Survey Abstract

• Online survey conducted in the first quarter of 2025 in Germany with a core sample of around 15,000 respondents
• Eliciting values for changes in protected natural areas and high‑nature‑value farmland, accounting for respondents’ current endowment of natural capital.
• Methods: Discrete choice experiment, life‑satisfaction valuation, and travel‑cost analysis
• The full dataset will be made publicly available as open-source

Life Satisfaction

Agricultural Intensification

• Basic background information on the impacts of intensified agriculture on biodiversity
• Illustrating examples of intensification measures and their effects
• Explaining the purpose of landscape changes to help restore biodiversity (next slides)

Characteristics HNV

Characteristics PA

HNV

PA

Choice Experiment Attribute Levels

*Note: For the size of protected areas and high nature value farmland, half of the respondents received the attribute levels listed for this vector, while for the other half the levels are twice as high

Choice Set Example

Choice Set Map

Choice Set Attributes

Data Downloads

RStudio Setup

• Open RStudio.
• Create a new folder on your computer (e.g. NV_Data).
• In RStudio, go to File > New Project > New Directory > New Project
• Name the project (e.g. NV_Data) and save it in the folder you created
  • Benefit: The .Rproj keeps all files (data, scripts, outputs) in one place and automatically sets the working directory to the project folder.
• Optional: Create subfolders for a better organization
  • data/ (for raw datasets)
  • scripts/ (for R scripts)
  • output/ (for results, plots, and processed data)

Data structure all_datasets.RData

• Download all_datasets.RData and store it in your newly created Folder.
• Open RStudio and load all_datasets.RData (e.g. from the Files pane).
• After loading your Environment will show that the file contains three data frames:
  • all_data has one observation per respondent sampled (incl. dropouts and screenouts)
  • complete_data is a subsample of all_data that excludes dropouts and screenouts
  • database has ten observations per respondent (excl. dropouts and screenouts)

Data frame database

• This is the data frame required for analyzing the DCE
• Differs from complete_data because it contains one observation for each of the 10 choice sets answered.
• All other variables remain constant across the 10 observations per respondent.

DCE Variables

Codebook Excerpt

Data Exploration I

The file all_datasets.RData is already cleaned, but there are two things to keep in mind.

1. Different numbers of observations

In the Environment pane we see different numbers of observations, as explained before.

• However, the number of observations of database should be ten times as high as those of complete_data. Checking the difference:

  nrow(database) - 10 * nrow(complete_data)

• Reason: database still contains observations whose geolocations (database$lat, database$lon) were removed after anonymization because they were outside of Germany.

• We can remove these observations to exclude them from our sample:

  database_cleaned <- database[!is.na(database$lat), ]

Data Exploration II

2. Different numbers of variables

We also see different numbers of variables across the three data frames. Manually inspecting variables via names() is not very useful in the case of 450+ variables.

• We first determine which variables exist in all three data frames:

  commonvars <- Reduce(intersect,
    list(names(all_data), names(complete_data), names(database))
  )

• With 453 variables shared across all three data frames, we can use setdiff() to retrieve the variables that are present only in complete_data:

  setdiff(names(complete_data), commonvars)

• All of these are supplementary information derived from the respondent’s geolocation, so it is not problematic if these are not present.

Data Exploration III

Understand the Data Structure

• To get a first idea of the data structure use dplyr::glimpse():

  install.packages("dplyr")
  library(dplyr)
  dplyr::glimpse(complete_data)

Relevant Variables for DCE Analysis

Seminar Work: The setting

Friedrich Merz wants to assess the potential gains of expanding PA and HNV in Germany. He is confronted with several lobby organisations that lobby for and against an expansion and he has also a tight household. Further, several counties fight about the potential money they gain or have to pay. To get a better idea on whether a law on an PA and HNV expansion is a good idea, he asks an engineering office to assess the potential costs. The costs are given and amount to 1500 Euro per ha. But Merz also wants to know the benefits. How much do people actually benefit from it? Veronika Grimm tells him about the potential of economic valuation and recommends you as an expert in it. Merz calls you and asks you to analise the ValuGaps dataset and guide him on the decision. Unfortunately, he is not an economist, and also quite busy, so that he does not give you more details. He expects a presentation of you in one month.

Your Tasks

• Prepare a presentation to guide the German government on a new law to expand PA and HNV

• Use the ValuGaps dataset to derive economic values for the expansion

• Estimate willingness to pay for 100 ha more PA and and HNV

  • Using choice experiments

  • Using life satisfaction

• Use different functional forms of the utility function and compare model performance

• Only based on your results from the choice experiment and the life satisfaction, what would you recommend?

• Hint: A recommendation is not necessarily a statement on whether or not to do it. It can also relate to the validity of the methods, or a reminder of taking a specific argument into consideration.

Some hints

• Use the scripts from the tutorials as templates
• For some inspiration on the utility functions, see https://mpra.ub.uni-muenchen.de/125429/1/MPRA_paper_125429.pdf
• When writing about the results, make sure to shortly summarize the data, explain the method and then correctly interpret the results
• Hand in your code (that should be executable for us and include many comments in order to understand what you did) alongside your paper
• Look at the material from Econometrics again and google or ask ChatGPT (without uploading the data set) when you have coding questions