Rethinking ILUC Part 2: Lessons for California and Other Policymakers

By Veronica Bradley, Director of Environmental Science

After CARB’s Public Forum on Land Use Change (LUC) disappointed those hoping to advance low-carbon, crop-based fuels, I want to share how our industry believes policymakers can move forward, whether by refining ILUC modeling or reconsidering the methodology entirely.

In my previous piece, I identified shortcomings in California’s approach but omitted the history of the state’s induced LUC (ILUC) values, which can double the carbon intensity (CI) score of some crop-based fuels. ILUC is a theoretical estimate of emissions attributed to global land conversion modeled as induced by increases in crop-based fuel demand. ILUC modeling arose from reasonable concerns that large-scale displacement of fossil fuels with crop-based fuels could incentivize expansion of cropland onto natural ecosystems, thereby releasing stored carbon and threatening biodiversity. But the trajectory of modeling this risk has shown over time to be less and less of the threat originally raised.

To evaluate this risk, economic equilibrium models have paired market forecasts with global carbon stock data, constructing counterfactual scenarios of how agriculture and energy markets could theoretically evolve under heightened demand for biofuel crops. This is where those who attended CARB’s Public Forum saw debates and presentations on Purdue University’s GTAP-BIO model, International Institute for Applied Systems Analysis’ GLOBIOM model, and Pacific Northwest National Lab’s GCAM model. While a potentially useful exercise, these models are inherently uncertain. Their complexity, reliance on assumptions, and lack of real-world verification make their results best understood as scenario-based indicators of potential risk—not as direct measures of greenhouse gas emissions or environmental harm.

ILUC values from model results do not reflect actual environmental harm.

Many attributes of the modeling from their very basic structure to the parameters they rely on create the variability in their outputs. For example, GTAP-BIO, which is used in California per its regulatory text, is a computable general equilibrium (CGE) model, which means it looks at the global economy as a whole and models how all markets react to the increased demand for crop-based fuels from labor to agriculture to oil to land. GLOBIOM and GCAM, which were also highlighted at the Forum are partial equilibrium (PE) models, models that focus on certain markets, like forestry and agriculture, with greater detail than a CGE model but without the capability to model the entire global economy. What does this translate to? Higher ILUC results from PE models than CGE models based on that very fundamental structure. If you restrict the ways in which markets can adapt to a new normal there will be more impact. This does not mean the lower numbers associated with CGE models are wrong; they simply reflect a hypothetical outcome. It’s a disingenuous argument, then, to say that GTAP-BIO underestimates the risk of ILUC simply because the output is lower than other models assessed.

The agricultural community is quick to highlight that intensification (increases in yield from new technology) has generated significant increases in feedstock supply, without the need to utilize or convert new acres in other regions of the world.

A 45% jump in soybean yields since the 1990s added 15 billion lbs of soybean oil in 2025—exceeding U.S. biomass-based diesel demand in 2024—without requiring a single acre of new cropland.

EPA has nicely laid out the myriad ways in which various ILUC models differ and what may drive material differences in outcomes but not for the untrained eye. I was early in my understanding of ILUC modeling when they published their report, but as I kept asking “why,” the reasons became obvious. The different models exist because they’re asking slightly different questions. And, frankly, different experts have different opinions about how economic markets react or how best to represent changes to the various markets biodiesel and renewable diesel play in—namely, agriculture and energy.

Nevertheless, ILUC values have been incorporated into regulatory frameworks, effectively functioning as precautionary adjustments to account for potential land conversion risk. California and the other low-carbon fuel states as well as ICAO’s CORSIA take these modeled values and add them to the carbon intensity of fuel pathways based on their feedstock, and in the case of CORSIA, sometimes also based on the regional origin of that feedstock. This use underscores these models’ purpose: they are tools for risk mitigation, not definitive evidence of biofuels’ actual climate impact. As such, other less cumbersome tools may be more appropriate to incorporate into clean fuel policy. They certainly should be less controversial than the black box penalties that fuel producers must tack onto their supply chain CI scores regardless of how sustainable their feedstock’s actual supply chain is.

