Forest owners' willingness to accept contracts for ecosystem service provision is sensitive to additionality

Ecological Economics 113 (2015) 15–24

Contents lists available at ScienceDirect

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

Analysis

Forest owners' willingness to accept contracts for ecosystem service provision is sensitive to additionality Suzanne Elizabeth Vedel a,⁎, Jette Bredahl Jacobsen a,b, Bo Jellesmark Thorsen a,b a b

Department of Food and Resource Economics, Faculty of Science, University of Copenhagen, Rolighedsvej 23, DK-1958 Frederiksberg C, Denmark Centre for Macroecology, Evolution and Climate, University of Copenhagen, Rolighedsvej 23, DK-1958 Frederiksberg C, Denmark

a r t i c l e

i n f o

Article history: Received 2 April 2014 Received in revised form 28 January 2015 Accepted 13 February 2015 Available online xxxx Keywords: Payment for environmental services Biodiversity Groundwater Recreational access

a b s t r a c t A key prerequisite to ensure that payment for ecosystem services is effective is that the management measures landowners are paid to undertake are in fact additional to the status quo and hence bring about a change in provision. We investigated Danish forest owners' preferences for conditional contracts for the provision of ecosystem services in Natura 2000 policies in a sample covering 12.5% of the total private forest area. This involves allowing old trees to decay naturally, setting aside forest areas, accepting a fixed percentage of broadleaves and increasing access for the public. Forest owners may already provide some of these, e.g., if they derive private benefits from them, in which case additionality becomes an issue. This study investigates the link between forest owners' current management and their willingness to accept (WTA) payments for providing specific ecosystem services by eliciting current practice prior to a choice experiment on contracts. For most of these ecosystem services, owners differentiate their WTA significantly according to their current management. Owners who did not provide extended access had a mean WTA of €14/ha/year for accepting access up to 15 m from roads and paths and €28/ha/year for accepting access everywhere in their forest. However, forest owners who already allow extended access have a mean WTA around zero. © 2015 Elsevier B.V. All rights reserved.

1. Introduction The enhanced provision of ecosystem services from forests relies to a great extent on the forest management decisions made by private forest owners. In many European countries, a large percentage of the forest area is privately owned. Changes in private forest owners' management practices can be influenced by changes to the legal framework, e.g., demanding specific changes on private land. However, in recent years, focus on the potential of economic instruments to bring about efficient and legitimate changes in forest management has grown, particularly payments for ecosystem services1 (PES, see, e.g., Engel et al., 2008; Wunder et al., 2008). According to Wunder et al. (2008), a PES contract is: (a) a voluntary transaction where (b) a well-defined environmental service or land use likely to secure that service (c) is ‘bought’ by at least one service buyer, (d) from at least one service provider, (e) on the condition that the service provider secures service (or management change) provision. Wunder et al. (2008) also stress that for the performance of a PEStype contract to be efficient, it is important that the actions or services paid for are indeed additional in the sense that they have not already ⁎ Corresponding author. E-mail address: [email protected] (S.E. Vedel). 1 In the literature, payment for ecosystem services and payment for environmental services are both used extensively, but appear to refer largely to the same concepts and ideas.

http://dx.doi.org/10.1016/j.ecolecon.2015.02.014 0921-8009/© 2015 Elsevier B.V. All rights reserved.

been carried out (provided), or would have been carried out (provided) in the absence of the contract. While the latter is difficult to observe in a counterfactual way in any specific case, the former may be observable to some extent ex ante, at least if information is not inherently asymmetric. The research question addressed here is as follows: By obtaining information on forest owners' current management practices, is it possible to obtain ex ante estimates of the WTA for specific management changes that take into account the additionality aspect? Preference heterogeneity among forest owners is well documented, also in Denmark (Boon et al., 2004), and it is likely that some forest owners will derive private benefits from the management changes and the resulting changes in the forest ecosystem and services provided which we investigate here (untouched forest, allowing trees to decay naturally, increasing area with broadleaves and granting access). Therefore, their net loss is below the direct opportunity costs and financial losses, which may have an effect on their WTA and will certainly influence information rents when flat rate payment schemes are applied (Wunder et al., 2008; Juutinen et al., 2013) and information is private (Latacz-Lohman and van der Haamsvoort, 1997; Juutinen et al., 2013). A few papers examine evidence for preference heterogeneity in real conservation auctions and the related efficiency aspects (e.g., Vukina et al., 2008). Juutinen et al.'s study (2013) is the study which most closely resembles the current paper, as they calculate the likely information rents using a small sample of actual bids for the conservation of a small number of stands, and their likely financial value. Several studies have shown that

16

S.E. Vedel et al. / Ecological Economics 113 (2015) 15–24

owners who have expressed environmental preferences or biodiversity concerns have reduced compensation claims when entering conservation contracts (Mäntymaa et al. 2009; Rabotyagov and Lin, 2013). More broadly, preference heterogeneity has been documented in both stated and revealed preference studies, which highlights the scope for potential targeting of contracts (Wilson and Hart, 2000; Vanslembrouck et al., 2002; Hudson and Lusk, 2004; Hackl et al., 2007; Ruto and Garrod, 2009; Broch and Vedel, 2012; Juutinen et al., 2008; Juutinen and Ollikainen, 2010), although none have concerned themselves with linking WTA measures with landowners' current management activities. Those who have come closest are Broch and Vedel (2012) who estimated the effect of already having forest on the property on farmers' WTA for afforestation schemes. In this study we investigated forest owners' willingness to undertake specific management actions to enhance ecosystem services, with or without payments, on their property. We used a questionnaire including a choice experiment (CE) replication of contract choices. The ecosystem services chosen here, and the related forest management measures, are based on on-going policy debates in Denmark regarding access, biodiversity protection, e.g., in relation to NATURA 2000 initiatives, and the provision of clean groundwater for drinking water purposes. Prior to the choice experiment, we elicited information about the forest owners' current forest status and management practices, and we also elicited their own statements about current activities pertaining to the attributes of the CE. The CE method has previously been applied to elicit landowners' preferences for policy initiatives regarding nature conservation on private land. Although the CE method has been applied in a vast amount of environmental valuation studies, its application in eliciting landowners' preferences for contracts is more limited (Hudson and Lusk, 2004; Horne, 2006; Ruto and Garrod, 2009; Espinosa-Goded et al., 2010; Broch and Vedel, 2012; Vedel et al., 2015). The paper is structured as follows: first the econometric method and data collection are presented. A presentation of the results follows including a discussion about any potential interactions related to the forest owners' current management practices. This is then followed by the discussion and finally the conclusion. 2. Econometric method Choice experiments are based on the random utility model (McFadden, 1973) and Lancaster's consumer choice theory (Lancaster, 1966). For a more detailed description of choice modeling in general, see Train (2009), as we only describe it briefly here. We assume that a respondent will choose an alternative k over another j, given that the former provides greater utility. The utility depends on the attributes of a specific alternative, but may also vary depending on the individual's characteristics, which in our case are described by the forest owner's current forest management practices. We gather variables describing all of these components in the vector x. To allow for heterogeneity between respondents, we model the choices using a random parameter logit model allowing for multiple choices of each respondent, and take advantage of the panel structure of the N choice sets on each individual which allows us to estimate individual βs. Thus, the probability that the i'th respondent will produce the sequence of N observed choices is: Z Pr ¼

