Biofuels, Hunger, and Climate: Fixing California's LCFS Analysis
- Apr 6, 2015 6:00 pm GMTJul 7, 2018 9:16 pm GMT
- 420 views
The politics surrounding biofuels and biomass-based power generation are so fraught with controversy that the Washington Post recently labeled bioenergy “a familiar obstacle to good policymaking.” Who could disagree, then, that decisions about which types of bioenergy to encourage should be based on dispassionate scientific analysis rather than the parochial interests of lawmakers? Instead of “picking winners,” shouldn’t we let lifecycle analyses (LCAs) sort out which fuels make environmental sense?
Maybe not. An important new study in the journal Science finds that agencies in the United States and Europe use LCAs in ways that obscure the extent to which policies that promote biofuel production “are relying on reductions in food consumption to generate GHG [greenhouse gas emissions] savings.” The study’s timing is auspicious, in that it tackles an issue that currently confronts the California Air Resources Board (ARB).
The new study—“Do biofuel policies seek to cut emissions by cutting food?”—examines the lifecycle models that ARB, the US Environmental Protection Agency, and the European Commission use to assess whether the net GHG emissions from biofuels meet the emission reduction requirements of California’s Low Carbon Fuel Standard, the US Renewable Fuel Standard, and Europe’s Renewable Energy Directive. Timothy Searchinger of Princeton University and the other modelers that authored the Science study note that when farm crops are used to make biofuel instead of food or animal feed, “some of the food may not be replaced, meaning that someone will eat less or less well.” The disruption in food and feed markets is felt primarily in developing countries: “any reduction in global food consumption is likely to disproportionately affect the some groups of the poor because they can less afford high food prices.”
What percentage of crops diverted to biofuel refineries are not replaced in the food market? After digging into the LCAs, the authors found
These models estimate that roughly 25 to 50% of the net calories in corn or wheat diverted [from food and feed markets] to ethanol are not replaced but instead come out of food and feed consumption … Replacing fewer crops leads to lower GHG emissions from [land use changes]. It also reduces direct emissions of carbon dioxide (CO2) by people and livestock.
The impact that reduced food consumption has on biofuels’ modeled emissions performance is, in a word, significant. According to the study:
Without crediting reduced food consumption, none of [the models relied upon by EPA, the European Commission, or California ARB] would project lower GHGs for ethanol than for gasoline …. In that sense, the lower GHGs for ethanol depend of reductions in food consumption.
When the US Congress and then-President George W. Bush drastically expanded the scope and duration of the Renewable Fuel Standard (RFS) in 2007, they agreed, somehow, that lifecycle greenhouse gas (GHG) emissions should be the key criterion for determining which biofuels qualified as “advanced” under the RFS. (A grandfathering provision in the legislation spares the vast majority of corn ethanol refiners from having to prove their product reduces GHG emission relative to gasoline; not coincidentally, EPA calculations analyzed here and here show that the net GHG emissions from corn ethanol produced during 2010-2015 exceed those from an energy equivalent volume of gasoline by 28%.) Other policies, including California’s LCFS and Europe’s Renewable Energy Directive, followed suit by preferencing biofuel types based on an assessment of the fuels’ lifecycle GHG emissions.
Environmental groups like the the one I work for have long supported the use of LCAs to model the net GHG emissions associated with the production of different biofuels. In theory, lifecycle analysis helps steer investment toward climate-beneficial technologies and provides regulators with a transparent method for ensuring that their policies are promoting biofuels that either meet minimum performance benchmarks (in the case of the RFS) or reduce GHG emissions more efficiently than other fuel options (in the case of the LCFS). In theory, rewarding biofuels based on their LCA-modeled emissions performance makes more sense then trying to pick winners, given how inept we’ve proven to be at predicting how biofuels will impact other markets or when biofuels will become commercially available. And, at least in theory, LCA-backed policies that become more stringent over time will eventually dislodge corn ethanol and the other kinds of conventional biofuels that are most disruptive to food markets.
