Are there Carbon Savings from U.S. Biofuel Policies? The Critical Importance of Accounting for Leakage in Land and Fuel Markets.

• Author: Bento, Antonio M.

1. INTRODUCTION

   Although the costs of comprehensive U.S. federal climate legislation, such as a cap-and-trade program, are shown to be rather small (CBO 2009), a variety of political obstacles continue to block its passage. Policymakers have instead relied on sectoral and regional approaches to reduce greenhouse gas (GHG) emissions. (1) A major concern associated with sectoral and regional approaches to climate policy relates to their effectiveness in reducing GHG emissions (Bushnell, Peterman, and Wolfram 2008; Goulder and Stavins 2011). Such approaches are incomplete, in that only a subset of polluting sectors or regions are regulated. As a consequence they are likely to generate carbon leakage. Carbon leakage occurs as sectors or regions not covered by the regulation respond to the regulation (directly or indirectly) (Goulder, Jacobsen, and van Benthem 2012). When it comes to sectoral approaches to climate policy, policies that call for the expansion of liquid biofuels have been especially scrutinized by environmental groups and the popular press. Yet, to date very few studies have examined the carbon leakage that results from biofuel policies, and typically only consider a single source of leakage. (2)

   The purpose of this paper is to provide comprehensive estimates of carbon leakage from the Renewable Fuel Standard (RFS) for conventional biofuels. The RFS mandates quantities of conventional and advanced biofuels, with each biofuel class defined according to its lifecycle emissions savings relative to gasoline. (3) The current RFS was established in 2007 when the Volumetric Ethanol Excise Tax Credit (VEETC)--the long-standing federal biofuel subsidy--was in place. However, the VEETC was allowed to expire at the end of 2011, leaving the RFS as the primary biofuel support program in the U.S. Our analysis of the RFS explicitly accounts for these changes in policy regime, and reviews the impact of current proposals to eliminate the RFS for conventional biofuels altogether.

   This paper addresses three related questions. First, what are the effects of the RFS on land and fuel markets? Second, what is the impact of the RFS on overall GHG emissions, and how does carbon leakage in land and fuel markets cause overall emissions to deviate from the intended emissions savings anticipated by legislators at the time the RFS was passed in 2007? Third, what is the impact of the change in policy regimes and current proposals to eliminate the RFS on overall GHG emissions and leakage due to the RFS?

   Several prior studies have examined the emissions impacts of biofuels, although none have simultaneously examined these impacts in the context of past, current, and proposed policy regimes. One strand of the literature relies on lifecycle methods, without reference to a particular biofuel policy. For example, in their seminal work, Farrell et al. (2006) argue that the lifecycle emissions savings of ethanol relative to gasoline are 18%. Many studies recognize that biofuel policies can lead to various multi-market adjustments. However, most develop models to explicitly capture adjustments in either fuel (Khanna, Ando, and Taheripour 2008; de Gorter and Just 2009; Rajagopal, Hochman, and Zilberman 2011; Hochman, Rajagopal, and Zilberman 2011; Drabik and De Gorter 2011; Thompson, Whistance, and Meyer 2011; Rajagopal and Plevin 2013) or land markets (Searchinger et al. 2008; Hertel et al. 2010; EPA 2010a) either abstracting from adjustments in the excluded markets altogether or assuming constant adjustments and/or emissions factors in the excluded market per unit of biofuel added. For example, Thompson and coauthors (2011) analyze the RFS in a framework that includes world fuel markets and U.S. agricultural markets, but do not link the emissions calculations directly to land market adjustments. Similarly, Rajagopal and Plevin (2013) perform a Monte Carlo analysis to quantify uncertainties in the GHG impacts of biofuel policies using a model of world fuel markets that includes emissions resulting from land market impacts as uncertain parameters that are constant per unit of fuel. The U.S. Environmental Protection Agency's (EPA) Regulatory Impact Analysis of the RFS (EPA 2010a), which is the most comprehensive analysis of the RFS to date, considers the GHG implications of biofuels expansion using several sophisticated domestic agricultural and global land use models, but does not quantify the GHG implications that result from adjustments in fuel markets. There are a few studies that consider both land and fuel markets. A set of studies uses the Biofuel and Environmental Policy Analysis Model (BEPAM), which integrates U.S. land and world fuel markets, to analyze first and second generation biofuel policies along a number of dimensions. Chen et al. (2012) examines the changes in domestic land use and emissions resulting from the RFS in 2022. Chen et al. (2011) compares the welfare implications of the RFS, LCFS and a carbon tax in 2030. Huang et al. (2013) examines the welfare and GHG impacts of combining the RFS with a LCFS and carbon price policy. A common feature of the BEPAM analyses is that biofuel policies will cause large expansions in cellulosic ethanol and feedstocks, and relies on assumptions by the EIA regarding future penetration of E-85 automobiles.