0

2

 0  31 exp β x i ikn @ ∏ 4X  5Aϕðβjb; W Þd β J 0 n¼1 exp βi xi jn j N

ð1Þ

where we assume a constant scale and ϕ(β|b, W) is a normal distribution function for β, with mean b and covariance W. Furthermore, we allow for a zero-mean error component (σ) with spread in order to capture the uncertainty associated with choosing the status quo as opposed to one of the alternative contracts (Greene and Hensher, 2007; Ferrini and Scarpa, 2007).

We calculate the marginal rate of substitution (WTA) using the Delta method (Greene, 2002), assuming no distribution around the price parameter. This is common practice to avoid the problem of identifying the distribution of a parameter calculated as the ratio between two random parameters. We tested models assuming a log-normal distribution around the price parameter and found that other parameters remained similar, as did the mean price parameter. Thus, the potential bias derived from assuming a fixed price is unproblematic. In our model, we include variables which describe the current forest status and forest management practices related to the main attributes of the conservation contract alternatives. We interacted these variables with the relevant main attribute variables to obtain estimates of their effect on the stated WTA. Our general hypothesis is that a forest owner who has already set aside part of his/her forest as untouched or fully accessible, will reveal a different stated WTA from those who have not. Specifically, we would expect the WTA to be lower than for forest owners who have not already implemented such actions, but not necessarily zero, as the contract implies a restriction on future actions. We note that in Denmark, existing menus of contracts for payment for ecosystem services apply flat rates; however, site and context-specific opportunity cost calculations for individual measures are also applied, which reduces the scope for rents at poor sites with low opportunity costs, and enhances the incentive compatibility of our experimental context. 3. Data collection Forest owners were sampled based on contact details obtained from the Danish National Forest Inventory, which applies a spatial grid to systematically select a random set of measurement plots which is therefore representative of the Danish forest area (Söderberg and Johannsen, 2000). This also allows us to identify a random set of forest owners, although it should be noted that the probability of being on the list depends on the amount of forest owned, i.e. owners of larger forest estates appear on the list with several plots. This allowed us to contact a random set of all forest owners, rather than relying on membership of forestry or agricultural organizations, or previous participation in subsidy schemes. From the total sample of contact details, a random set of forest owners was selected across a systematic stratification based on the number of times a forest owner appeared in the data, with owners of larger forests appearing more often due to the spatial grid applied. Thus, we oversampled the population of larger forest owners. As in several other European countries, forest ownership is characterized by extremely uneven property sizes. There are more than 25,000 forest owners in Denmark, but more than 20,000 own forest areas smaller than 10 ha. Yet, the largest 500 owners together own more than 50% of the total of private forests in the country. This implies a challenge for a study like this as these owners of large areas are very interesting from a policy perspective; however, randomly drawing the addresses may well result in a sample which only represents a small part of the forest area. Therefore, we decided to oversample the population of owners of large forests by allowing the probability of being selected to depend on the number of plots in the National Forest Inventory. The data were collected through an online survey using the software SurveyXact in the period June–August 2012. The forest owners were contacted by letter which invited them to participate in the survey and included a brief explanation as to how potential respondents had been identified. The letter also included statements of support for the survey from the Director of the Danish Forest Association and the Director of the Danish Forest Owners Association. Furthermore, the letter contained a leaflet with brief information about the survey and the name of the website, where they could log-on to fill out the questionnaire online. The forest owners could also contact us and ask for a paper version of the questionnaire (three forest owners did this — one questionnaire was returned completed). Respondents were enticed to take part by offering them entry into a lottery if they completed the survey. The owners of larger forests had a higher probability of winning

S.E. Vedel et al. / Ecological Economics 113 (2015) 15–24

a gift voucher (one in ten), while for the owners of smaller forests, seven gift vouchers were randomly distributed between those who had completed their questionnaires. The respondents in each sample were informed about the number of owners potentially entering the lottery. The total value of the gift vouchers was approximately €400, which could be spent at grocery stores across the country. In total, 1429 forest owners were contacted by letter with owners who had not filled out the questionnaire being sent a reminder by post three to four weeks later. These efforts altogether resulted in a sample of forest owners who, in total, own 12.5% of the private Danish forest area. The size of the sample makes our data very policy relevant, though we should stress that they are not likely to be statistically representative in a random sample sense. However, we can correct for differences in, e.g., property size and other aspects. The design of the questionnaire was based on experience from earlier studies on forest owners and other types of landowners (Boon et al., 2004; Broch and Vedel, 2012). We tested the questionnaire in a focus group of forest owners which resulted in a redesign of parts of the questionnaire. The change especially concerned questions connected with costs and income related to the forest, which was considered burdensome for forest owners to report on. The attributes were selected to align with current national policy debates in relation to Natura 2000, regulations in certification schemes and ecosystem services currently being implemented nationally (Jacobsen et al., 2013), and discussed with the focus groups. When accepting a requirement on the percentage of broadleaves, areas with conifers should be converted to broadleaved areas when a stand has reached maturity — until the specified percentage of broadleaves has been reached. Increased access to private forest areas for the public has been debated with NGOs advocating either permission to walk up to a maximum of 15 m off roads and paths in the forest, or access on foot everywhere. At present, the public are allowed to walk anywhere in publicly owned forests, while they are allowed to walk and cycle on forest roads and paths from 6 o'clock to sunset in privately owned forests (larger than 5 ha). The questionnaire was structured as follows: First it asked questions regarding the respondent's forest and property including harvest, administration and decision-making related to forest management. After this followed questions regarding previous experiences with subsidy schemes and issues which may have affected their motivation for participating. Then the CE was introduced and each forest owner was presented with 6 choice sets including 2 alternatives and a status quo option (see example of choice set in Appendix A). The final part of the questionnaire consisted of socio-economic questions regarding the respondent, their household and membership of organizations. Throughout the questionnaire, the respondent had the opportunity to go back to previously answered pages in the questionnaire if they felt they needed to revise their responses at a later stage. The attributes and levels were described in detail just prior to the choice experiment. They are summarized in Table 1. The ecosystem services which the specific attribute promotes were stated briefly to avoid inducing certain opinions in forest owners, e.g., ‘the aim of allowing

