. Furthermore, additionally excluding all respondents that only use their bike to reach Grunewald substantially reduces the consumer surplus estimated. For the sample including the residents of the city of Berlin and the greater Berlin area, CS per visit then amounts to 12.64 €, whereas the CS for Berlin residents then only amounts to 14.76 € per visit. Since we observe a significant effect of the survey mode, we also calculated consumer surpluses separately for the internet sample (N = 316) and for the on-site sample (N = 567), including both respondents from the city of Berlin as well as respondents from the entire larger urban area of Berlin. The results reveal that consumer surplus for the internet sample amounts to 8.93 € while it amounts to 23.09 € for the onsite sample. This indicates that the elasticity of demand for visits to Grunewald is substantially larger for those respondents sampled over the internet than for those respondents sampled on-site. 4. Discussion In this paper, we applied the TCM to a large urban forest site, Grunewald, in Berlin, Germany. The results show an overall significantly positive and non-negligible consumer surplus related to the recreational use of this urban forest site. The magnitudes of the estimated consumer surpluses vary with the subsamples considered but are almost unchanged for robustness checks regarding the underlying assumptions

262

C. Bertram, N. Larondelle / Ecological Economics 132 (2017) 255–263

of the travel cost method. This study therefore emphasizes the applicability of the travel cost method in urban surroundings when distances travelled are large, a significant share of the people use motorized or public transport, and the method is applied to large and unique areas with limited substitutes. This is supported by the observation that the recreational value of the urban forest site derived using the TCM in our application is comparable to results from TCM analyses applied in rural areas. Comparing our estimates of consumer surplus to the results of such studies, we find that our estimates take on intermediate values. Two examples of results from travel cost studies referring to recreation in German forests include Elsasser (1996), who reports a mean consumer surplus of 9.87 €2000, and Löwenstein (1991), who reports a mean consumer surplus of 51.89 €2000. Results differ among the countries where the study was carried out. Most consumer surpluses calculated for the UK, for example, are lower than our estimates while those for Sweden and Finland tend to be higher (see Zandersen and Tol (2009) for a meta-analysis). Referring to concerns voiced by one reviewer, we would like to discuss shortly the role of opportunity costs of time, which are usually considered in travel cost studies. In our application, travel time might play an important role in determining the demand for Grunewald visits, particularly when visits during the week are considered when people typically work long hours and might be constrained on their leisure time. In addition, time costs could be a relevant fraction of total travel costs for people that use bikes as means of transport in which case direct travel costs are rather small. In the applied travel cost literature, it is common that a certain share of the monthly wage rate is used as a proxy for time costs. An alternative way would be to consider travel time as a separate variable in the regressions in addition to direct travel costs (Parsons, 2013). In our application, both ways were not possible since sufficiently precise information on income was not available and travel time would be highly correlated with direct travel costs. The consumer surplus estimates might thus be considered conservative lower bound estimates. Based on the significant positive consumer surpluses estimated, this analysis provides a good basis for argumentation for forest managers and conservationists aiming to keep Grunewald as one of the biggest recreational sites in Berlin, providing multiple benefits to its increasing number of visitors. To highlight this importance, we project an annual visit number for Grunewald based on the median frequency of visits in our sample and the total population of Berlin. We use the median so as not to overstate the number of visits as the median (36 visits per year) is lower than the weighted mean (81 visits per year). Multiplying the median number of visits with the total population of Berlin (3.5 million), we project 126 million visits to Grunewald per year. Note, however, that this calculation is based on the assumption of having a representative sample of the Berlin population. On the one hand, this is probably not the case so that the total annual number of visits might be overstated. However, the projection is only slightly higher than an older estimate which amounted to 100 million visits per year back in 1995 (Meierjürgen, 1995) and is still used by forest officials today. Using this projected annual number of visits however to calculate an annual consumer surplus created by the urban forest of Grunewald is delicate. As we stated earlier the limitations of the study do not allow us to assume that we worked with a representative sample. Given these limitations in addition to the uncertainties regarding visitation numbers, we do not recommend scaling up consumer surplus. An annual number of visits of around 100 million visits imply a very intensive use of the forest Grunewald and a very high recreational value. This intensive use brings a considerable amount of user conflicts and management challenges to maintain the ecosystem's stability as well as ongoing recreational values for visitors (Coles and Bussey, 2000; Konijnendijk, 2003; Nielsen and Møller, 2008). This is even more the case as the urban forest site Grunewald was awarded “forest region of the year” in 2015 by the German national forest association. Consequently, visitor numbers are expected to increase as press coverage

widens. It is therefore even more important to express the large value which Grunewald offers to recreational users.

