Rocky Mountain Farmers Union v. Richard W. Corey

OPINION

Whether global warming is caused by carbon emissions from our industrialized societies is a question for scientists to ponder. Whether, if such a causal relationship exists, the world can fight or retard global warming by implementing taxes or regulations that deter carbon emissions is a question for economists and politicians to decide. Whether one such regulatory scheme, implemented by the State of California, is constitutional under the United States Constitution’s Commerce Clause is the question that we consider in this opinion.

Plaintiffs-Appellees Rocky Mountain Farmers’ Union et al. (“Rocky Mountain”) and American Fuels & Petrochemical Manufacturers Association et al. (“American Fuels”) separately sued Defendant-Appellant California Air Resources Board (“CARB”), contending that the Low Carbon Fuel Standard' (“Fuel Standard”), Cal.Code Regs. tit. 17, §§ 95480-90 (2011), violated the dormant Commerce Clause and was preempted by Section 211(o) of the Clean Air Act, 42 U.S.C. § 7545(o), known as the federal Renewable Fuel Standard (“RFS”). In three rulings issued in December 2011, the district court held that the Fuel Standard (1) facially discriminated against out-of-state ethanol; (2) im-permissibly engaged in the extraterritorial regulation of ethanol production; (3) dis*1078criminated against out-of-state crude oil in purpose and effect; and (4) was not saved by California’s preemption waiver in the Clean Air Act. See Rocky Mountain Farmers Union v. Goldstene (“Rocky Mountain Ethanol ”), 843 F.Supp.2d 1071, 1090, 1093 (E.D.Cal.2011); Rocky Mountain Farmers Union v. Goldstene (“Rocky Mountain Preemption”), 843 F.Supp.2d 1042, 1070 (E.D.Cal.2011); Rocky Mountain Farmers Union v. Goldstene (“Rocky Mountain Crude”), Nos. CV-F-09-2234 LJO DLB, CV-F-10-163 LJO DLB, 2011 WL 6936368, at *12-14 (E.D.Cal. Dec. 29, 2011). The district court applied strict scrutiny, and although it reasoned that the Fuel Standard served a legitimate state purpose, it concluded that CARB had not shown that its purpose could not be achieved in a nondiscriminatory way. Rocky Mountain Ethanol, 843 F.Supp.2d at 1093-94; Rocky Mountain Crude, 2011 WL 6936368 at *15-16. The district court granted American Fuels’s motions for summary judgment on its Commerce Clause claims, and it granted Rocky Mountain’s request for a preliminary injunction, finding that Rocky Mountain was likely to succeed on the merits of its Commerce Clause challenge and raised “serious questions” about whether the Fuel Standard was preempted by the RFS. Rocky Mountain Ethanol, 843 F.Supp.2d at 1103. The appeals of the orders were consolidated.

We hold that the Fuel Standard’s regulation of ethanol does not facially discriminate against out-of-state commerce, and its initial crude-oil provisions (the “2011 Provisions”) did not discriminate against out-of-state crude oil in purpose or practical effect. Further, the Fuel Standard does not violate the dormant Commerce Clause’s prohibition on extraterritorial regulation. We vacate the preliminary injunction and remand to the district court to consider whether the Fuel Standard’s ethanol provisions discriminate in purpose or in practical effect. If so, then the district court should apply strict scrutiny to those provisions. If not, then the district court should apply the balancing test established in Pike v. Bruce Church, Inc., 397 U.S. 137, 90 S.Ct. 844, 25 L.Ed.2d 174 (1970), to the Fuel Standard’s ethanol provisions. The district court is directed to apply the Pike balancing test to the 2011 Provisions for crude oil. Id. To prevail under that test, Plaintiffs-Appellees must show that the Fuel Standard imposes a burden on interstate commerce that is “clearly excessive” in relation to its local benefits. Id. at 142, 90 S.Ct. 844.

I

A

California has long been in the vanguard of efforts to protect the environment, with a particular concern for emissions from the transportation sector. Since 1957, California has acted at the state level to regulate air pollution from motor vehicles. Motor & Equip. Mfrs. Ass’n v. EPA (“MEMA ”), 627 F.2d 1095, 1109 n. 26 (D.C.Cir.1979) (citing 1957 Cal. Stats., chap. 239, § 1). Based on this expertise, “[t]he first federal emission standards were largely borrowed from California.” Id. at 1110 & n. 34.

