Citizens’ Valuation of Urban Afforestation as a Function of the Trade-Offs between CO2 Reduction, Co-Benefits and Land Use Conflicts

The objective of this study is to conduct a stated choice experiment to understand the valuation of urban afforestation of German citizens. Specifically, I aim to identify the relative importance of different attributes of urban afforestation, such as the co-benefits of urban trees, land conflicts, location of tree, and carbon sequestration potential and aim to i) identify the attributes of urban afforestation that are most important to urban residents, ii) estimate the economic values that urban residents attach to these attributes, and iii) explore how demographic and socio-economic factors as well as locational variables (i.e. the tree population of the neighborhood) influence choices and can help explain taste heterogeneities.

Study participants will be invited to answer a survey on city climate mitigation and adoption that also included a stated choice experiment on urban afforestation. The survey is structured in four main parts: (i) general introduction with some pre-experimental questions, (ii) introduction to the choice context with some behavioral, context-related questions, (iii) the choice experiment tasks and follow-up questions to the tasks, and (iv) socio-demographic questions. Within the choice experiment, each participants answers six choice sets in which respondents are asked to choose among three alternatives, i.e., two urban afforestation measures (alternative 1 and 2) and a status quo option.

In the choice experiment, each participant will respond to six choice sets. In each set, respondents must choose among three alternatives: two urban afforestation measures (alternative 1 and 2) and a status quo option.

The urban afforestation measures are defined by six attributes with varying levels:
1. Average carbon sequestration: 213 t/CO2, 1067 t/CO2, 2.134 t/CO2, 4.268 t/CO2, 6.401 t/CO2
2. Location: roadside, park, public square, brownfield
3. Alternative land usage: parking space, residential buildings, none
4. Local co-benefits: focus urban heat regulation, focus biodiversity
5. Neigborhood: in participants neighborhood, not in participants neighborhood
6. Financial contribution (in EUR): 12, 24, 48, 60, 120, 240

Trees absorb CO2 from the air as they grow and store it in their leaves and wood. The attribute “Average carbon sequestration” defines the different levels of CO2 reductions and, consequently, contributions to climate protection based on how extensively the urban afforestation measures are implemented. To make the CO2 reduction more tangible, participants see, for each reduction amount, how often they would have to fly the distance from Berlin to Rome to emit the same amount of CO2. For the attribute “Location,” participants are informed about possible locations for additional tree planting. There is a choice of planting along roadsides, in parks, in public squares, or on previously unused brownfields. The attribute “Land conflict” then highlights that space is at a premium in cities, and trees compete for limited space with various elements of urban infrastructure. Therefore, planting additional trees can impact the number of parking spaces or the amount of available building land for new housing. The attribute “Local co-benefits” highlights that urban afforestation can achieve additional local benefits. By selecting certain tree species, some additional benefits can be particularly emphasized. For example, there are trees that are particularly popular with insects and birds, having a positive effect on urban biodiversity (focus on biodiversity). Other trees have a better cooling effect and thus help to reduce heat islands, especially (focus on heat islands). The attribute “Neighborhood” allows differentiating between the values assigned to global and local components of tree planting. Participants are reminded that the closer the tree planting occurs, the more likely they will benefit from local co-benefits. At the same time, they are also more affected by potential land use conflicts, which may result in longer parking searches, more expensive parking, or rising rents due to a lack of housing, or pollen or leaves on sidewalks. For climate mitigation, however, location does not matter since CO2 emissions are distributed throughout the atmosphere. Lastly, the attribute “Financial contribution” informs participants about the funding of the initiative. The measure is financed in part by the city and in part by its citizens through a mandatory annual contribution per household. The annual contribution varies depending on how extensively the measure is implemented. If no measures are implemented, no contribution needs to be paid.

2023-10-29

2024-02-28

Discrete choice experiment among two alternatives (and a status quo option) with six attributes (carbon sequestration, location, local co-benefits, alternative land usage, neigborhood, financial contribution). Each participants answers six choice sets. In each choice set, respondents are asked to choose among three alternatives, i.e., two afforestation measures (alternative 1 and 2) and a status quo option.

Primary Outcome Choice: A variable "choice" takes the value 1 for the chosen alternative and 0 for the alternative 1 or 2 that was not chosen. For the status quo alternative, the dummy variable ‘status quo’ is constructed which takes the value one if the respondent chose the status quo option.

Primary Outcome WTP: I will in addition estimate the mean WTP for each attribute by dividing the estimated mean of the random parameters by the estimated fixed parameter of the attribute "Financial contribution".

In addition to the primary outcomes, some sub-group analyses are planned based on the additional included survey items:
- visits to urban green spaces, satisfaction with urban green spaces
- climate attitudes
- economic preferences, such as time, altruism, reciprocity, trust...
- socio-economics such as age, gender, income, car ownership
- financial and political consequentiality

Lastly, the experimental data are planned to be matched with observational data. The Berlin administration provides an extensive database on the state of trees in the city area. This database is differentiated by neighborhoods and informs about the amount of trees in each neighborhood, ii) amount if trees per kilometer of street, iii) distance between trees in meters, iv) share of newly planted trees, v) felling rate, and iv) percentage change in total tree stock. With these data and the great variety, assumptions about decreasing marginal returns for additional urban nature can be tested on a fine granular level.

I am employing a discrete choice experiment within an online survey to evaluate the valuation of urban afforestation by citizens of a major German city. The survey will be administered by the survey institute Bilendi to a specifically chosen group of participants from their panel database, with a target of 1,000 responses. Participants will be notified about the survey via email, with the added incentive of earning points that can be exchanged for vouchers, cash, or donations.

The sample is restricted to individuals residing in Berlin, ensuring a representative distribution in terms of gender, education (high vs. medium/low), and age to reflect the demographics of the German population.

The final data collection is scheduled for February 2024. A pre-test was conducted in November 2023 with a broader population sample of approximately 260 participants. For the pre-test, a d-efficient design was generated using the Stata command "dcreate" for a multinomial logit model with zero priors for main effects. The priors obtained from the pre-test were subsequently used to estimate a final d-efficient design. The ultimate choice experiment comprises 150 choice sets distributed across 25 blocks. Individuals are randomly assigned to one of the 25 blocks and answer the included six choice sets in a random order.

Participants are randomly assigned to one of the 25 blocks by the survey software and respond to the included six choice sets in a random order. Order effects will be accounted for in the regression analysis.

Individual level.

No

No clusters.

Sample Size: 1000 individuals.

Not applicable.

Not applicable.