17

5 trees per/ha to age and decay naturally is to increase biodiversity’. All attributes were presented along with the various levels associated with them — except for ‘lower property tax’ where the levels were omitted in order to prevent an anchoring effect. The payment vehicle, lower property tax, is described as a permanent reduction in the property tax per ha per year for the whole property. Respondents were told that it would be registered in the deed and would persist perpetually, along with the other elements in the contract. This is how existing permanent conservation measures are dealt with (e.g., existing contracts for leaving an area untouched or allowing trees to decay naturally are perpetually binding). Using reduced property taxes as an incentive to provide ecosystem services has been part of recent policy debates and is, therefore, likely to be an acceptable payment vehicle. The levels were chosen to have realistic ranges compared with current property tax ranges in place (which typically vary between 100 and 200 DKK/ha per year, depending on the site class). The level of compensation was both below and above current levels. 3.1. Design The design for the CE was optimized in NGene 1.0.2 for DB-efficiency. A MNL model with the main effects was assumed for the design using zero priors and no interactions. Without a priori information on distributions of random parameters in a random parameter logit (RPL) model, this was chosen as the best option. The attributes access and broadleaves were dummy-coded, whereas untouched forest area, allowing old trees to decay naturally and price were continuously coded. The final design contained 24 choice sets divided into 4 blocks, which resulted in 6 tasks to be answered by each respondent. One level in one choice set (0 to 5 for old trees to decay naturally) was changed to avoid having an alternative identical to the status quo. Within the same block, a 5 was changed to a 0 (old trees to decay naturally) in order to maintain the balance within the block. The D-error at the generation stage was 0.001244. The posterior D-errors for the two models applied here are 0.01908 (main effects model) and 0.03199 (model with interactions). 3.2. Response rate In total, 1429 people received the letter inviting them to participate in the questionnaire. In the final sample used for the data analyses, we chose to eliminate 4 respondents who stopped during the CE part. The final sample includes 283 respondents, which corresponds to a response rate of 19.8%. Importantly, the forest owners in the final sample own 12.5% of the total private forest area in Denmark. 4. Results 4.1. Main effects model Table 2 shows the results of a main effect RPL model along with the WTA estimates. Note that the compensation requirements are per

Table 1 Attributes investigated in the choice experiment. Attributes

Variable name

Coded levels

Levels of agreement on minimum management restrictions

Set aside an area of the forest Allowing 5 old trees to decay naturally Agree to an area with broadleaves of at least

0,7,15 0,5 3 dummies (0,1)

No change No change 0% broadleaves

7% of the forest 15% of the forest Allowing 5 old trees per hectare to decay naturally 25% broadleaves of total 50% broadleaves of 75% broadleaves of forest area total forest area total forest area

Increase the public's access

UNTOUCH DEAD BLEAVE25 BLEAVE50 BLEAVE75 ACC

0,1,2

No change

Lower property tax

PRICE

0,25,50,75,100,125,150,175 0 DKK

Access for the public on foot up to 15 m off roads and paths 25 50 75 DKK DKK DKK

Access for the public on foot everywhere in the forest 100 DKK

125 DKK

150 DKK

175 DKK

18

S.E. Vedel et al. / Ecological Economics 113 (2015) 15–24

Table 2 RPL model of main effects (based on 1000 Halton draws). WTA is based on the Delta method and converted using 1 € equals 7.45 DKK. Coefficient

Z

Random parameters in utility functions UNTOUCH −.05173** −2.10 DEAD .12837** 2.44 BLEAVE25 −.38207 −1.10 BLEAVE50 −.22934 −.62 BLEAVE75 −.75759** −2.15 ACC −1.73080*** −6.85 Non-random parameters in utility functions ASC 4.84554*** 7.25 PRICE .01426*** 7.56

WTA (€/ha/year)

WTA 95% confidence interval

0.48** −1.21** 3.60 2.16 7.13** 16.30***

(0.05–0.92) (−2.2–−0.22) (−2.64–9.84) (−4.64–8.96) (1.21–13.05) (11.28–21.3)

−45.61***

(−57.95–−33.28)

Standard deviations of normal distributed random parameters UNTOUCH .13039*** 4.70 DEAD .18802* 1.72 BLEAVE25 1.35748** 2.15 BLLEAVE50 1.01828** 1.97 BLEAVE75 .73181 0.98 ACC 1.45746*** 4.83 Standard deviations of error component σ*10 5.24778*** 7.01 283/0.5782 No. of respondents/Adj. pseudo R2 −786.8/0.5763 Log-likelihood/R2 adjusted 2 −1865.4/2157.2 Restricted LL/χ ***, **, * → Significance at the 1%, 5%, and 10% levels.

hectare for the entire forest, even if the restriction (such as untouched forest) only applies to a part of the forest. The marginal utility of income is assumed to be constant and equal for all respondents to facilitate the WTA calculation. Changing this assumption to, e.g., a log normal distribution does not affect the estimates of the other parameters. As shown in Table 2, forest owners require €0.48/ha/year (for the entire property) in compensation through tax relief for each percentage point of forest set aside. Thus, if 7 ha is set aside as untouched in a 100 ha forest, the compensation would be (0.48 × 7 × 100=) €338.3/ year in total for the owner. Applying a risk adjusted market interest rate of 2% for forest investments (Thorsen, 2010; Brukas et al., 2001), this corresponds to approximately €2416/ha of set aside forest land, which is somewhat below what would typically be the financial opportunity costs (Jacobsen et al., 2013). Interestingly, allowing trees to decay naturally is an aspect which forest owners are willing to accept negative compensation for, i.e. €-6/ha/year for setting aside 5 trees per hectare after the final harvest. Having a requirement on the percentage of broadleaves in the forest, however, has no effect until it reaches 75%, when forest owners require €7.1/ha/year for the entire property. For allowing the public access up to 15 m off roads and paths, forest owners require €16/ha/year for the entire forest, and €32/ha/year for allowing access everywhere on the forest floor, which makes this ecosystem service expensive to target. We note, however, that even though the access levels are not linear conceptually, they are modeled as such here as a linear tendency was found in dummy-type models. Through the ASC (alternative specific constant), we see that owners in general require €46 as a baseline payment to enter a contract (with only status quo levels off all attributes). For all the random parameters in Table 2, except compensation for 75% broadleaved cover and allowing 5 trees to decay naturally per ha, we see considerable heterogeneity in the population. 4.2. Accounting for current management and lack of additionality: a model with interactions The large heterogeneity observed for the random main effect parameters in Table 2 questions whether some of the unexpectedly low compensation requirements found, in particular for setting aside forests and