5. Conclusions This paper has demonstrated the applicability of the TCM in an urban context. The lower bound consumer surplus for the whole sample amounts to 14.95 € per visit in the baseline specification. It has to be noted, however, that the demand for visits to Grunewald is less elastic if only Berlin residents are taken into account compared to when residents from the larger urban area of Berlin are considered. Consequently, consumer surplus would be greater if only Berlin residents were considered. Similarly, demand is more elastic for the internet sample than for the on-site sample. Besides other important considerations when quantifying the benefits that an urban forest provides for citizens, such as quantifying its cooling potential or assessing its value as habitat for animal and plant species using biophysical indicators, economic valuation can play an important role in providing information on people's preferences and their valuation of an urban forest as recreational area. Values such as those calculated here with the travel cost method can give urban forest administrations powerful arguments when challenged by decreasing public funds for more incurring tasks due to intensified use by an increasing population. Nevertheless it needs to be stressed that, even more perhaps with increasing visitor numbers, these values will always capture only a small range of the benefits that an urban forest provides. In particular, the TCM is only able to capture direct use values related to recreational use. Indirect use values brought about by regulating ecosystem services or non-use values would have to be considered in addition to gain a full picture of the benefits that an urban forest site like Grunewald provides for the residents of a city.

Acknowledgements The first part of the title of this manuscript is from John Muir's “Our National Parks”. We thank two anonymous reviewers for helpful comments and our colleague André Mascarenhas for remarks on an earlier version of the paper. The German Federal Ministry of Education and Research provided welcome financial support through the project “Urban Biodiversity and Ecosystem Services” (URBES; 01LC1101A), funded within the European BiodivERsA framework, and WAHYKLAS, funded within the German Waldklimafonds, 28W-C-4-031-01.

References ADAC (Allgemeiner Deutscher Automobil-Club, 2013. ADAC Autokosten 2013. Amt für Statistik, Berlin-Brandenburg, 2013. Statistik Berlin-Brandenburg [WWW document]. URL. http://www.statistik-berlin-brandenburg.de/home.asp. Baumgardner, D., Varela, S., Escobedo, F.J., Chacalo, A., Ochoa, C., 2012. The role of a periurban forest on air quality improvement in the Mexico City megalopolis. Environ. Pollut. 163:174–183. http://dx.doi.org/10.1016/j.envpol.2011.12.016. Brack, C.L., 2002. Pollution mitigation and carbon sequestration by an urban forest. Environ. Pollut. 116, 195–200. Brander, L.M., Koetse, M.J., 2011. The value of urban open space: meta-analyses of contingent valuation and hedonic pricing results. J. Environ. Manag. 92, 1763–2773. Bratman, G.N., Daily, G.C., Levy, B.J., Gross, J.J., 2015. The benefits of nature experience: improved affect and cognition. Landsc. Urban Plan. 138:41–50. http://dx.doi.org/10. 1016/j.landurbplan.2015.02.005. Brühbach, D., 2009. Ökonomische Vorteile Des Fahrradfahrens für Unternehmen. http:// www.fahrrad-fit.de/default.asp?Menue=225. Bujosa Bestard, A., Riera Font, A., 2009. Environmental diversity in recreational choice modelling. Ecol. Econ. 68:2743–2750. http://dx.doi.org/10.1016/j.ecolecon.2009.05. 016. Cameron, A.C., Trivedi, P.K., 2009. Microeconometrics using Stata. Stata Press, College Station, Texas, USA. Chaudhry, P., Tewari, V., 2006. A comparison between TCM and CVM in assessing the recreational use value of urban forestry. Int. For. Rev. 8:439–448. http://dx.doi.org/10. 1505/ifor.8.4.439. Clawson, M., Knetsch, J., 1966. Economics of Outdoor Recreation. The John Hopkins University Press, New York.

