The carbon credit system allows companies to finance their ambitions to reduce or remove their greenhouse gas emissions. The integrity of these credits, especially in the voluntary carbon markets, has come under scrutiny over a key factor: additionality.
A project is ‘additional’ if it has helped in reducing green house gas (GHG) emissions. In the context of Reducing Emissions from Deforestation and Forest Degradation (REDD) projects, it means that if it were not for the project the forests would have been cleared.
Researchers at the Climate Policy Initiative (CPI) and the Pontifical Catholic University of Rio de Janeiro (PUC-RIO) have proposed a new method to define baselines and project deforestation on private Amazonian properties in Brazil, that will help to assess ‘additionality’ in REDD projects better. This approach brings economic considerations into land-use dynamics, going beyond statistical models.
Brazil is a likely leader in generating carbon credits, particularly via nature-based solutions like forest conversation.
Checking feasibility
Historically, assessments have relied on statistical and spatial data, often extrapolating historical trends in deforestation. But data could be tweaked to help inflate credit generation. This impacts the credibility of the entire ecosystem.
The proposed model requires landowners in forests to evaluate the most profitable use of their land: maintaining standing forest, engaging in a REDD project, or converting to agriculture. Their decision is influenced by agricultural and carbon prices, transportation costs, agricultural productivity, and carbon stocks, the authors say.
The proposed method calculates the opportunity cost of deforestation. It also helps remove bias as it is agnostic to time windows or reference regions. It operates over a historical period (2010 to the most recent data) and uses a large sample of more than 13,000 properties across the Amazon.
Applying the proposed method to the current set of REDD projects in the Brazilian Amazon shows that 77 per cent of the carbon traded for REDD projects is indeed ‘additional’. . These initiatives have helped prevent the release of an estimated 0.5 giga tons of carbon dioxide (CO2) equivalent.
Notably, 23 per cent of carbon stocks would have remained protected even without project incentives like REDD.
‘Additionality’ varies across regions, according to the study. Non-forested and deforested regions exhibit high rates of additionality (98.5 per cent and 93.8 per cent, respectively), largely because these areas, already consolidated by agribusiness, face immense pressure to clear remaining forest given lower transportation costs and better infrastructure. Conversely, forested regions show a lower additionality of 79.4 per cent.
Researchers cite the example of the municipality of Portel in Pará. Despite high carbon stocks, Portel’s properties in REDD project areas have low agricultural productivity and high transportation costs, rendering agriculture economically unviable. The CPI/PUC-RIO model predicted that these areas would remain forested regardless of carbon credit revenues, thus exhibiting low additionality – or 57 per cent as per the analysis.
So, properties with greater area, higher agricultural productivity, and lower transportation costs are most suitable for REDD projects, as they face the highest risk of deforestation.
What exactly do the researchers include in the new methodology?
The model defines a baseline for a given area, representing what would have happened to the area if the project had not existed.
The model’s core assumption is landowner profit maximisation, and considers various economic conditions and property characteristics, including, agricultural prices; carbon prices; transportation costs (from property to nearest port); distance to the nearest highway; agricultural productivity; carbon stocks of the area; property area; and transition costs; geographic factors; regional effects such as local policies, climatic conditions and economic trends).
Calculating additionality
It first estimates transition probabilities. The model first obtains the probabilities of a property transitioning from forest to agriculture, or to a REDD activity, estimating the likelihood of a property moving from forest to one of the three uses (REDD, forest, or agriculture). Data for these transitions are collected from sources such as MAPBIOMAS (which has records of land converted from forest to agriculture) and Verra projects (from forest to REDD).
Then, equations are developed to describe the revenue associated with each land use: agriculture, standing forest, and REDD.
Next comes the land-use prediction equation, which relates the ratio of probabilities (forest to REDD vs. forest to agriculture) to the present value difference in revenues between REDD and agriculture.
The model parameters are obtained using observed data. With these parameters, the model can evaluate various future scenarios by considering different prices, costs, and the probability of a property owner choosing forest, agriculture, or REDD over time.
The study proposes a baseline scenario where carbon projects (REDD) do not exist. In this baseline, the probability of a given property being deforested without carbon credit revenues is calculated.
Additionality is then calculated as the difference between the amount of deforestation in this baseline scenario and the amount deforested in a “scenario of interest” (the current scenario with REDD projects or a hypothetical one with a specific carbon price).
Published on September 15, 2025