Closed ncx-gitbot closed 1 year ago
NCX response: Leakage is poorly studied across existing IFM projects, especially short-term harvest deferral projects. We agree that leakage is a possible outcome of purposefully delaying a harvest. Based on comments received, we have updated the methodological approach to include a more conservative deduction. We look forward to working with other developers and academic researchers to explore methods of measuring leakage directly in the future.
Commenter Organization: CarbonPlan
Commenter: Danny Cullenward
2021 Deferred Harvest Methodology Section: No Section Indicated
Comment: The Proposed Methodology does not adequately justify the leakage deductions applied to carbon credits issued for harvest deferrals. To the extent baseline models can accurately predict the probability of timber harvest, the same models can and should be used to provide leakage estimates. Leakage and additionality are closely related. To the extent a landowner is actually deferring harvest, leakage risks increase because deferred harvests from one parcel can lead to increased harvests elsewhere. Accordingly, it is important to connect an additionality finding to the specific mechanisms that might result in leakage. The Proposed Methodology considers but fails to adequately address two types of leakage: (1) leakage from activities shifting within a landowner’s operations and (2) leakage from the market effects of reducing harvest. 15 First, the Proposed Methodology assumes there will be no activity shifting leakage based on the premise that owners/managers must enroll the entirety of their holdings, 16 but uses imprecise language that fails to ensure this outcome. The Proposed Methodology asserts that a participating landowner must enroll all forested property within the “program area” of the credited project. However, the Proposed Methodology allows project proponents to define each project’s “program area.” 17 In other words, project proponents get to define what satisfies the Proposed Methodology’s requirement of enrolling all relevant lands on a project-by-project basis, without any programmatic guardrails. With such broad and flexible definitions, it is easy to imagine a program area that intersects only a portion of a large landowner’s holdings and thus allows activity-shifting leakage that would nevertheless be ignored under the Proposed Methodology.Second, the Proposed Methodology adopts default Verra leakage deductions to account for market-wide effects of deferring harvest. 18 These deductions are not based on clear evidence, nor calibrated to the specific mechanics of the Proposed Methodology. It is not uncommon for forest offset programs to use simple assumptions precisely because it is so challenging to robustly measure leakage outcomes. For example, in a 2019 public letter reviewing the limited number of studies estimating forest carbon leakage rates, Duke University Professor Brian Murray wrote that: “The empirical work is not easy and I do not pretend that the estimates from my work with others, generated more than ten years ago [in 2004], focused on hypothetical programs are precise estimates of what happens today with real programs. But to my knowledge, they are the only (or perhaps one of a few) peer-reviewed estimates of carbon leakage in US regional programs out there.” 19 Although we are sympathetic to the challenge deep uncertainty presents for climate policy decisions, and appreciate Dr. Murray’s characterization of the limited evidence available today, the Proposed Methodology is premised on the notion that harvest rate probabilities can be accurately characterized in baseline modeling. If accurate, these harvest prediction models should also be capable of calculating market leakage. 20 If baseline harvest models are not capable of characterizing leakage, however, this implies that they are also incapable of accurately predicting the risk of harvest and should not be used as the basis for crediting.
Proposed Change: We recommend Verra either (1) introduce an activity shifting leakage factor, or (2) establish clear and appropriate guardrails on the definition of a program area. We also recommend that Verra (3) require any model deemed sufficient to predict baseline harvest risk probabilities for the purposes of additionality findings also be used to evaluate the adequacy of assumed leakage factors for market effects.