C. Bertram, N. Larondelle / Ecological Economics 132 (2017) 255–263 Clement, J.M., Cheng, A.S., 2011. Using analyses of public value orientations, attitudes and preferences to inform national forest planning in Colorado and Wyoming. Appl. Geogr. 31:393–400. http://dx.doi.org/10.1016/j.apgeog.2010.10.001. Coles, R.W., Bussey, S.C., 2000. Urban forest landscapes in the UK - progressing the social agenda. Landsc. Urban Plan. 52:181–188. http://dx.doi.org/10.1016/S01692046(00)00132-8. Dobbs, C., Escobedo, F.J., Zipperer, W.C., 2011. A framework for developing urban forest ecosystem services and goods indicators. Landsc. Urban Plan. 99:196–206. http://dx.doi.org/10.1016/j.landurbplan.2010.11.004. Dwyer, J.F., Peterson, G.L., Darragh, A.J., 1983. Estimating the value of urba forests using the travel cost method. J. Arboric. 9, 182–195. EASAC, 2009. Ecosystem Services and Biodiversity in Europe. EASAC Policy Report 09. Elmqvist, T., Fragkias, M., Goodness, J., Güneralp, B., Marcotullio, P.J., McDonald, R.I., Parnell, S., Al., E., 2013. Global Urbanisation, Biodiversity and Ecosystem Services. Challenges and Opportunities. Springer. Elsasser, P., 1996. Der Erholungswert des Waldes : monetäre Bewertung der Erholungsleistung ausgewählter Wälder in Deutschland. (Dissertati). Schr Forstökon Vol. 11. EPA, U.S., 2000. Projecting Land-Use Change: A Summary of Models for Assessing the Effects of Community Growth and Change on Land-Use Patterns. European Commission, 2011. Mapping Guide for a European Urban Atlas. Freemann, M., 2003. Economic valuation: what and why. A Primer on Nonmarket Valuation. Springer, pp. 1–25. Garrod, G.D., Willis, K.G., 1992. Valuing goods' characteristics: an application of the hedonic price method to environmental attributes. J. Environ. Manag. 34, 59–76. Gómez-Baggethun, E., Barton, D.N., 2013. Classifying and valuing ecosystem services for urban planning. Ecol. Econ. 86:235–245. http://dx.doi.org/10.1016/j.ecolecon.2012. 08.019. Hanley, N., Barbier, E.B., 2009. Pricing Nature - Cost-Benefit Analysis and Environmental Policy. Edward Elgar Publishing Limited, Cheltenham, UK. Hörnsten, L., Fredman, P., 2000. On the distance to recreational forests in Sweden. Landsc. Urban Plan. 51:1–10. http://dx.doi.org/10.1016/S0169-2046(00)00097-9. Institut national de la staistique et des études économiques, 2014e. Ile de France - Paris Statistics. Jansson, A., Nohrstedt, P., 2001. Carbon sinks and human freshwater dependence in Stockholm County. Ecol. Econ. 39, 361–370. Jim, C.Y., Chen, W.Y., 2009. Leisure participation pattern of residents in a new Chinese City. Ann. Assoc. Am. Geogr. 99:657–673. http://dx.doi.org/10.1080/ 00045600903066482. Konijnendijk, C.C., 2003. A decade of urban forestry in Europe. For. Policy Econ. 5: 173–186. http://dx.doi.org/10.1016/S1389-9341(03)00023-6. Larondelle, N., Haase, D., Kabisch, N., 2014. Mapping the diversity of regulating ecosystem services in European cities. Glob. Environ. Chang. 26:119–129. http://dx.doi.org/10. 1016/j.gloenvcha.2014.04.008. Liu, W., Chiou, C., Lin, J., 2014. Application of the travel cost method to measure the urban forest park' s recreational value with first-visit satisfaction taken into account. J. Food Agric. Environ. 12, 1013–1024. Loewenstein, N.J., Loewenstein, E.F., 2005. Non-native plants in the understory of riparian forests across a land use gradient in the southeast. Urban Ecosyst. 8, 79–91. Löwenstein, W., 1991. Die Monetäre Bewertung der Fernerholung im Südharz mit der Reisekostenmethode. In: Bergen, V., Brabänder, H.D., Bitter, A.W., Löwenstein, W. (Eds.), Beiträge zum Symposium-Monetäre Bewertung der landeskulturellen Leistungen der Forstwirtschaft, first ed. J.D.Sauerländer's Verlag, Göttingen, p. 162. McPhearson, T., Andersson, E., Elmqvist, T., Frantzeskaki, N., 2014. Resilience of and through urban ecosystem services. Ecosyst. Serv. 1–5. http://dx.doi.org/10.1016/j. ecoser.2014.07.012.

