I4R x R2E-Defor: Robustness and Replicability in the Deforestation Literature

An expertise-enhanced meta-replication

How robust is the evidence on what drives deforestation and which policies can reduce it?R2E-Defor systematically examines the replicability and robustness of 14 influential studies estimating causal effects on deforestation, published between 2012 and 2022 in leading economics and interdisciplinary journals. Independent replicators follow a common protocol, supported by the I4R core team and seven deforestation experts.

Why does this matter?

Evidence on deforestation informs climate and conservation policy, such as market-based carbon pricing ambitions. Yet the heavy reliance on complex datasets and methodologies suggests that findings might depend on analytical choices and on which satellite-based data are used to measure forest loss. Testing the sensitivity of these findings is crucial for how they are used to inform policy. Also, beyond deforestation, we want to improve our understanding of how to integrate domain expertise in replication research.

What we’re doing

For each study, replicators:

• Computationally reproduce the original results and audit the code for errors
• Systematically vary reasonable and conservative analytical choices using a common robustness protocol
• Test the replicability of estimates using alternative satellite-based deforestation datasets

A central feature is the close collaboration with renowned deforestation experts, combining expertise in causal inference, remote sensing, and the policy contexts. Experts review the replications at different stages, for example during a dedicated two-day workshop where replicators and experts jointly examined replication methodologies and initial results.

Our experts

The project’s expert team comprises:

• Jill Caviglia-Harris (Salisbury University)
• Paul Ferraro (Johns Hopkins University)
• Subhrendu Pattanayak (Duke University)
• Erin Sills (North Carolina State University)
• Richard Tol (University of Sussex)
• Laura Villalobos (Salisbury University)
• David Wüpper (University of Bonn)

What comes next

• 14 individual replication reports as standalone scholarly contributions
• A special issue at a leading field journal
• A meta-paper synthesizing evidence on reproducibility, analytical robustness, and sensitivity to alternative deforestation data

The individual reports will be released gradually, with the meta-paper planned for 2027.