When instituting uniform federal regulations for air pollution in the Clean Air Act, “Congress consciously chose to permit California to blaze its own trail with a minimum of federal oversight.” Ford Motor Co. v. EPA 606 F.2d 1293, 1297 (D.C.Cir.1979). Section 209(a) of the Clean Air Act expressly prohibited state regulation of emissions from motor vehicles. 42 U.S.C. § 7543(a). But the same section allowed California to adopt its own standards if it “determine[d] that the State standards will be, in the aggregate, at least as protective of public health and welfare as applicable Federal standards.” Id. § 7543(b). Other states could choose *1079to follow either the federal or the California standards, but they could not adopt standards of their own. Id. § 7507. The auto industry strenuously objected to this waiver provision and was “adamant that the nature of [its] manufacturing mechanism required a single national standard in order to eliminate undue economic strain on the industry.” MEMA 627 F.2d at 1109 (quoting S.Rep. No. 403, at 33 (1967)). But Congress decided to encourage California “to continue and expand its pioneering efforts at adopting and enforcing motor vehicle emission standards different from and in large measure more advanced than the corresponding federal program; in short, to act as a kind of laboratory for innovation.” Id. at 1111. So California’s role as a leader in developing air-quality standards has been explicitly endorsed by Congress in the face of warnings about a fragmented national market.

Continuing its tradition of leadership, the California legislature enacted Assembly Bill 32, the Global Warming Solutions Act of 2006. The legislature found that “[g]lobal warming poses a serious threat to the economic well-being, public health, natural resources, and the environment of California.” Cal. Health & Safety Code § 38501(a). These threats included “exacerbation of air quality problems, a reduction in the quality and supply of water to the state from the Sierra snowpack, [and] a rise in sea levels resulting in the displacement of thousands of coastal businesses and residences.” Id. This environmental damage would have “detrimental effects on some of California’s largest industries, including agriculture, wine, tourism, skiing, recreational and commercial fishing and forestry” and would “increase the strain on electricity supplies.” Id. § 38501(b).

Faced with these threats, California resolved to reduce its greenhouse gas (“GHG”) emissions to their 1990 level by the year 2020, and it empowered CARB to design emissions-reduction measures to meet this goal. Id. § 38501(e), (g). In Assembly Bill 32, the legislature told CARB to issue regulations, including scoping and reporting requirements to achieve maximum technologically and economically feasible reductions, see, e.g., id. § 38561(a), a cap and trade program to enforce limits on carbon emissions from a variety of domestic sources, id. § 38562(c), and regulations seeking to reduce GHG emissions from the transportation sector, see, e.g., id. § 38562(a); Cal.Code Regs. tit. 13, § 1961.1.

The Assembly Bill 32 scoping plan required CARB to consider “the relative contribution of each source or source category to statewide greenhouse gas emissions.” Cal. Health & Safety Code § 38561(e). In California, transportation emissions account for more than 40% of GHG emissions—the state’s largest single source. Cal. Exec. Order No. S-01-07 (January 18, 2007). Given the relative import of these emissions, CARB adopted a three-part approach designed to lower GHG emissions from the transportation sector: (1) reducing emissions at the tailpipe by establishing progressively stricter emissions limits for new vehicles (“Tailpipe Standards”), CaLCode Regs. tit. 13, § 1961.1 (2001); (2) integrating regional land use, housing, and transportation planning to reduce the number of “vehicle miles traveled” each year (“VMT Standards”), see Cal. Gov’t Code § 65080; and (3) lowering the embedded GHGs in transportation fuel by adopting the Fuel Standard to reduce the quantity of GHGs emitted in the production of transportation fuel, Cal.Code Regs, tit. 17, §§ 95480-90.

The Tailpipe and VMT Standards work on the demand side: they aim to lower the consumption of GHG-generating transportation fuels. The Fuel Standard, *1080by contrast, is directed at the supply side, creating an alternate path to emissions reduction by reducing the carbon intensity1 of transportation fuels that are burned in California.