263

McPherson, E.G., Nowak, D., Heisler, G., Grimmond, S., Souch, C., Grant, R., Rowntree, R., 1997. Quantifying urban forest structure, function, and value: the Chicago urban forest climate project. Urban Ecosyst. 1, 49–61. Meierjürgen, U., 1995. Freiraumerholung in Berlin – Die Entwicklung Eines Forstpolitischen Ansatzes für Eine Raumplanung Des Stadtnahen Erholungswaldes. University Freiburg. Nielsen, A.B., Møller, F., 2008. Is coppice a potential for urban forestry? The social perspective. Urban For. Urban Green 7:129–138. http://dx.doi.org/10.1016/j.ufug.2008.02. 005. Ninan, K.N., Inoue, M., 2013. Valuing forest ecosystem services: what we know and what we don't. Ecol. Econ. 93:137–149. http://dx.doi.org/10.1016/j.ecolecon.2013.05.005. Ovaskainen, V., Kniivil, M., 2005. Consumer versus citizen preferences in contingent valuation : evidence on the role of question framing ∗. Aust. J. Agric. Resour. Econ. 49, 379–394. Paoletti, E., 2009. Ozone and urban forests in Italy. Environ. Pollut. 157, 1506–1512. Parsons, G.R., 2013. Travel Cost Methods. In: Shogren, J.F. (Ed.), Encyclopedia of Energy, Natural Resource, and Environmental Economics. Elsevier, Amsterdam, pp. 349–358. Sander, H., Polasky, S., Haight, R.G., 2010. The value of urban tree cover: a hedonic property price model in Ramsey and Dakota Counties, Minnesota, USA. Ecol. Econ. 69:1646–1656. http://dx.doi.org/10.1016/j.ecolecon.2010.03.011. Senatsverwaltung für Stadtentwicklung und Umwelt, 2012. Population Prognosis for Berlin and Districts 2011–2030 (Berlin). Senatsverwaltung für Stadtentwicklung und Umwelt Berlin, 2015. Baden und Hundeauslauf an Schlachtensee und Krumme Lanke. Smith, V.K., Kaoru, Y., 1990. Signals or Noise ? Explaining the Variation in Recreation Benefit Estimates. Am. Agric. Econ, Assoc. Statistische Ämter des Bundes und der Länder, 2014. Region and Population - Household [WWW Document]. (online source. URL http://www.statistik-portal.de/statistikportal/de_jb01_jahrtab4.asp, accessed 5.12.15). Termansen, M., Mcclean, C.J., Søndergaard, F., 2013. Modelling and mapping spatial heterogeneity in forest recreation services. Ecol. Econ. 92:48–57. http://dx.doi.org/ 10.1016/j.ecolecon.2013.05.001. Tyrväinen, L., 2001. Economic valuation of urban forest benefits in Finland. J. Environ. Manag. 62:75–92. http://dx.doi.org/10.1006/jema.2001.0421. Tyrväinen, L., Pauleit, S., Seeland, K., de Vries, S., 2005. Benefits and uses of urban forests and trees. In: Konijnendijk, C.C., Nilsson, K., Randrup, T.B., Schipperijn, J. (Eds.), Urban Forests and Trees, pp. 81–114. United Nations, Department of Economic and Social Affairs, P.D, 2014,. World Urbanization Prospects: The 2014 Revision, Highlights (ST/ESA/SER.A/352). United States Census Bureau, 2010. New York City Census Data. Velarde, M.D., Fry, G., Tveit, M., 2007. Health effects of viewing landscapes – landscape types in environmental psychology. Urban For. Urban Green. 6:199–212. http://dx. doi.org/10.1016/j.ufug.2007.07.001. Vesely, É., 2007. Green for green : the perceived value of a quantitative change in the urban tree estate of New Zealand. Ecol. Econ. 3:605–615. http://dx.doi.org/10.1016/ j.ecolecon.2006.12.019. von Grünigen, S., Montanari, D., 2014. Erholung im Schweizer Wald: monetärer Wert und Determinanten. Schweiz. Z. Forstwes. 165:113–120. http://dx.doi.org/10.3188/szf. 2014.0113. Wang, S., Fu, B., 2013. Trade-offs between forest ecosystem services. For. Policy Econ. 26: 145–146. http://dx.doi.org/10.1016/j.forpol.2012.07.014. Zandersen, M., Tol, R.S.J., 2009. A meta-analysis of forest recreation values in Europe. J. For. Econ. 15:109–130. http://dx.doi.org/10.1016/j.jfe.2008.03.006. Zandersen, M., Bartczak, A., Czajkowski, M., Giergiczny, M., Termansen, M., 2012. Guide on Economic Instruments & Non-market Valuation Methods. POLFOREX, Warsaw.