allowing mature trees to age and decay naturally, may be due to some forest owners deriving private benefits from these measures or — more importantly from an additionality and efficiency point of view — perhaps having already implemented these measures. To be able to correct for this, we needed information on the current management practices and this was elicited prior to the introduction of the choice experiment. Table 3 shows the presence of a number of relevant current management practice characteristics on the owners' properties. As we have specific hypotheses for the heterogeneity linked to current management practices, we estimated an RPL model of the results including interaction effects. The results are shown in Table 4. We see the same pattern for the main effects as in the previous main model, though the levels of WTA obviously change as the main effects now only cover the subsample of respondents who have not already implemented the management measures. We discuss all parameters in detail, relating them to the current management aspects. The current level of having 50% broadleaves was selected to maintain the number of interaction variables at a sensible level compared to the sample size. 4.2.1. Additionality in access Allowing access to either up to 15 m off roads and paths (level 1) or everywhere in the forest (level 2) is an expensive management change to get the forest owners to accept. The main effect estimate shows that a forest owner who at present does not provide the public extended access rights in his forest requires €14/ha/year to accept each of these levels. However, there is considerable variation with current management practice, and we have interacted with two characteristics: The current access level allowed in the forest, as stated by the forest owner, and the hunting interest on the property, again according to the owner. The current access level is a dummy variable which takes the value 1 if the owner has stated that he already grants the public access to the entire forest. Similarly, hunting interests is also a dummy variable, which takes the value 1 if the respondent or his/her family uses the forest for hunting or the hunting rights are let out to hunters. We find that the forest owners who already allow access to the entire forest, on average, require significantly less compensation for this service (approximately €18 less for each level), which results in an insignificant level of WTA. In contrast, the effect of owners' hunting interests on their property has a slight tendency to increase the costs of additional access, though it is not significant in this model. Furthermore, from the standard deviation of access, we see that while the effect of current management captures some of the heterogeneity across owners, there is still considerable preference heterogeneity among owners. We tried to model the access distribution more flexibly by testing the inclusion of higher order effects, but this did not significantly improve the model fit (see Fosgerau and Bierlaire, 2007 and Scarpa et al., 2008 for a description of this method). 4.2.2. Untouched forest areas We note from Table 3 that 60% of the respondents stated that they have a minimum of 5% untouched forest area on their property. This is somewhat more than the total designated and contracted untouched forest area in the country, which today constitutes 1.6% of the total (Ejrnæs, 2009), but it is closer to the average finding of unmanaged forest from the Danish National Forest Inventory (around 5% of forest appears to Table 3 Percentage of owners who currently provide/partly provide the investigated ecosystem services. Management for ecosystem services

Variable name

Percentage of owners

Already have min. 5% of untouched forest Already allow 5 trees per ha to decay naturally Already allow access everywhere Have more than 50% broadleaves Have active hunting interests on the property

HAVEUNTOUCH HAVEDEAD HAVEACC HAVE N 50% HUNT

60.1% 49.5% 9% 29.4% 73.5%

S.E. Vedel et al. / Ecological Economics 113 (2015) 15–24

19

Table 4 RPL model including interaction effects (based on 1000 Halton draws). WTA is based on the Delta method, and converted using 1 € equals 7.45 DKK. Coefficient

Z

WTA (€/ha/year)

95% confidence interval

Random parameters in utility functions UNTOUCH DEAD BLEAVE25 BLEAVE50 BLEAVE75 ACC

−.12794*** .00515 −.38272 −.42662 −.80802* −1.59201***

−3.62 .07 −1.03 −1.01 −1.91 −4.17

1.11*** −0.044 3.33 3.72 7.05** 13.88***

(0.5–1.73) (−1.3–1.21) (−2.82–9.49) (−3.36–10.8) (0.07–14.01) (7.18–20.57)

Non-random parameters in utility functions ASC PRICE HAVEUNTOUCH*UNTOUCH HAVEDEAD*DEAD HAVE N 50%*BLEAVE25 HAVE N 50%*BLEAVE50 HAVE N 50%*BLEAVE75 HAVEACC*ACC HUNT*ACC

5.05534*** .01540*** .12628*** .19996** .08352 .26644 −.04896 2.04179*** −.59265

7.66 7.67 3.13 2.04 .13 .32 −.08 3.22 −1.48

−44.05***

(−56.16–−31.95)

−1.10*** −1.74** −0.72 −2.32 0.43 −17.80*** 5.17

(−1.82–−0.38) (−3.41–−0.07) (−11.59–10.14) (−16.69–12.04) (−10.42–11.27) (−28.55–−7.04) (−1.59–11.92)

Standard deviation of normal distributed random parameters UNTOUCH .11519*** DEAD .28694*** BLEAVE25 .70594 BLEAVE50 1.21319** BLEAVE75 .90320 ACC 1.30613*** Standard deviations of error component σ*10 No. of respondents/Adj. pseudo R2 Log-likelihood/R2 adjusted Restricted LL/χ2

5.29179 283/0.5868 −771/0.5841 −1865/2189

3.77 2.97 .88 1.96 1.56 4.39

7.25

***, **, * → Significance at the 1%, 5%, and 10% levels.

be unmanaged). In the interactions model in Table 4, we find the main to be €1.1/ha/year, which corresponds to a present value of €5570/ha at a 2% discount rate. For the group who already claim to do this, we see a large discount of €1.1/ha/year, which effectively eliminates the mean WTA effect.

4.2.3. Allowing trees to decay naturally We see that the approximately 50% of forest owners who stated that they already allow trees to decay naturally in their forest (see Table 3) had a negative mean compensation requirement of € 1.7/ha/year for allowing a tree to decay naturally. This is a good explanation as to why the main effect parameter DEAD was unexpectedly negative in the main effect model, and though it was negative in the interaction model, it became insignificant. Consequently, it indicates that even forest owners who currently do not undertake this action may not require much compensation.

4.2.4. Minimum percentage of broadleaves Forest owners whose forests already consist of a minimum of 50% broadleaves were interacted with the three attribute levels (25%, 50% and 75%) of the requirement on broadleaves. None of these were significant which shows that forest owners who to a large extent already fulfill this requirement do not reduce the compensation they demand for this requirement. However, based on the main effects of both models, accepting a requirement on broadleaves of 25% or 50% was not significant. However, above a certain threshold, namely a requirement for a minimum of 75% broadleaves, they do require approximately €7/ha/year. We also note that there is significant heterogeneity associated with the main effect requirement of 50%, which indicates that although the average owner does not require compensation for this, a proportion of the sample does.