B

On January 18, 2007, the California governor issued Executive Order S-01-07, which directed CARB to adopt regulations that would reduce the average GHG emissions attributable to California’s fuel market by ten percent by 2020. The Fuel Standard, developed in response, applies to nearly all transportation fuels currently consumed in California and any fuels developed in the future. Id. § 95480.1(a). In 2010, regulated parties were required to meet the Fuel Standard’s reporting requirements but were not bound by a carbon intensity cap. Id. § 95482(a).2 Beginning in 2011, the Fuel Standard established a declining annual cap on the average carbon intensity of California’s transportation-fuel market. Id. § 95482(b). By setting a predictable path for emissions reduction, the Fuel Standard is intended to spur the development and production of low-carbon fuels, reducing overall emissions from transportation.

To comply with the Fuel Standard, a fuel blender must keep the average carbon intensity of its total volume of fuel below the Fuel Standard’s annual limit. Id. § 95482(a). Fuels generate credits or deficits, depending on whether their carbon intensity is higher or lower than the annual cap. Id. § 95485(a). Credits may be used to offset deficits, may be sold to other blenders, or may be carried forward to comply with the carbon intensity cap in later years. Id. § 95485. With these offsets, a blender selling high carbon intensity fuels can comply with the Fuel Standard by purchasing credits from other regulated parties; no regulated party is required to sell any particular fuel or blend of fuels with a certain carbon intensity or origin. To build a durable and effective marketplace to stimulate the development of alternative fuels, the Fuel Standard created a market for trading, banking, and borrowing Fuel Standard credits. Id.; see also ISOR ES-1. CARB expects that the demand for credits will encourage producers, wherever they are located, to develop fuels with lower carbon intensities for use within the California market.

i

The Fuel Standard uses a “lifecycle analysis” to determine the total carbon intensity of a given transportation fuel. Because GHGs mix in the atmosphere, all emissions related to transportation fuels used in California pose the same local risk to California citizens. “ ‘That these climate change risks are widely-shared does *1081not minimize [California’s] interest’ in reducing them.” Rocky Mountain Ethanol, 843 F.Supp.2d at 1093 (quoting Massachusetts v. ERA 549 U.S. 497, 522, 127 S.Ct. 1438, 167 L.Ed.2d 248 (2007)) (alteration in original) (internal quotation marks omitted). One ton of carbon dioxide emitted when fuel is produced in Iowa or Brazil harms Californians as much as one emitted when fuel is consumed in Sacramento. The Tailpipe Standards control only emissions within California. Without lifecycle analysis, all GHGs emitted before the fuel enters a vehicle’s gas tank would be excluded from California’s regulation. Similarly, the climate-change benefits of biofu-els such as ethanol, which mostly come before combustion, would be ignored if CARB’s regulatory focus were limited to emissions produced when fuels are consumed in California.

With a one-sided focus on consumption, even strong tailpipe-emissions standards would let GHG emissions rise during fuel production. Tailpipe standards could sharply reduce emissions from each individual vehicle without reducing net GHG emissions. In the extreme, rising emissions from production could raise total GHG emissions, completely subverting tailpipe-emissions limits. As an example, CARB analyzed the carbon intensity of ethanol produced in the Midwest using coal for electricity and heat. That method of production yields a carbon intensity more than twenty-percent higher than gasoline. See CaLCode Regs. tit. 17, § 95486(b)(1), tbl. 6 (“Table 6”). No tailpipe standard could capture that difference. If the ethanol were credited for the carbon dioxide absorbed during cultivation of the corn feedstock, it would look superi- or to gasoline from a GHG perspective at the tailpipe. But any shift from gasoline to that form of ethanol would increase net GHG emissions and subject California to greater risk.

To avoid these perverse shifts, CARB designed the Fuel Standard to account for emissions associated with all aspects of the production, refining, and transportation of a fuel, with the aim of reducing total, well-to-wheel GHG emissions. See id. § 95481(a)(38). When these emissions are measured, CARB assigns a cumulative carbon intensity value to an individual fuel lifecycle, which is called a “pathway.” Id. § 95481(a)(14).

