Stated preference data on the insurance value of forests in Switzerland

We present stated preference data for improved forest management from seven Swiss municipalities in the Cantons of Grisons and Valais. The data was collected between October 2019 and February 2020 using an online questionnaire. We invited 10289 households to participate and received 939 responses. The online questionnaire consisted of two main parts – (1) an online choice experiment and (2) questions on the sociodemographic characteristics of the responding households. The choice experiment confronted households with twelve consecutive choice tasks. Each choice task consisted of three options with a varying degree of avalanche and rock fall risk reduction due to improved forest management. The options further differed with respect to the way the charges for the improved forest management are determined and assigned to the households. We additionally included a cost attribute to estimate the respondents' willingness to pay. At the end of the choice experiment we asked five de-briefing questions and eight attitudinal questions. Additionally, we asked the responding households to state their willingness to take risks. The sociodemographic characteristics collected in the second part of the questionnaire allowed us to analyse their impact on the choices we observed in the first part of the questionnaire. An analysis of the choice data and further interpretive insights are presented in the article “Determining the insurance value of ecosystems: A discrete choice study on natural hazard protection by forests”.

© 2020 The Author(s The questionnaire targeted households in seven Swiss municipalities in the Canton of Grisons and Valais, which are exposed to avalanche and rock fall risk. Description of data collection Around 10,300 invitation letters were sent to households in seven Swiss municipalities. Each letter contained login details to an online questionnaire, which we administered via Sawtooth. The questionnaire consisted of two parts, a choice experiment and questions on sociodemographic characteristics of the responding households. Data  Value of the Data -The data allow for eliciting preferences of households for funding improved forest management that is geared towards reducing avalanche and rock fall risk. Additionally, it helps to see how variations in sociodemographic characteristics affect these preferences. Eventually, the data provide a demand side perspective on the insurance value of ecosystems. -The data helps to inform policy makers as well as the private sector (insurance companies) when it comes to incorporating ecosystem services into climate change adaptation and disaster risk management agendas. -The data can be used to operationalize the insurance value of ecosystems. In particular, this dataset enables a demand side perspective on the insurance value of ecosystems and helps to analyze preferences and the willingness to pay based on sociodemographic characteristic. -Preferences and willingness to pay estimates on the demand side can be linked with insights from the supply side (the provisioning of ecosystem services) in order to identify feasible strategies and policies.

Data Description
We conducted a discrete choice study to elicit the preferences of Swiss households for forest management that is geared towards the reduction of natural hazards. We focused on seven municipalities in the Cantons of Grisons and Valais. In each of these municipalities, households face avalanche and rock fall risks. The first part of our online questionnaire consisted of a discrete choice study in which respondents face 12 consecutive choice tasks. In each choice task, respondents are asked to choose one among three alternatives. Each of these alternatives had five attributes with two to six attribute levels. The attributes and their respective levels are presented in Table 1 in Section 2.1 . After the choice experiment, we asked five de-briefing questions as shown in Table 2 in Section 2.1 , followed by eight attitudinal questions as shown in Table 3 .
In the second part of the survey, we asked respondents about their sociodemographic characteristics, e.g., age, gender, monthly disposable income per household, type of residence, property situation etc. The questions on the sociodemographic characteristics are shown in Table 4 in Section 2.2 below. Following [ 1 , 2 ], we additionally asked respondents to state their willingness to take risks on a scale ranging from 0 to 10, with 0 referring to "completely unwilling to take risks" and 10 to "very willing to take risks". The here presented data is thus a combination of the choices respondents made within the choice experiment and their stated sociodemographic characteristics. For reasons of confidentiality we anonymized the data by removing all fields that would enable personal identification (IP-address, email address, comments, password, username). The complete questionnaire, the dataset and data description are available on the Environmental Data Platform EnviDat of the Swiss Federal Institute for Forest, Snow and Landscape Research WSL (doi:10.16904/envidat.175). "DIVES_Questionnaire.pdf" shows the complete questionnaire. "DIVES_choice_data.csv" is the dataset and "data_description.pdf" provides its detailed description.

Experimental Design, Materials and Methods
In October 2019 we sent invitation letters to all households in our seven case study municipalities in the Cantons of Grisons, GR, and Valais, VS (VS: St. Niklaus, Grächen, Täsch, Zermatt, GR: Davos, Albula/Alvra/Bergün Filisur). For the households in GR and the municipality of Täsch, Table 1 Attributes and their respective attribute levels (readapted from [3]    The questionnaire consisted of two main parts: (1) A choice experiment with twelve consecutive choice tasks and the associated de-briefing and attitudinal questions. (2) Questions on the sociodemographic characteristics of the household.

The choice experiment
The choice experiment consisted of twelve consecutive choice tasks. In each choice task respondents had to choose between three alternatives. Options 1 and 2 described situations in which better forest management leads to reduced avalanche and rock fall risk. The status quo in each choice task described a situation without improved forest management, hence there is no change in the avalanche and rock fall risk faced. To identify relevant attributes and credible attribute levels we collaborated with insurance and re-insurance professionals and academics from different disciplines. In the course of workshops and project meetings we identified five relevant attributes and the respective levels. These are shown in Table 1 .
We used Ngene to generate a D-efficient design that varies the attribute levels in Options 1 and 2 and has a constant status quo across twelve choice tasks [4] . For a detailed description of the attribute levels and the choice experiment please refer to the accompanying publication [3] . Table 4 Questions on the sociodemographic characteristics of the responding households.

Sociodemographic characteristics
Type of question Answer options Gender Select -Male -Female -Other Year of birth Input field Numeric YYYY Monthly household income Select -Less than CHF 20 0 0 -CHF 20 0 0 to below CHF 40 0 0 -CHF 40 0 0 to below CHF 60 0 0 -CHF 60 0 0 to below CHF 80 0 0 -CHF 80 0 0 to below CHF 10 0 0 0 -CHF 10 0 0 0 to below CHF 120 0 0 -CHF 120 0 0 to below CHF 140 0 0 -CHF 140 0 0 to below CHF 160 0 0 -CHF 160 0 0 to below CHF 180 0 0 -More than CHF 180 0 0 Subsequent to the choice experiment we asked the responding household five de-briefing questions to understand whether the households understood the experiment and considered it realistic. For each question the possible response ranged from "strongly disagree" to "strongly agree" with the opportunity not to state any answer. Table 2 below shows the debriefing questions and the possible answers.
Next, we asked eight attitudinal questions to get an idea of the subjective risk exposure of the responding households as well as subjective valuations and opinions on natural hazards, forest management and their link to climate change. They are shown in Table 3 .
Finally, we asked the responding household to indicate their willingness to take risks. Following [ 1 , 2 ], we used a scale ranging from 0 to 10, with 0 referring to "absolutely no willingness to take risks" and 10 to "very high willingness to take risks".

Sociodemographic characteristics
In the second part of the questionnaire we asked the responding households to share some sociodemographic characteristics. The questions and the respective answer options are presented in Table 4 .

Ethics Statement
The participation in the survey was voluntary and respondents were informed that the data will be analyzed anonymously. In participating in the survey and submitting the questionnaire each respondent gave her/his informed consent.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships which have, or could be perceived to have, influenced the work reported in this article.