4.3. Policy simulation As we asked forest owners to imagine entering the proposed contract with their largest coherent forest area in mind, we are able to link their preferences to a specific forest size. Based on the model with heterogeneity, we estimate the forest owners' individual WTA for each attribute of the contract and link this to their individual forest area. This allows us to construct a ‘supply curve’ for the management change. Figs. 1 and 2 show the relation between compensation levels and the expected potential cumulative forest area under contract for two of the ecosystem services investigated here, namely ‘untouched forest areas’ for biodiversity protection and ‘enhanced access.’ Fig. 1 shows that a compensation level of approximately €188 would be required if all owners in our sample were to enter a contract to set aside 5% of their forest area. Note that owners who have only chosen the status quo are allocated to the high end of the WTA range by the estimation method, but this may be a conservative estimate of the true WTA of this group. Fig. 2 shows that compared to setting aside forest, it would be considerably cheaper to get all forest owners in the sample to accept a contract granting the public access on foot to their entire forest (approximately €56/ha/year2). In comparison, owners of approximately 40% of the forest area in our sample require close to zero compensation for granting increased access, whereas forest owners who could be expected to enter an ecosystem service contract for 5% untouched forest at the same low price, own a much smaller share of the forest area in the sample. If the aim is to attract the owners of 50% of the forest area, then compensation of approximately €19/ha/year is required for increased access, and €115/ha/year for each hectare of untouched forest. 2 Note that although this graph also starts out in the negative scale with regard to WTA, the overall costs should be seen in connection with the ASC which determines the baseline compensation for entering the proposed contract for ecosystem service provision.

20

S.E. Vedel et al. / Ecological Economics 113 (2015) 15–24

250 200 150 100 50 WTA 0 0

500

1000

1500

2000

2500

3000

-50 -100 -150 -200

Area in hectares

Fig. 1. An estimation of the link between individual forest owners' WTA, their compensation requirements for setting aside 5% of their forest and the increasing marginal compensation as the cumulative area targeted increase within the sample.

The above results are calculated for our sample of forest owners, representing some 12.5% of the privately owned forest area in Denmark. Appendix B shows a similar estimation of ‘upscaled supply curves’ and increasing marginal costs of provision for untouched forest and increased access. The figures were derived by upscaling the sample investigated here to all private forest owners in Denmark (See Appendix B, Table B3). To account for the different distribution of forest property sizes in our sample compared to the population, forest owners were divided into 10 size classes corresponding to official statistics (Larsen and Johannsen, 2002). The sampled population in each size class was then assumed to represent all private owners in the size class. A graph of the cumulative forest area covered by a contract was made by assigning identical portions of the entire forest area in each size class to the individual WTA estimates of our sample in that size class. When looking at the simulation for the entire private forest area, we find that €29/ha/year would be required to persuade owners to grant complete access to 50% of the private forest area. To persuade forest owners of half of the forest area to enter into a contract to set aside 5% of their area would

require approximately €77/ha/year. Besides this, as shown in Table B3, there is no clear relation between forest owners of different size categories and the average individual WTA among owners in these subgroups. 5. Discussion The ownership of Denmark's forests is fragmented with a large number of people each of whom typically owns a fairly small area. This implies that policy initiatives regarding enhanced biodiversity measures or recreational opportunities for the public often have to involve a large number of forest owners to be successful. Persuading forest owners to implement specific initiatives on their property through participation in voluntary schemes is a relevant policy instrument for achieving many of the national and international goals for biodiversity protection and conservation and for providing the public with recreational opportunities. The Danish forests serve as key habitats for more than 50% of the red-listed species in Denmark (Stoltze and Pihl, 1998) and are hence

70 60 50 40 30 20 WTA 10 0 0

10000

20000

30000

40000

50000

60000

-10 -20 -30 -40

Area in hectares

Fig. 2. An estimation of the link between individual forest owners' WTA for public access on foot everywhere in their forest and the increasing marginal compensation as the cumulative area targeted increase within the sample.

S.E. Vedel et al. / Ecological Economics 113 (2015) 15–24

of great importance for halting biodiversity decline. Furthermore, in the Danish landscape, which is characterized by intensive agricultural production, the forests play an important role in providing areas for recreation and leisure time activities for the public. With 75 million annual forest visits, they are important for the public to experience peaceful and quiet surroundings, experience nature and have a break from everyday life (Jensen and Koch, 2004). In a historical perspective, the public's interest in pursuing recreational opportunities and leisure time activities in forests has given rise to conflicts with landowners and their wish to maintain private property rights. To a great extent, the recreational options have been provided through public provision based on state ownership of forests — especially close to urban areas. However, with approximately 70% of the forest area in Denmark being privately owned, the recreational use of private forest areas remains an important issue for a large share of the Danish population. 5.1. The provision of enhanced access Our results show that forest owners on average are reluctant to grant increased access. Owners require on average €33/ha/year for granting the public full access on foot to their forests. Many of the viewpoints on this topic expressed in the open-ended statements in the questionnaire also reflect a strong aversion to increased public access. Some forest owners with relatively small areas stated that they did not want people to trespass close to their homes and expressed concerns about maintaining privacy. The issue of enhanced public access rights in private forests has been part of environmental policy debates for several decades, and our results illustrate that forest owners remain largely opposed to such initiatives and to a significant degree compared to other potential initiatives. We also know that the compensation requirement is higher where demand for recreational access is higher (Broch et al., 2013). Interestingly, 9% of our sample provides enhanced access voluntarily. Recent analyses of the willingness-to-pay for such enhanced access revealed that the general public is in fact divided on the value of enhanced recreational access (Campbell et al., 2014). However, recreational ecosystem services are generally highly valued (Bateman et al., 2013), also in Denmark (DØRS, 2014), and the value of improving the supply of recreational ecosystem services could be large in some areas. For many forest owners, hunting is important both financially and due to personal interest, and hunting experiences are often sensitive to disturbances from other forest guests. In the current model, the sign of the effect is as expected, but it is not significant. The greatest effect of reducing compensation for access was seen among the approximately 9% of forest owners who stated that they already allowed extended access; for this group the mean compensation dropped to zero. 5.2. The enhanced provision of biodiversity protection Turning to attributes related to enhancing biodiversity, we find that, on average, the forest owners require €0.48/ha/year for the entire forest per percentage of land set aside, which corresponds to a present value of €2416/ha set aside at a 2% discount rate. Setting aside an area is a major management change which may entail large opportunity costs for some owners, depending on the characteristics of the forest area in question, so the required compensation appears low for this change compared to less restrictive measures (Jacobsen et al., 2013). It should, however, be noted that there is considerable heterogeneity in these WTA measures within the sample of forest owners, and when accounting for the part of heterogeneity related to forest owners who already have set aside parts of their land, the results change. In fact, using an interaction dummy for owners who claim to already do this, we find that this group, in total, does not require compensation. For the remaining owners, however, the WTA increases to €1.1/ha/year, which corresponds to a present value of €5570/ha at a 2% discount rate. This may still be lower than the financial value of old, productive stands. Our estimates here can be compared with a recent cost-effectiveness analysis carried out by