The importance of lifecycle analysis is shown clearly by the diversity of the California fuel market, which includes fuels made with many different source materials, called “feedstocks,” and production processes. As of June 2011, CARB has performed lifecycle analyses of fuels made from petroleum, natural gas, hydrogen, electricity, corn, sugarcane, used cooking oil, and tallow. Id. § 95486(b)(1). Fuels made from these feedstocks generate or avoid emissions at different stages of their production, transportation, and use, depending on when the conversion to fuel requires or displaces energy. An accurate comparison is possible only when it is based on the entire lifecycle emissions of each fuel pathway.

Recognizing the need for a reliable method to compare the lifecycle emissions of diverse fuels, the Argonne National Laboratory developed the Greenhouse Gases, Regulated Emissions, and Energy Use in Transportation Model (“GREET”).3 GREET, first published in 1996 and revised and peer reviewed several times since, incorporates comprehensive data on the lifecycle emissions of various fuels. The Environmental Protection Agency *1082(“EPA”) uses GREET for lifecycle analysis in the RFS, which mandates the use of low-carbon-intensity biofuels in the United States fuel supply. See 78 Fed.Reg. 14190, 14209 (Mar. 5, 2013). State agencies in Oregon, Minnesota, and New York have also used GREET to estimate emissions from the production of alternative fuels. In designing the Fuel Standard, CARB used GREET as the basis for its lifecycle-emissions model for fuels used in California. That peer-reviewed model, called CA-GREET, incorporates detailed information about local conditions, including California’s stringent environmental regulations and low-carbon electricity supply-

To provide a baseline against which to compare future reductions, CARB measured the average carbon intensity of the 2010 gasoline market at 95.86 grams of carbon-dioxide equivalent per mega joule (“gC02e/MJ”) of energy. Cal.Code Regs, tit. 17, § 95486(b). In 2011, the carbon intensity cap was set 0.25% below the 2010 average. Id. § 95482. From 2011 to 2020, each annual limit will be a further reduction from that baseline. Id. § 95482(b). After reviewing ethanol sales in different markets during 2011, the Oil Price Information Service reported that fuels with lower carbon intensities received a price premium in California. So this program is starting to work as intended.

The Fuel Standard gives regulated parties two methods to comply with its reporting requirements. First, CARB issued a schedule of “default pathways” for a range of fuels that it anticipated would appear in the California market. These default pathways provided average values for the CA-GREET factors for these anticipated fuels. The resulting default pathways for ethanol appear in Table 6, which we attach as Appendix One. Under Method 1, regulated parties who sell fuel under a default pathway may rely on that pathway in reporting the carbon intensity of the conforming fuel. Id. § 95486(b).

Second, the Fuel Standard allows regulated parties to register individualized pathways using Method 2A or 2B. Id. § 95486(c), (d). Under Method 2A, a regulated party relies in part on a default pathway but proposes a replacement for one or more of the pathway’s average values. Id. § 95486(c). Under Method 2B, a regulated party proposes a new, individualized pathway. Id. § 95486(d). To qualify for Method 2A, the proposed pathway must have a carbon intensity at least 5 gC02e/MJ less than the default pathway it seeks to replace, and it must be expected to supply more than 10 million gasoline-equivalent gallons per year in California. Id. § 95486(e)(2). There is no such threshold for Method 2B. Id. § 95486(e). Once CARB approves a Method 2A or 2B pathway, the pathway remains available for use without further documentation unless there is a material change. Id. § 95484(c)(2)(D). Thus fuel producers can take advantage of default and individualized carbon intensity values, and choose what is most advantageous.

