Ready-made oTree applications for the study of climate change adaptation behavior

Catharina Wolff von Bülow, Xiufeng LiuJournal of Behavioral and Experimental Economics 2020 · DOI

TL;DR

This paper introduces CAT (Climate Change Adaptation Applications), a set of ready-to-use oTree modules designed to experimentally study how individuals and groups make decisions about adapting to climate change. By simulating realistic scenarios and incorporating behavioral economics principles, CAT provides a flexible and accessible tool for researchers to investigate adaptation behaviors and inform policy.

What Problem Does This Paper Address?

Climate change adaptation is a complex decision-making process for individuals and communities, involving balancing costs against uncertain benefits in high-stakes, low-probability scenarios. While research has explored various adaptation strategies, there's a need for effective tools to better understand the behavioral constraints, biases, and uncertainties that influence these decisions at the individual and community levels. Existing experimental frameworks often require significant customization, hindering broader application in climate change research. This paper addresses this gap by developing a standardized, web-based toolkit that simplifies the experimental design and implementation of studies on climate change adaptation behavior.

Methods

The core of this work is the development of CAT, a set of eight ready-to-use oTree applications. oTree is an open-source platform for running online social science experiments. CAT leverages oTree's capabilities to create a flexible framework for studying climate change adaptation. Each CAT app is designed to simulate specific aspects of adaptation decisions, incorporating various "treatments" that correspond to known behavioral biases and interventions. These include:

  • Control Treatment: A baseline for comparison.
  • Ambiguity Treatment: Introduces uncertainty by providing probability distributions but not exact probabilities.
  • Loss Aversion Treatment: Frames choices in terms of potential losses.
  • Myopia Treatment: Models short-sighted decision-making where immediate outcomes are prioritized.
  • Coordination Treatments (with and without ambiguity): Examines how group dynamics and the potential for cost-saving collective action influence adaptation.
  • Communication Treatment: Investigates the role of communication in group adaptation decisions.
  • Survey App: Collects demographic and attitudinal data from participants.

The paper also outlines four experimental design structures (A-D) that can be employed with CAT, allowing researchers to tailor their studies to specific hypotheses. A pilot experiment with 75 subjects was conducted to evaluate CAT's functionality and gather initial insights.

Datasets

The paper does not introduce new datasets for training. Instead, it describes the data collected from a pilot experiment involving 75 participants. This pilot experiment serves as a benchmark for evaluating the CAT applications and generating initial findings on climate change adaptation behavior.

Key Insights and Ideas

  • Adaptation Increases with Risk: Consistent with economic theory and previous studies, the adaptation rate generally increases as the perceived risk level rises.
  • Uncertainty Drives Adaptation: When risk information is unknown (ambiguity), adaptation rates tend to be higher compared to situations with known risk. This suggests that uncertainty can be a stronger motivator for adaptation than precise risk information.
  • Group Dynamics Polarize Decisions: In group settings, decisions tend to polarize towards either unanimous adaptation or no adaptation at all. This contrasts with some previous findings and might be influenced by speculative behavior or the desire to align with group norms.
  • Individual Preferences Matter: Subject characteristics like age, gender, and property ownership significantly influence adaptation decisions, highlighting the importance of considering individual-level factors in policy design. For instance, homeowners are more likely to adapt than renters, and women tend to be more risk-averse than men.
  • Communication's Nuanced Role: While communication in groups can encourage collective adaptation, its effect is not always pronounced and can be influenced by the presence of coordination mechanisms.

Key Results

The pilot experiment with 75 subjects provided initial evidence supporting several hypotheses:

  • Risk Level: Adaptation rate increases with risk level (H1a supported).
  • Loss Frame: A loss frame generally leads to less adaptation, though this effect can be mitigated at high risk levels (H1b partly supported).
  • Myopia: A longer time horizon has a limited impact on reducing adaptation, with some risk levels showing less adaptation (H1c partly supported).
  • Ambiguity: Adaptation rate increases with uncertainty (H1d supported).
  • Group Ambiguity: Individuals in groups are more ambiguity-seeking than those acting alone (H2 supported).
  • Coordination: Coordination treatments lead to more polarized decisions and increased collective adaptation as group risk rises (H3a, H3b, H3c supported).
  • Communication: Communication treatments show some support for increased adaptation and more frequent unanimous decisions (H4a, H4b, H4c partly supported).

The logit regression analysis indicated that risk level, property type, gender, and age are statistically significant predictors of individual adaptation decisions.

Limitations and Future Work

The authors acknowledge several limitations and suggest future research directions:

  • Bonini's Paradox: As models become more realistic by incorporating more factors, they can become as complex as the real process they aim to explain. The authors recommend limiting the number of treatments per session and focusing on key variables to maintain clarity.
  • Instruction Clarity: Subjects sometimes preferred instructions embedded within the program rather than a separate booklet. Future implementations could incorporate pop-up screens for instructions.
  • Speculative Behavior: The pilot test suggested that keeping risk constant when collective adaptation is not chosen might encourage speculative behavior, which could be addressed in future app designs.
  • Future CAT Enhancements: CAT could be improved by adding features like varying adaptation costs within groups, modeling different degrees of adaptation effectiveness, and incorporating options to postpone decisions to gather more information.
  • Broader Applications: CAT could be extended to other behavioral economics games (e.g., public goods games, housing market games) and tested in field settings to investigate adaptation in real-world flood scenarios and the role of political intervention.
  • Cross-Regional Experiments: Comparing adaptation decisions between individuals directly affected by climate disasters and those who are not could provide valuable insights.

Summary updated

BibTeX
@article{bulow2020ready,
  author = {Catharina Wolff von Bülow and Xiufeng Liu},
  title = {{Ready-made oTree applications for the study of climate change adaptation behavior}},
  journal = {Journal of Behavioral and Experimental Economics},
  year = {2020},
  doi = {10.1016/j.socec.2020.101590},
}
Figure