21

the Danish Economic Councils (DØRS, 2012), who found that the most effective measure to protect 200 endangered species was to set aside 47,000 ha of forest in Denmark. Based on simple capital budgeting analysis of the forest production, they found the opportunity costs to be in the range of €270–400/ha/year when setting aside older broadleaved forest areas. Our results here indicate considerably lower opportunity costs, in particular for the first 1000 ha. This is probably because forest owners would tend to select their least productive areas, i.e. forests on wetlands for set aside. Of course, investigating whether the areas forest owners are willing to set-aside (or perhaps have already set aside) at a given compensation level are also areas of interest for biodiversity protection purposes remains a key issue. If this is the case, then taking into account the effect of current management practices documented in this study can significantly reduce the potential inefficiency caused by a lack of additionality in contracting and provision. If the opposite is the case, further policy improvements would be needed to efficiently enhance biodiversity protection. Allowing 5 trees/ha to age and decay naturally after the final harvest is considered a small-scale management change, and the main effects model which ignored a lack of additionality showed that forest owners on average required a negative compensation for such a policy attribute. However, turning to the model with interaction effects, we found that this effect can be attributed entirely to respondents who already do this, resulting in an insignificant parameter estimate for this attribute for other forest owners. Nevertheless, the result indicated that allowing 5 trees/ha to decay naturally is something forest owners, on average, would be willing to do without receiving compensation. This finding is in contrast to current schemes that offer payment for setting such trees aside in connection with final harvest or regeneration (Jacobsen et al., 2013). 5.3. Broadleaf requirements On several forest lands, conifer stands have higher volume productivity and can be established at lower costs. Therefore, species such as Norway Spruce (Picea abies L.) and Sitka Spruce (Picea sitchensis Bong. (Carr.)) are quite widely used in monocultures. Thus, the percentage of broadleaves varies significantly across the country. During the 1990s and 2000s, various policies encouraged an increase in the area of broadleaves for several reasons, including concerns for forest stability (in particular risks of wind throw), greater biodiversity, but also concerns regarding groundwater quality and quantity. Groundwater recharge under broadleaved forests is considerably higher and of better quality (lower nutrient loading) than under coniferous forests (RaulundRasmussen and Hansen, 2003). Once again, the public has a significant willingness-to-pay for such enhanced groundwater provision (Campbell et al., 2014). Thus, we expected that, on average, forest owners would demand compensation for requirements regarding species choice (broadleaf requirements), and that it would vary depending on current levels of broadleaves. In the model which ignores the current status, we found that owners were on average willing to accept a broadleaf requirement of up to 50% without significant compensation, but demanded €7/ha/ year for the entire forest for accepting a requirement for a minimum of 75% broadleaves on their forest land. It should be noted, though, that there was significant variation among forest owners for the 50% requirement level, whereas owners seemed more united in demanding compensation for a 75% broadleaf requirement. Turning to the model which accounts for current forest management, we found, interestingly, that the existing percentage of broadleaves on the property did not influence the compensation requirement significantly. Overall, the compensation required here appears to be low when accounting for the general differences in the economics of coniferous vs broadleaved forests on much of the Danish forest soils. However, coniferous species may face greater challenges from climate change than the broadleaved alternatives (Hanewinkel et al., 2013), which may increase

22

S.E. Vedel et al. / Ecological Economics 113 (2015) 15–24

forest owners' expected benefits from such changes. Furthermore, the percentage of broadleaves (in particular older trees) on a forest property positively affects hunting land rents (Meilby et al., 2006; Lundhede et al., forthcoming). This could also be factored in by forest owners with diverse objectives and motivations for ownership and management. 5.4. Policy contributions Having obtained the results shown in Figs. 1 and 2, it is natural to ask how this may improve policy design or at least policy evaluation. In the EU, payments for actions such as the NATURA 2000 forest schemes have to be based on direct cost and income foregone measures (Jacobsen et al., 2013) to avoid ‘overcompensating’ forest owners. Modest menus of contracts and flat rate schemes are designed to keep costs — and potential overcompensation — at a low level. However, our results suggest that some forest owners may be willing to enter into at least some such agreements at even lower costs. This, and the large variation in WTA, also means that at a given flat-rate subsidy, there could be a considerable rent harvested by many forest owners. In particular, little additionality is gained from contracting with many of these properties, the exception being that with a contract, the public protects the ecosystem service provision from, e.g., a change in ownership. Such problems of additionality have been documented elsewhere, e.g., Juutinen et al. (2013) found that only 66% of forest owners would have harvested their forest during a 10 year period in the absence of the specific conservations scheme for setting aside forest land for that period. Thus, surveys like this may add information that can allow policy makers to better target the relevant segments of forest owners. For example, our results indicate that rather than targeting expensive enhanced access rights everywhere on private lands, it may be more beneficial for the government and interested NGOs to identify forest owners who already allow this, and negotiate more favorable terms with them. Finally, two additional issues deserve a comment in relation to the use of survey instruments, targeted auctions and for improving the cost-effectiveness of policies. The first is that targeting some owners implies the exclusion of others, which in turn implies the risk of crowding out some owners' private incentives for provision of ecosystem services among those who are not in focus. The second is an issue originally raised by Carson and Groves (2007) who point out that to be able to elicit agents' true preferences — whether in a real auction or in a survey — the instruments used must be incentive compatible. This usually implies a design that involves avoiding the use of the information obtained about the specific agent to extort all possible rents. In an auction, this can be implemented by applying a first-rejected-offer-setsthe-price scheme. In a survey like this, incentive compatibility is ensured by the policy context that forest owners are well aware of, namely that it is quite likely that the survey will affect actual policy design. 5.5. Caveats and limitations The main caveat of our study is that it relies on the voluntary responses of a random sample of forest owners and, therefore, may suffer from significant selection bias relative to a fully representative sample. For example, the number of owners that provide untouched forest areas as well as the compensation levels for setting aside individual trees may indicate some bias towards respondents with some interest in providing ecosystem services, especially related to biodiversity. However, although our final sample consisted of only 283 forest owners, they own in total 56,316 ha. This accounts for approximately 12.5% of the Danish private forest land. Thus, despite the issue of self-selection bias, the forest properties investigated here constitute a non-trivial proportion of the privately owned Danish forest area, and hence policy relevance is high in spite of the small and potentially not completely representative sample, something which is illustrated in the above policy oriented simulations.