ii

Ethanol is an alcohol produced through fermentation and distillation of a variety of organic feedstocks. Most domestic ethanol comes from corn. Brazilian sugarcane dominates the import market. See 75 Fed. Reg. 14670, 14743, 14746-47 (Mar. 26, 2010). Ethanol production is a resource-intensive process, requiring electricity and steam. Id. at 14745. Steam is usually produced on site with coal or natural gas in dedicated boilers. Id. The choices of type of feedstock, source of electricity, and source of thermal energy affect the carbon intensity of the fuel pathway. To illustrate, ethanol made with sugarcane, hydroelectricity, and natural gas would produce *1083lower emissions than ethanol made from corn and coal. Id. To determine the total carbon intensity values for each ethanol pathway, the CA-GREET model considers the carbon intensity of factors including: (1) growth and transportation of the feedstock, with a credit for the GHGs absorbed during photosynthesis; (2) efficiency of production; (3) type of electricity used to power the plant; (4) fuel used for thermal energy; (5) milling process used; (6) offsetting value of an animal-feed co-product called distillers’ grains, that displaces demand for feed that would generate its own emissions in production; (7) transportation of the fuel to the blender in California; and (8) conversion of land to agricultural use.

On Table 6, CARB separates these factors into those that are correlated with location and those that are not, using a regional identifier as a shorthand for the factors correlated with location. The milling process, co-product, and source of thermal energy are not correlated with region, so they are labeled individually. Factors related to transportation, efficiency, and electricity are correlated with a plant’s location in the Midwest, Brazil, or California. For example, California ethanol plants are newer and more efficient on average than those in the Midwest, using less thermal energy and electricity in the production process. Also, the electricity available on the grid in the Midwest produces more emissions in generation than electricity in California or Brazil because much of the electricity in the Midwest is generated by coal-fired power plants. By contrast, California receives most of its power from renewable sources and natural gas, and Brazil relies almost entirely on hydroelectricity.4

Emissions from transporting the feedstock and the refined fuel are related to location, but they are not directly proportionate to distance traveled. Transportation emissions reflect a combination of: (1) distance traveled, including distance traveled inside California to the fuel blender; (2) total mass and volume transported; and (3) efficiency of the method of transport. California ethanol produces the most transportation emissions because California grows no corn for ethanol, so its producers import raw corn, which is bulkier and heavier than the refined ethanol shipped by producers in Brazil and the Midwest. Brazilian ethanol produces fewer emissions than the 7,500 miles it travels would suggest because ocean tankers are very efficient.5 Midwest ethanol, going one third of that distance, produces the least.6 As a result, total transportation emissions for California ethanol are 8.1 gC02e/MJ, compared to 5.5 for Brazil and 4.8 for the Midwest. Brazilian GREET Pathways 6. This advantage in transportation is reflected in the location of ethanol plants, which are mainly located in the Midwest near sources of corn. 75 Fed. *1084Reg. at 14745. California producers gain a larger credit for distillers’ grains because those grains are consumed in California, so they do not travel as far from the plant to the point of consumption.

We attach two excerpts from Table 6 as appendices. Appendix One reproduces the ethanol pathways from the Midwest, California, and Brazil in Table 6. Appendix Two breaks out two default corn ethanol pathways from Table 6, individually showing each of the regionally correlated factors that determine the carbon intensity values of those pathways. The ethanol pathways detailed in Appendix Two both use a dry-mill production process with natural gas as a heat source and produce dry distillers’ grains as a co-product. As shown in these tables, California’s combination of more efficient plants and greater access to low-carbon electricity outweighs Midwest ethanol’s lower transportation emissions, leaving California ethanol with a 7.2 gC02e/MJ lower carbon intensity for the factors correlated with region. California ethanol producers import their corn from the Midwest, so the two regions have identical carbon intensity assessments for land-use changes. Those factors, combined with the feedstock, milling method, treatments of distillers’ grains, and heat source, determine the carbon intensity of each default pathway.