   Our study differs from earlier work in several ways. First, we develop an analytic and numerical multi-market model that consistently integrates fuel, food and land markets. We link this multi-market model with a sectorally disaggregated emissions model. While some studies (e.g. Chen et al. (2011) and Huang et al. (2013)) also examine the impact of the RFS on total emissions using models that integrate land and fuel markets, our goal is to understand how the emissions consequences of the RFS differ from those intended. We derive an analytical formula that decomposes the overall change in GHG emissions that result from an increase in the RFS into intended emissions savings and carbon leakage. Intended emissions savings are calculated with standard lifecycle methods that reflect the GHG emissions savings resulting from replacing a unit of gasoline with a unit of ethanol, scaled up by the amount of ethanol added to the economy as a result of the RFS. Carbon leakage emerges from adjustments in land and fuel markets, both domestic and international, as the RFS impacts key prices. This decomposition is of critical importance to public policy as it directly illustrates the dangers of including LCA metrics in federal legislation as a criteria to select biofuel feedstocks. (4) The analytical formula guides the presentation of our simulation results, and provides a consistent frame of reference for comparing the magnitudes of leakage under various policy regimes and parameter assumptions. Our numerical results uncover a codependency between land and fuel market leakage that reflects the underlying economic relationships. For example, policy regimes with less land market leakage emerge because the policy causes a smaller increase in the price of corn per liter of ethanol added. As a result, the price of blend fuel is more likely to decline, resulting in larger fuel market leakage. This suggests that the integration of land and fuel markets is critical for estimating leakage from either market, and for quantifying the total change in GHG emissions due to biofuel policies.

   Second, we examine the RFS through the lens of past, current, and proposed policy regimes and therefore are able to shed light on how policy interactions and changes in policy play an important role in the direction and magnitude of leakage due to the RFS. Relative to a baseline that includes the VEETC we consider two policy regimes, one in which the RFS is added to the preexisting VEETC and a second regime in which the RFS replaces the pre-existing VEETC. We include the VEETC in the baseline in our central analysis because this allows us to understand the emissions implications of the RFS from the perspective of policymakers at the time the RFS was enacted. The first policy regime allows us to isolate the impact of just adding the RFS to the economy and reflects the policies in place prior to 2011. The second regime allows us to isolate the impact of replacing the VEETC with the RFS--jointly removing the VEETC while imposing the RFS--and reflects the policies in place from the end of 2011. To understand the implications of current legislative proposals to eliminate the conventional RFS entirely, we consider a third policy regime that examines the impact of adding the RFS to a baseline without the VEETC in place.

   Third, by focusing on the RFS for conventional biofuels through 2015 our estimates of the emissions resulting from the RFS will be unencumbered by assumptions regarding second generation biofuels. We are able to safely ignore second generation biofuels because mandated and realized volumes of second generation biofuels are likely to be negligible over the time horizon of our study. (5) An analysis of the RFS through 2022 would require strong assumptions to dictate the emergence of second-generation biofuels, such as farmers' willingness to plant second-generation feedstocks, the yields of second generation feedstocks, the marginal costs of producing second generation biofuels, and the emergence of E-85 vehicles. These assumptions will also affect price adjustments in land and fuel markets, and therefore leakage, due to the RFS.

   Our central finding is that the expansion of biofuels mandated by the RFS can increase or decrease GHG emissions depending on the policy regime being evaluated. Relative to a baseline that includes the VEETC, the RFS increases emissions by 4.5 tgC[O.sub.2]e in 2015 with our central parameters. (6) Emissions increase because the intended emissions savings due to the RFS are offset by considerable leakage in land and fuel markets, 80% and 60% of intended emissions savings...