U.S. soy exports go downand world soy prices rise

Market-Mediated Land Use Change ConsequencesOf Crop-Based Biofuel Production

Andrew Jones, Richard Plevin, Michael O’Hare, & Alexander Farrell • University of California – Berkeley, CA

Supported by EPA STAR Graduate Research Fellowship and California Air Resources Board

INTRODUCTIONGrowing biofuel feedstocks on prime agricultural land influences theglobal food system and causes indirect, off-site changes in land use andland cover via agricultural market signals. In addition to emissions ofgreenhouse gases, these changes affect biodiversity, regional climate, soilquality, and water quality.

New biofuel and low-carbon fuel policies, such as the EnergyIndependence and Security Act of 2007 and California’s Low Carbon FuelStandard, require that the life-cycle greenhouse gas emissions of fuels bemeasured and reported. In addition, there are several efforts under way toestablish sustainability criteria for biofuels. Thus it is essential that weunderstand the indirect effect of biofuel production on land use and landcover, as well as the impacts of those changes on greenhouse gasemissions and other sustainability metrics.

Estimating market-mediated land use effects requires both economic andbio-physical models, which are fraught with methodological challengesand data uncertainties. Characterizing these uncertainties will inform thepolicy making process and guide research toward issues that will increasethe ability of decision-makers to mitigate climatic and ecological effects ofnew fuels.

Process Emissions

U.S. corn farmer switchesfrom corn/soy to corn/corn

New Demand

SIZE OF EFFECT PRELIMINARY RESULTS

Considering land use change !

Searchinger et al 2008 demonstrated thatthe indirect land use emissions of biofuelsmay be very large, possibly outweighingall other sources of greenhouse gases inthe life-cycle of corn ethanol andswitchgrass ethanol grown on prime cornland.

Net energy and net GHG estimates for 6 studies of corn ethanol, as well as 3cases. Gasoline is shown for reference. The cellulosic case is switchgrassgrown on prime crop land. Adapted from - Farrell et al, 2006.

time

GHG Conversion

Operation

Gasoline

Time30yr

1.0Discounting (5%)CalamityHorizonGeneric

Figure2:Possible social costof physical GHG release functions. Conventionaleconomic discounting is shown for comparison (see text)

tc

Traditional life-cycle assessmentconsiders industrial processesand direct agricultural emissionsas well as upstream processessuch as fertilizer production.

Market-Mediated effects arecaused by price changes, andhave often been ignored in life-cycle assessment. They can beassessed with economicequilibrium models.

Develop a meta-model that permits comparisons across studies as well as montecarlo and sensitivity analysis

Characterize the distribution of final effects given what we know about inputparameters.

Identify what conditions must be true to minimize the effect

Identify parameters that contribute most to final uncertainty.

Biophysical Uncertainties Economic Uncertainties Conceptual Uncertaintiesecosystem types converted elasticities of substitution amortizationcarbon stocks per land technological innovation derating of futureportion of carbon released baseline demands discounting of futurealbedo changes availability of lands consistency of GWI valueshydrocycle changes trade policies role of land reversionnitrogen cycle disruption regulationsrole of short lived gases investment dynamics

Schematic GHG releases over time Possible Social Cost Funtions

Potential GHG release from land conversion, biofuel production, and ultimate landreversion are shown in red, compared to constant emissions from gasoline in black

METHODS

Indirect ProcessEmissions

Indirect Land CoverChange Emissions

Soy farmers everywhereuse more inputs to

increase yields

Intensification

Price Adjustment

World consumptionof soybeandecreases

Substitution

Additional land in Brazil(for instance) is put into

soy production

Extensification

We performed monte carlo simulation on the land use analysis of Searchinger et al. 2008.The economic modeling results were held constant, while assumptions about the partition ofecosystem types affected in each region were varied along with assumptions about above-ground carbon stocks, soil carbon stocks, and the percentage of soil carbon lost due toconversion. We examined the literature cited by Searchinger et al. to estimate distributionsfor each of these parameters. We found that uncertainty in physical parameters did notqualitatively affect the result that corn ethanol and switchgrass ethanol grown on corn landcreate more greenhouse gas emissions than gasoline.

Distribution of Life-Cycle GHG Emissioms from Corn Ethanol

Gas = 92

Consideration of uncertainties in carbon stocks and ecosystem types affected did not qualitatively affect the ranking ofcorn ethanol relative to gasoline.

Contribution of the Most Influential Parameters to the Final Variance

More than 75% of the variance isexplained by 4 parameters (out of ~120)

Inclusion of uncertainty in additionalparameters such as crop yield andgreenhouse gases from corn productioncreated more final variance, but still didnot qualitatively affect the result.

DISCUSSIONThe market-mediated climatic land use effect of crop-based biofuels appears to be verylarge and policies seeking to mitigate climate change via biofuels must consider it. Whileuncertainties in physical parameters such as carbon stocks did not qualitatively influencemodel results, uncertainties in the economic model have the potential to do so, particularlyif a larger share of the demand shock is met by intensification and substitution rather thanextensification. More research is needed to characterize and resolve the uncertainties inmodeling this complex phenomenon.