PES programs typically involve problems of asymmetric information. Therefore, landowners have an incentive to act strategically by asking bids above their true opportunity costs (e.g., Latacz-Lohman and van der Haamsvoort, 1997). This study relies on stated preference techniques so we cannot rule out that some owners may have had an incentive to answer strategically, in spite of observing aspects of incentive compatibility in the design (Carson and Groves, 2007). However, the overall compensation levels are not high compared to the estimated costs of these and related initiatives, e.g., for leaving single trees or undertaking environmentally friendly reforestation measures (see, e.g., Jacobsen et al., 2013), which suggests that strategic answering is not a significant problem. 6. Conclusion Enhancing the provision of biodiversity and recreational opportunities in forests relies to a great extent on private forest owners' own initiatives, changes in rules and regulations or, as increasingly stressed in the environmental policy arena over the last decades, owners' willingness to voluntarily accept PES-type contracts specifying conditional payments for the provision of specific ecosystem services or management changes. The performance of this latter type of instrument hinges on several crucial assumptions (Wunder et al., 2008). An important one of these is the additionality of the management measures paid for. If payments are made to forest owners for undertaking actions or meeting requirements which they already do, then payments are essentially of no or little consequence. Thus, it is important for policy design and evaluation to account for forest owners' current forest management practices when investigating potential WTA levels. This study has shown that forest owners do have heterogeneous preferences for providing different types of ecosystem services related to the Natura 2000 framework. The required management changes include setting aside parts of the forest, setting aside individual old trees to age and decay naturally, minimum requirements on broadleaved species and an increase in recreational access. Moreover, for the majority of these management actions, forest owners reduce their required WTA significantly if they already undertake these actions — in most cases the WTA drops to close to zero. Despite potential self-selection bias, the study is based on a sample of forest owners who together own 12.5% of the Danish private forest area. This highlights the scope for targeting the provision of ecosystem services and improving cost efficiency when combined with analyses of the value created on the benefits side. Our results demonstrate that several owners carry out forest management practices that, at least in terms of the written descriptions, seem to be in line with the forest management practices required in, e.g., the NATURA 2000 and the Convention on Biological Diversity. For these owners, flat rate payments for such actions would imply nearly no additionality and would be highly inefficient. Thus, our results suggest that other types of policies should be considered, e.g., targeting subsets of forest owners with rewards or branding not strictly in the form of compensation or other actions for this particular group. We note, however, that if such actions are coupled with either the exclusion of or higher compensation payments for other types of forest owners to undertake similar actions, there is a real risk of crowding out private incentives and inducing a potential negative change in practice for the ‘biodiversity oriented’ forest owners. This concern needs to be balanced with the concern for efficiency loss due to a lack of additionality when designing the economic instruments. Acknowledgments The authors appreciate the helpful suggestions from two anonymous reviewers. The research is funded by the EU FP7 project NEWFOREX (No:243950). Furthermore, Jette B. Jacobsen and Bo J. Thorsen acknowledge support from the Danish National Science Foundation for the Center for Macroecology, Evolution and Climate.

S.E. Vedel et al. / Ecological Economics 113 (2015) 15–24

23

Appendix A Fig. A.1 shows an example of a choice set as it was presented to the respondents on screen where icons are used in the left hand column together with text to describe the attributes. Changes which are referred to in exact figures or percentages are in bold.

Fig. A.1. Example of a choice set as it was displayed to respondents. Choice between subsidy schemes (1 out of 6). If the subsidy scheme, for example, includes a lower property tax of 75 DKK per hectare, and your largest forest is 100 ha, then your compensation will amount to 7500 DKK every year (in cash after tax) for the whole forest. Which of these subsidy schemes would you prefer for your (whole) largest forest?

Appendix B

Fig. B1. Up-scaling results to the national private forest area, and simulating a ‘supply curve’ showing the increasing required compensation for setting aside 5% of a forest the greater the forest area which is targeted.

Fig. B2. Up-scaling results to the national private forest area, and simulating a ‘supply curve’ showing the increasing required compensation for granting the public access on foot the greater the forest area which is targeted.

24

S.E. Vedel et al. / Ecological Economics 113 (2015) 15–24

Table B3 The average WTA of forest owners in each size category. No. of hectares

No. of respondents in sample

Total no. of forest owners in Denmark

Average WTA access (€)/ha and year

No. who already provide access

Average WTA untouched (€)/ha and year

No. who already provide untouched forest

b2 2–5 5–10 10–20 20–50 50–100 100–250 250–500 500–1000 N1000

36 58 41 52 34 11 16 13 10 12

8437 8319 4640 2512 1375 358 275 107 67 27

21.94 23.84 26.57 22.13 26.56 32.23 26.61 18.82 27.73 15.61

5 4 2 7 2 0 1 2 0 3

64.23 73.56 51.97 64.38 55.12 72.50 83.86 72.68 47.65 83.11

15 24 18 19 17 4 5 6 6 5

References Bateman, I.J., Harwood, A.R., Mace, G.M., Watson, R.T., Abson, D.J., Andrews, B., Binner, A., Crowe, A., Day, B.H., Dugdale, S., Fezzi, C., Foden, J., Hadley, D., Haines-Young, R., Hulme, M., Kontoleon, A., Lovett, A.A., Munday, P., Pascual, U., Paterson, J., Perino, G., Sen, A., Siriwardena, G., van Soest, D., Termansen, M., 2013. Bringing ecosystem services into economic decision-making: land use in the United Kingdom. Science 341 (6141), 45–50. Boon, T.E., Meilby, H., Thorsen, B.J., 2004. An empirically based typology of forest owners in Denmark — improving the communication between authorities and owners. Scand. J. For. Res. 19 (4), 45–55. Broch, S.W., Vedel, S.E., 2012. Using choice experiments to investigate the policy relevance of heterogeneity in farmer agri-environmental contract preferences. Environ. Resour. Econ. 51, 561–581. Broch, S.W., Strange, N., Jacobsen, J.B., Wilson, K.A., 2013. Farmers' willingness to provide ecosystem services and effects of their spatial distribution. Ecol. Econ. 92, 78–86. Brukas, V., Thorsen, B.J., Helles, F., Tarp, P., 2001. Discount rate and harvest policy: implications for Baltic forestry. For. Policy Econ. 2, 143–156. Campbell, D., Vedel, S.E., Thorsen, B.J., Jacobsen, J.B., 2014. Heterogeneity in the demand for recreational access — distributional aspects. J. Environ. Plan. Manag. 57, 1200–1219. Carson, R.T., Groves, T., 2007. Incentive and informational properties of preference questions. Environ. Resour. Econ. 37, 181–210. DØRS, 2012. Biodiversitet (Biodiversity). Chapter 2, Økonomi og Miljø 2012. (Economics and Environment 2012). The Danish Economic Councils, Copenhagen Denmark, pp. 141–287 (http://www.dors.dk/sw9038.asp). DØRS, 2014. Værdi af rekreative områder (Value of recreational areas). Chapter 4. Økonomi og Miljø 2014. (Economics and Environment 2014). The Danish Economic Councils, Copenhagen, Denmark. Ejrnæs, R., 2009. Notat til By og Landskabsstyrelsen med udkast til kapitel 1 til 4. Landerapporten til CBD-sekretariatet om tilstand, udvikling og trusler for Danmarks biodiversitet: overview of biodiversity status, trends and threats in Denmark. Engel, S., Pagiola, S., Wunder, S., 2008. Designing payments for environmental services in theory and practice: an overview of the issues. Ecol. Econ. 65, 663–674. Espinosa-Goded, M., Barreiro-Hurlé, J., Ruto, E., 2010. What do farmers want from agrienvironmental scheme design? A choice experiment approach. J. Agric. Econ. 61 (2), 259–273. Ferrini, S., Scarpa, R., 2007. Designs with a priori information for nonmarket valuation with choice experiments: a Monte Carlo study. J. Environ. Econ. Manag. 53, 342–363. Fosgerau, M., Bierlaire, M., 2007. A practical test for the choice of mixing distributions in discrete choice models. Transp. Res. B 41, 784–794. Greene, W.H., 2002. Econometric Analysis. Fifth edition. Prentice Hall, New Jersey (1026 pp.). Greene, W.H., Hensher, D.A., 2007. Heteroscedastic control for random coefficients and error components in mixed logit. Transp. Res. E 43, 610–623. Hackl, F., Halla, M., Pruckner, G.J., 2007. Local compensation payments for agrienvironmental externalities: a panel data analysis of bargaining outcomes. Eur. Rev. Agric. Econ. 34 (3), 295–320. Hanewinkel, M., Cullmann, D.A., Schelhaas, M.-J., Nabuurs, G.-J., Zimmermann, N.E., 2013. Climate change may cause severe loss in the economic value of European forest land. Nat. Clim. Chang. 3, 203–207. Horne, P., 2006. Forest owners' acceptance of incentive based policy instruments in forest biodiversity conservation—a choice experiment based approach. Silva Fenn. 40 (1), 169–178. Hudson, D., Lusk, J., 2004. Risk and transaction cost in contracting: results from a choicebased experiment. J. Agric. Food Ind. Organ. 2 (1), 1–17. Jacobsen, J.B., Vedel, S., Thorsen, B.J., 2013. Assessing costs of multifunctional NATURA2000 management restrictions in continuous cover beech forest management. Forestry 86, 572–582. Jensen, F.S., Koch, N.E., 2004. Twenty-five years of forest recreation research in Denmark and its influence on forest policy. Scand. J. For. Res. 19 (4), 93–102.