According to the International Energy Agency (IEA), growing the sustainable fuels industry will bring millions of jobs to the regions where these fuels are produced and, importantly, deliver “sustained emissions reductions while improving domestic energy security and creating new economic development opportunities.” IEA suggests this is all possible so long as we use transparent and robust carbon accounting methodologies that enhance comparability and interoperability. The right methodologies will enable performance-based policies and incentives that reward progress in measurable emission reductions.

With this in mind, Clean Fuels is taking IEA’s recommendations to heart. We are working on a framework that captures the concerns of ILUC risks while rewarding stewardship, transparency, and continuous improvement across our supply chains. We believe that this approach should reflect key principles that guard against unintended consequences while enabling continuous improvement. This framework and its implementation should therefore:

1. Reflect the risk tolerance of the environment in which it operates. U.S. programs in states with a heavy agriculture presence have more to gain in economic development than risk of inducing deforestation, so their policies should reflect that balance.
2. Mitigate the risk of converting high carbon stock and biodiverse lands. If the real concern is tropical deforestation because tropical forests are our planet’s lungs, sequester the most carbon, and hold a disproportionate number of the world’s known species, then we must focus on what is happening on the ground there.
3. Be technology neutral. As I previously put forth, the problem is not a soy plant, it’s anything taking up the land of previously high-carbon stock and biodiverse land.
4. Dynamically capture changing conditions at different geographical scales. A U.S. produced biodiesel made from oil processed in the U.S. from soybeans grown in the U.S. where the local, state, and federal governments have not only restrictive land use laws but also strict enforcement mechanisms, should have a bright green light. Conversely, a biodiesel made from soybean oil that has its roots in a region that has questionable land use laws or questionable enforcement of the laws on the books should be treated less favorably.
5. Reward process improvements. Effective decarbonization demands balancing social, economic, and environmental priorities while relentlessly pursuing better outcomes. This means centering our decisions on practices that improve the long-term sustainability of the land we depend on.
6. Consider the informal economy and fraud prevention. Even with good conservation laws in the books, governments’ ability to prevent deforestation is only as good as their enforcement mechanisms that deter bad actors.

We are actively working to flesh out how these principles can be operationalized to create a simplified and transparent system for policymakers to implement in their low-carbon fuel programs.

In the interim, we also recognize that complex regulatory processes may slow or limit the ability of agencies to act in the near term. California just finalized its 3-year rulemaking process without touching ILUC values and thereby continuing their reliance on the GTAP-BIO model. But the LCFS also states that CARB should use the best available economic and scientific information, which means that ILUC modeling, where employed, should be regularly updated. We encourage CARB to do what it can to stay true to that mission and for other jurisdictions who see modeled ILUC values as the best fit for their program to reflect the most recent understanding of:

1. Relationships between global agricultural and energy economies.
2. Global land cover and land use classifications and data trends.
3. Land sector carbon stocks.
4. Carbon cycling during land conversion.

These four aspects are all contained in the models mentioned above and are essential to ensuring the models reflect our latest and best understanding of the world we live in. While California relies on global and energy economic information from 2004 for its ILUC values, the modelers at Purdue have updated GTAP-BIO to better reflect the evolution of agriculture and energy markets since then, most recently with the global economy’s state in 2017. They have also recently updated their carbon stock and carbon cycling methodologies to reflect the latest soil carbon databases and recommendations from the Intergovernmental Panel on Climate Change.

At Clean Fuels, we fully embrace making progress each and every day to have science better inform the decisions we make to support the energy transition. We look forward to collaborating with stakeholders who share our dual focus: making near-term policy improvements while working toward larger, lasting reforms that advance economic, social, and environmental wellbeing.