## by Katherine Ricca

**Background and Research Objectives**

The true carbon savings of corn ethanol as a biofuel substitute for gasoline are highly contested. Experts and policymakers alike are unable to agree on the total carbon intensity (CI) for the fuel, a measure of the carbon emissions produced per unit of energy. This uncertainty is largely due to emissions generated by indirect land use change (ILUC), in which changes in agricultural land used to support biofuel production in one region affect land use decisions in others. ILUC CI cannot be measured empirically, therefore the uncertainty largely stems from the assumptions required by modeling, presenting an avenue for estimation improvement. Accurate estimation of the total CI of corn ethanol is critical because key biofuel legislation allocates subsidies based on CI values.

The objective of my thesis is therefore two-fold:

- To decompose the uncertainty in ILUC CI estimates produced by assumptions in the field leading ILUC CI modeling framework.
- To translate this uncertainty into emissions consequences through a policy analysis of the California Low Carbon Fuel Standard (LCFS).

**Dataset Production**

To begin, I must construct a dataset of input and ILUC CI output values simulated by the ILUC CI modeling framework. I do so with Monte Carlo analysis, a methodology that assesses the uncertainty of a model by running a series of random trials that vary the values of input parameters to produce different output values.

**Objective #1: Decomposing Uncertainty in Modeling**

To address the first objective, I regress the varied parameter input values from the Monte Carlo dataset on the ILUC CI output values to determine the impact of each parameter on the final output value of the modeling framework. These relationships are not immediately obvious because the ILUC CI modeling framework consists of a series of models that transform initial inputs through multiple stages.

Results of the regression demonstrate that a single parameter, yield price elasticity, contributes to the vast majority of the variance in ILUC CI estimates. Yield price elasticity captures the response of agricultural yields to changes in commodity prices. Its importance is therefore justified given that increasing or decreasing yield due to price changes will affect the amount of land needed to produce crops for biofuel production, therefore impacting ILUC emissions. Such a strong influence of this single parameter suggests that future research directed at refining the modeling framework should focus funding and time on refining the yield price elasticity value.

**Objective #2: Assessing the Policy Impact of Uncertainty**

To address the second objective, I leverage three key facts about the LCFS. First, the LCFS defines the total CI of corn ethanol as the sum of direct emissions from production and ILUC emissions. Second, the subsidies of the LCFS are determined based on the total CI assumed for corn ethanol by the policy, with a smaller total CI yielding a greater subsidy. Third, the subsidy incentivizes corn ethanol producers to reduce the direct emissions from their refineries, with a greater subsidy generating greater direct emissions reduction. Together, these three facts demonstrate that the ILUC CI value of corn ethanol assumed by the LCFS directly impacts the emissions reductions incentivized by the policy. If the estimated ILUC CI value is too high or too low, the policy will result in forfeited or excess emission reductions respectively. Through a series of equations defined in my thesis, I can use the ILUC CI estimates from the Monte Carlo dataset to generate a range of potential direct CI reductions, and therefore potential emissions reductions under the policy.

Results suggest that the ILUC CI estimate and thus total CI estimate for corn ethanol currently used by the LCFS legislation is likely larger than the true value, generating a subsidy that is too low. This under-subsidy results in forfeited direct CI reductions and therefore forfeited annual emissions reductions of up to 200,000 MT CO_{2}e. These emissions totals are enough to be concerned and support efforts to refine the ILUC CI value of corn ethanol, yet still small enough to inspire confidence in the LCFS policy. As we look towards goals to achieve net-zero by 2050, this thesis ultimately contributes to the literature supporting the use of biofuels like corn ethanol as part of the energy transition.