Producers from all three regions have obtained individualized pathways under Methods 2A or 2B. Cal.Code Regs. tit. 17, § 95486(b). Most of the Midwest ethanol producers who have done so either co-generate heat and electricity or use a renewable source for thermal energy, either of which can dramatically reduce GHG emissions. Cf. 75 Fed.Reg. at 14745. As of mid-2011, CARB had approved ethanol pathways with carbon intensities ranging from 56.56 to 120.99 gC02e/MJ. The individualized pathway with the lowest carbon intensity was achieved by a Midwest producer through Method 2A. The default pathway with the lowest carbon intensity is only slightly higher: 58.40 gC02e/MJ for Brazilian sugarcane ethanol made with electricity generated on site. The highest carbon intensity, 120.99 gC02e/MJ, is for Midwestern wet-mill ethanol, using 100% coal for thermal energy. That is significantly higher than the 95.86 gC02e/MJ average carbon intensity of gasoline in 2010.

iii

The Fuel Standard also regulates crude oil and derivatives sold in California. Like the ethanol provisions, the 2011 Provisions required compliance with carbon intensity caps starting in January 1, 2011. Cal. Code Regs. tit. 17, § 95482(a). The 2011 Provisions remained in effect until December 31, 2011, when they were replaced by amended regulations. The 2011 Provisions are the subject of American Fuels’s challenge and the district court’s decision, so we do not discuss the amended provisions in detail.

Crude oil presents different climate challenges from ethanol and other biofuels. Corn and sugarcane absorb carbon dioxide as they grow, offsetting emissions released when ethanol is burned. By contrast, the carbon in crude oil makes a one-way trip from the Earth’s crust to the atmosphere. For crude oil and its derivatives, emissions from combustion are largely fixed, but emissions from production vary significantly. As older, easily accessible sources of crude are exhausted, they are replaced by newer sources that require more energy to extract and refine, yielding a higher carbon intensity than conventional crude oil. As extraction becomes more difficult, emissions from crude oil will only increase, but CARB expects that fuels with carbon intensity values fifty to eighty percent lower than gasoline will be needed to meet its *1085emissions-reduction targets. No matter how efficiently crude oil is extracted and refined, it cannot supply this level of reduction. To meet California’s ambitious goals, the development and use of alternative fuels must be encouraged.

With that in mind, CARB designed the 2011 Provisions to promote the development of alternative fuels rather than to encourage marginal emissions reductions from crude oil. Under the 2011 Provisions, no crude oil could be assessed a carbon intensity below the market average, but newer sources causing higher emissions were assessed at their individual carbon intensity. By design, this system required regulated parties to meet the Fuel Standard’s carbon-intensity-reduction targets by supplying alternative fuels or buying credits from the sellers of alternative fuels. This was intended to direct investment into low-carbon alternative fuels rather than into the most efficient sources of crude oil, which would still lag behind improvements from alternative fuels that decrease the harmful emissions of carbon dioxide and other GHGs. By distinguishing between existing and emerging sources, CARB also hoped to prevent the mere shift of high carbon intensity crude oils to other markets. This process, known as “fuel shuffling,” would reduce the carbon intensity of the California market by altering the world-wide distribution of fuels, but it would neither promote alternative-fuel development nor reduce net global GHG emissions.

The 2011 Provisions categorized crude oil in two ways: (1) as “existing” or “emerging” crude sources; and (2) as high-earbon-intensity crude oil (“HCICO”) or non-HCICO. “Existing” sources were those that made up at least two percent of California’s crude-oil market in 2006. All others were “emerging” sources. HCI-COs were sources that produced more than 15 gC02e/MJ of emissions in extraction, production, and transportation. All existing sources were assessed the average carbon intensity value of the 2006 California market—8.07 gC02e/MJ—regardless of their individual value. Emerging non-HCICOs were also assessed that average value no matter how low their actual carbon intensity values. Emerging HCICOs were assessed their individual values. This system of categories is illustrated in the table below:

Existing Emerging

Non-HCI- 2006 Average 2006 Average CO

HCICO (8.07) Individual Carbon Intensity

In the benchmark year of 2006, California produced 38.7% of the oil it consumed. That 38.7% consisted of 6.10% oil recovered through gas-injection (“Gas Injection”), 1.3% oil recovered through water-flood methods (“Water Flood”), 16.5% light crude (“California Primary”), and 14.8% oil extracted using thermal-enhanced oil-recovery techniques (“California TEOR”). At 14.8% California TEOR was the only HCICO that made up more than two percent of the 2006 market. It had an individual carbon intensity of 18.89 gC02e/MJ, but as an existing source, it was assessed the market-average carbon intensity of 8.07 gC02e/MJ during 2011. Light crude from Alaska and abroad supplied most of the balance, but Venezuela heavy crude (“Venezuela Heavy”), which has a carbon intensity higher than California TEOR, filled 0.63% of the 2006 market.

