Primary Submission Category: Causal inference for environmental impact evaluation
A Novel Perspective for Carbon Offsetting through the Potential Outcomes Framework
Authors: Megan Ayers, Luke Sanford,
Presenting Author: Megan Ayers*
Forest carbon credits are purchased by entities seeking to offset negative environmental impacts by investing in conservation or restoration projects. Because offsets are generated by projects that avoid emissions or add sequestration, this practice inherently makes causal assumptions and relies on estimates of causal effects, though rarely formalized in practice. Offset calculations depend on emission estimates in counterfactual “baseline” scenarios where no crediting agreements are made. If observed emissions are reduced compared to these estimates, then projects are deemed “additional” and credits are awarded. Accounting for uncertainty and potential biases associated with baseline estimates is critical for the effectiveness of carbon offsetting as an environmental intervention, but current methodologies are inconsistent and often implicitly require unrealistic assumptions. In our work, we seek to close the gap between existing carbon offset methodologies and causal inference by providing: 1. A formalization of carbon offsetting practices within the potential outcomes framework, with accessibility for carbon offset stakeholders in mind; 2. Reformulations of existing baseline estimation practices within this framework; and 3. Conditions for unbiased and/or consistent estimation of the true amount of carbon offset under each baseline methodology. Finally, we assess over 30 complex offset protocols using the framework to provide guidance on when to value existing offsets.