Juutinen, A., Ollikainen, M., 2010. Conservation contracts for forest biodiversity. Theory and experience from Finland. For. Sci. 56, 201–211. Juutinen, A., Mäntymaa, E., Mönkkönen, M., Svento, R., 2008. Voluntary agreements in protecting privately owned forests in Finland — to buy or to lease? For. Policy Econ. 10, 230–239. Juutinen, A., Mäntymaa, E., Ollikainen, M., 2013. Landowners' conservation motives and the size of information rents in environmental bidding systems. J. For. Econ. 19, 128–148. Lancaster, K.J., 1966. New approach to consumer theory. J. Polit. Econ. 74, 132–157. Larsen, P.H., Johannsen, V.K., 2002. Skove og plantager 2000 (Forests and plantations 2000). Statistics Denmark, Forest and Landscape, Danish Nature Agency 87-5011287-2 (171 pp.). Latacz-Lohman, U., van der Haamsvoort, C., 1997. Auctioning conservation contracts: a theoretical analysis and an application. Am. J. Agric. Econ. 79, 407–418. Lundhede, T.H., Jacobsen, J.B., Thorsen, B.J., 2015. A hedonic analysis of the complex hunting experience. J. For. Econ. http://dx.doi.org/10.1016/j.jfe.2015.01.001 (forthcoming). Mäntymaa, E., Juutinen, A., Mönkkönen, M., Svento, R., 2009. Participation and compensation claims in voluntary forest conservation: A case of privately owned forests in Finland. For. Policy Econ. 11, 498–507. McFadden, D., 1973. Conditional logit analysis of qualitative choice behaviour. In: Zarembka, P. (Ed.), Frontiers in Econometrics. Academic Press Inc., New York, pp. 105–142. Meilby, H., Strange, N., Thorsen, B.J., Helles, F., 2006. A hedonic analysis of the price of hunting rentals. Scand. J. For. Res. 21, 63–72. Rabotyagov, S.S., Lin, S., 2013. Small forest landowner preferences for working forest conservation contract attributes: A case of Washington State, USA. J. For. Econ. 19, 307–330. Raulund-Rasmussen, K., Hansen, K. (Eds.), 2003. Grundvand fra skove — muligheder og problemer. (Ground Water From Forests — Opportunities and Issues).. Skovbrugsserien 34. Skov & Landskab, Hørsholm, p. 122. Ruto, E., Garrod, G., 2009. Investigating farmers' preferences for the design of agrienvironmental schemes: a choice experiment approach. J. Environ. Plan. Manag. 52 (5), 631–647. Scarpa, R., Thiene, M., Marangon, F., 2008. Using flexible taste distributions to value collective reputation for environmentally friendly production methods. Can. J. Agric. Econ. 56, 145.162. Söderberg, U., Johannsen, V.K., 2000. Designing a sample-based National Forest Inventory for Denmark. Field manual for the sample based forest inventory. FSL, Report No.3. Stoltze, M., Pihl, S. (Eds.), 1998. Rødliste 1997 over planter og dyr i Danmark. (Red list of plants and animals in Denamrk1997). Miljø- og Energiministeriet, Danmarks Miljøundersøgelser og Skov- og Naturstyrelsen. Thorsen, B.J., 2010. Risk, returns and possible speculative bubbles in the price of Danish forest land? In: Helles, F., Nielsen, P.S. (Eds.), Proceedings of the Biennial Meeting of the Scandinavian Society of Forest Economics, 19–22. May, 2010, Gilleleje, Denmark, pp. 100–111 Train, K., 2009. Discrete Choice Methods with Simulation. 2nd edn. Cambridge University Press, Cambridge. Vanslembrouck, I., Van Huylenbroeck, G., Verbeke, W., 2002. Determinants of willingness of Belgian farmers to participate in agri-environmental measures. J. Agric. Econ. 53 (3), 489–511. Vedel, S.E., Jacobsen, J.B., Thorsen, B.J., 2015. Contracts for afforestation and the role of monitoring for landowners' willingness to accept. For. Policy Econ. 51, 29–37. Vukina, T., Zheng, X., Marra, M., Levy, A., 2008. Do farmers value the environment? Evidence from a conservation reserve program auction. Int. J. Ind. Organ. 26, 1323–1332. Wilson, G.A., Hart, K., 2000. Financial imperative or conservation concern? EU farmers' motivation for participation in voluntary agri-environmental schemes. Environ. Plan. A 32, 2161–2185. Wunder, S., Engel, S., Pagiola, S., 2008. Taking stock: a comparative analysis of payment for environmental services in developed and developing countries. Ecol. Econ. 65, 834–852.