In October 2011, CARB concluded that regulating crude oil by reference to the *10862006 market was infeasible and issued new provisions. The new provisions pursued the same goals with similar logic, but they eliminated the categories in the 2011 Provisions. Under the new system, all crude oil is assessed the same carbon intensity value, either the average of the California market in the year of sale or the average from 2010, whichever is higher. These amended provisions took effect on January 1, 2012.

On July 24, 2013, CARB issued a regulatory advisory that altered the treatment of 2011 sales of crude oil that had not yet been subject to lifecycle analysis (“Potential HCICOs”).7 Low Carbon Fuel Standard Regulatory Advisory 13-01, available at http://www.arb.ca.gov/fuelsAcfs/072413 lcfs-rep-adv.pdf. CARB had previously stated that credits related to those sales would be adjusted once lifecycle analysis was performed. See Low Carbon Fuel Standard Regulatory Advisory 10-04A, at 2-4 (June 22, 2011), available at http:// www.arb.ca.gov/fuelsAcfs/0701111efs-rep-adv.pdf. With Advisory 13-01, CARB instead told regulated parties that retroactive adjustment of credit balances would not be required. For sales during 2011, Potential HCICOs would be treated like non-HCICOs and assigned the average carbon intensity of the California market, essentially applying the amended provisions to Potential HCICOs one year earlier than planned. Advisory 13-01, at 2-3.

C

In December 2009, Rocky Mountain filed a complaint challenging the ethanol provisions of the Fuel Standard, alleging that they violated the dormant Commerce Clause and were preempted by the RFS. In February 2010, American Fuels challenged both the ethanol and the crude-oil provisions on similar grounds. Rocky Mountain sought a preliminary injunction on its Commerce Clause and preemption claims. American Fuels moved for summary judgment on its Commerce Clause claims. CARB filed cross-motions for summary judgment on all grounds.

On December 29, 2011, the district court granted Rocky Mountain’s request for a preliminary injunction and American Fuels’s partial motion for summary judgment, concluding that the Fuel Standard violated the dormant Commerce Clause by (1) engaging in extraterritorial regulation, (2) facially discriminating against out-of-state ethanol, and (3) discriminating against out-of-state crude oil in purpose and effect. The district court then determined that CARB did not show that the Fuel Standard could survive strict scrutiny.

The district court granted partial summary judgment in favor of CARB on its cross-motion, concluding that the Fuel Standard is a control or prohibition respecting a characteristic or component of a fuel under section 211(c)(4)(B) of the Clean Air Act, but it denied summary judgment on whether that section prevents scrutiny of the Fuel Standard under the Commerce Clause. CARB timely appealed. We stayed the district court’s judgments pending this appeal.

II

We review de novo a district court’s rulings on cross-motions for summary judgment. CRM Collateral II, Inc. v. Tricounty Metro. Transp. Dist. of Or., 669 F.3d 963, 968 (9th Cir.2012). A grant of summary judgment is appropriate *1087where “the movant shows that there is no genuine dispute as to any material fact and the movant is entitled to judgment as a matter of law.” Fed.R.Civ.P. 56(a). A district court’s resolution of federal constitutional claims is also reviewed de novo. Black Star Farms LLC v. Oliver, 600 F.3d 1225, 1229 (9th Cir.2010).

We review an order granting a preliminary injunction for abuse of discretion. Stormans Inc. v. Selecky, 586 F.3d 1109, 1119 (9th Cir.2009) (citation omitted). We will reverse if the order was based on clearly erroneous findings of fact or on an erroneous legal standard. Id.

Ill

Plaintiffs contend that the Fuel Standard’s ethanol and crude-oil provisions discriminate against out-of-state commerce and regulate extraterritorial activity. CARB disagrees and, in the alternative, contends that Section 211(c)(4)(B) of the Clean Air Act authorizes the Fuel Standard under the Commerce Clause. We address each claim in turn.