In practice, however, the reliance on LCA to regulate biofuels has become so problematic—and, occasionally, so misleading—that a once-heretical proposal is beginning to gain steam: Would we be better off trying to pick winners after all?
The internal mechanics of the models used to project biofuels’ lifecycle emissions—especially the models that attempt to quantify the emissions resulting from indirect land use change (ILUC) and other market-mediated impacts—are so monstrously complex that even the people who designed them can be hard-pressed to explain certain outputs. One recent and especially troubling example (described in more detail here) involves the extent to which water availability governs the geography of agricultural expansion. Researchers at Purdue University determined that the lifecycle model they developed “likely underestimated induced land use emissions due to ethanol production by more than one quarter,” due to its inability to account for irrigation constraints. However, when California ARB attempted to retool its own version of the Purdue model to account for water availability, the change in estimated emissions was negligible. It’s not clear why.
Developing the relevant data and determining which datasets to use (and which to exclude) in a lifecycle model are subjective exercises marked by a high degree of epistemic uncertainty, as are the processes of choosing and programming the relational assumptions that drive the model. Viewed in this context, a recent proposal by California ARB staff to reduce corn ethanol’s ILUC score can be more appropriately understood as the product of a subjective process—one that reflects the current availability of certain data and analyses that would contribute to a lower ILUC score, but fails to account for a host of countervailing factors that ARB knows are significant but is not yet able to model.
The new Science study illustrates the extent to which California ARB’s decision to count GHG reductions that result from reduced food consumption when analyzing the lifecycle emissions of biofuels is both subjective and consequential. Lifecycle GHG emissions from corn ethanol are 5% lower than those from gasoline when ethanol gets credit for the emission reductions that California’s LCA attributes to reduced food consumption. When those emission reductions are not counted, corn ethanol’s lifecycle emissions are 40% higher.
Previous studies have come to similar conclusions. A separate 2015 study led by Richard Plevin, one of the co-authors of the Science paper, notes that “ILUC emission estimates [in the LCA used by California ARB] depend on various modeling choices, such as whether a reduction of food consumption resulting from biofuel expansion is treated as a climate benefit.” Plevin et al. find that if ARB were to hold food consumption constant (or fixed) in developing countries (by assuming that governments would subsidize food imports, for example), ILUC emissions for corn ethanol would increase by more than 5 gCO2/MJ as compared to a scenario in which food consumption is not fixed. Total emissions from corn ethanol under a “food fixed” scenario increase by approximately 10 gCO2/MJ. Likewise, an earlier study co-authored by Thomas Hertel, one of the architects of the Purdue LCA that ARB uses, found that if food consumption were held constant in the model, estimated emissions from biofuel expansion would increase by 41%.
We’ve urged California ARB to recognize these limitations, as well as the necessary role that it plays in interpreting and acting upon modeling results. ARB has to exercise its best judgment in light of the overarching policy objective of the LCFS, which we understand to be a non-trivial reduction in GHG emissions from California’s transportation sector. Because corn ethanol’s lifecycle GHG emission reductions—which, at best, are very modest to begin with—depend on an assumption of reduced food consumption in developing countries, and because increased reliance on corn ethanol would frustrate the development of more innovative and effective compliance options, ARB’s proposal to reduce the ILUC score for corn ethanol undermines the objectives of the LCFS and should be tabled.
Despite their shortcomings, it’s too soon to give up on biofuel LCAs. Lifecycle GHG emission models can help characterize the GHG performance of biofuels in general terms, but only if policymakers at California ARB and other agencies ensure that LCAs steer us toward, rather than away from, our policy objectives. In the new Science paper, Searchinger et al. offer two suggestions: modelers should be transparent about food-versus-ILUC—type tradeoffs that occur within LCAs; and agencies should eliminate the food consumption reduction credit when calculating a biofuel’s lifecycle GHG score, assuming they “do not want to mitigate climate change in this way.”