The Commerce Clause provides that “Congress shall have Power ... [t]o regulate Commerce ... among the several States.” U.S. Const., art. I, § 8, cl. 3. This affirmative grant of power does not explicitly control the several states, but it “has long been understood to have a ‘negative’ aspect that denies the States the power unjustifiably to discriminate against or burden the interstate flow of articles of commerce.” Or. Waste Sys., Inc. v. Dep’t of Envtl. Quality of State of Or., 511 U.S. 93, 98, 114 S.Ct. 1345, 128 L.Ed.2d 13 (1994) (citing Wyoming v. Oklahoma, 502 U.S. 437, 454, 112 S.Ct. 789, 117 L.Ed.2d 1 (1992)). Known as the “negative” or “dormant” Commerce Clause, this aspect is not a perfect negative, as “the Framers’ distrust of economic Balkanization was limited by their federalism favoring a degree of local autonomy.” Dep’t of Revenue of Ky. v. Davis, 553 U.S. 328, 338, 128 S.Ct. 1801, 170 L.Ed.2d 685 (2008) (citations omitted). Within the federal system, a “courageous state may, if its citizens choose, serve as a laboratory; and try novel social and economic experiments without risk to the rest of the country.” New State Ice Co. v. Liebmann, 285 U.S. 262, 311, 52 S.Ct. 371, 76 L.Ed. 747 (1932) (Brandéis, J., dissenting). If successful, those experiments may often be adopted by other states without Balkanizing the national market or by the federal government without infringing on state power.

“The modern law of what has come to be called the dormant Commerce Clause is driven by concern about ‘economic protectionism—that is, regulatory measures designed to benefit in-state economic interests by burdening out-of-state competitors.’ ” Davis, 553 U.S. at 337-38, 128 S.Ct. 1801 (quoting New Energy Co. of Ind. v. Limbach, 486 U.S. 269, 273-74, 108 S.Ct. 1803, 100 L.Ed.2d 302 (1988)). For dormant Commerce Clause purposes, economic protectionism, or discrimination, “simply means differential treatment of instate and out-of-state economic interests that benefits the former and burdens the latter.” Or. Waste Sys., Inc., 511 U.S. at 99, 114 S.Ct. 1345. “[0]f course, any notion of discrimination assumes a comparison of substantially similar entities.” Gen. Motors Corp. v. Tracy, 519 U.S. 278, 298, 117 S.Ct. 811, 136 L.Ed.2d 761 (1997). If a statute discriminates against out-of-state entities on its face, in its purpose, or in its practical effect, it is unconstitutional unless it “serves a legitimate local purpose, and this purpose could not be served as well by available nondiscriminatory means.” Maine v. Taylor, 477 U.S. 131, 138, 106 S.Ct. 2440, 91 L.Ed.2d 110 (1986) (internal quotation marks omitted). Absent discrimination, we will uphold the law “unless the burden imposed on [interstate] com*1088merce is clearly excessive in relation to the putative local benefits.” Pike, 397 U.S. at 142, 90 S.Ct. 844.

A

The district court concluded that the Fuel Standard facially discriminated against out-of-state corn ethanol by (1) differentiating between ethanol pathways based on origin and (2) discriminating against out-of-state ethanol based on factors within the CA-GREET formula that were “inextricably intertwined with origin.” Rocky Mountain Ethanol, 843 F.Supp.2d at 1087.

i

Before we consider whether the Fuel Standard discriminates against out-of-state ethanol, we must determine which ethanol pathways are suitable for comparison. Tracy, 519 U.S. at 298,117 S.Ct. 811. Entities are similarly situated for constitutional purposes if their products compete against each other in a single market. Id. at 299, 117 S.Ct. 811. If they do, it is irrelevant whether they are made from different materials or if one poses a substantial competitive threat to another. Bacchus Imports, Ltd. v. Dias, 468 U.S. 263, 268-69,104 S.Ct. 3049, 82 L.Ed.2d 200 (1984).

The district court concluded that all Brazilian ethanol pathways and all CA-GREET factors correlated with origin were outside the bounds of comparison. The district court explained, “Because the [Fuel Standard] makes production process, feedstock and origin relevant, comparing pathways with different production p