The corporate average fuel economy (CAFE) program is the oldest and most misunderstood efficiency regime in American transportation law. It requires each automaker to meet a fleet-wide average measured in miles per gallon, but the operative policy lives in the arithmetic rather than the headline number. The statute sets the duty; the formula decides how the duty is satisfied. Corporate average fuel economy requirements live at 49 U.S.C. sections 32901 through 32919, Chapter 329 of Title 49, a title of the United States Code that originated in the energy legislation of 1975 and was substantially rewritten by the Energy Independence and Security Act of 2007. As of November 2016, that chapter governed every passenger car and light truck sold in the United States, and it did so through three devices that the rest of this article unpacks: a sales-weighted harmonic mean, separate fleets for cars and trucks, and, since 2007, targets that rise and fall with each vehicle’s footprint.

The program did not begin as environmental policy. It began as a response to an embargo. The Arab oil embargo of 1973 and 1974 tripled crude prices and exposed how thoroughly the American driving fleet depended on imported petroleum. New-car fleet economy had fallen from 14.8 miles per gallon in model year 1967 to 12.9 miles per gallon in 1974, and Congress concluded that the market alone would not rebuild the margin of safety. The answer became the Energy Policy and Conservation Act of 1975, a broad energy-security statute whose purposes clause, codified at 42 U.S.C. section 6201, aimed at conservation and reduced dependence on imported oil. Fuel economy entered federal law as energy policy, and the environmental meaning of the program arrived decades later, layered on top by a second regulator acting under a different statute. Congress wrote the first decade’s ambition directly into the statute: passenger-car targets of 18.0 miles per gallon for model year 1978, 19.0 for 1979, and 20.0 for 1980, a delegation to the Secretary of Transportation for the interim years, and a long-run mark of 27.5 for model year 1985 and thereafter.

Why Congress reached for a mandate in 1975

The embargo that produced the program began in October 1973, when Arab members of the Organization of Petroleum Exporting Countries cut off oil shipments to the United States in retaliation for American support of Israel in the Yom Kippur War. Crude prices roughly tripled within months, gasoline lines formed in American cities, and several states imposed odd-even rationing based on license plate numbers. The shortage was not only a price shock but a physical one: drivers waited in lines that wrapped around blocks, and the federal government confronted the possibility that the country’s transportation system, the foundation of its economy, depended on suppliers who had just demonstrated their willingness to use oil as a weapon. The crisis atmosphere of 1973 and 1974 is the indispensable background to everything the fuel-economy statute became.

The numbers behind the panic told a longer story than the embargo alone. American cars had been getting thirstier for years. New-car fleet economy stood at 14.8 miles per gallon in model year 1967 and had slid to 12.9 by 1974, as buyers moved toward larger, heavier, more powerful models. The market was delivering exactly what consumers wanted and exactly what the national interest could not afford. Voluntary conservation appeals had produced little, and the price controls of the era blunted the signal that higher crude prices might otherwise have sent to buyers. Congress concluded that only a binding requirement would move the fleet, and it chose a regulatory mandate rather than a price instrument: the statute would tell manufacturers what average their fleets must achieve, and leave the engineering to them.

The Energy Policy and Conservation Act of 1975 was a broad crisis statute, and the fuel-economy provisions were one instrument among several aimed at the same vulnerability. The act’s purposes clause, codified at 42 U.S.C. section 6201, framed the enterprise as energy conservation and reduced dependence on imported oil, language that matters because it fixes the program’s original meaning. This was not environmental legislation wearing energy clothes. The Congress of 1975 legislated against the embargo, and every structural choice in the fuel-economy chapter, from the fleet-wide averaging to the separate truck track to the domestic-import wall, was designed to squeeze imported barrels out of the transportation sector. The climate framing that later attached to the program was grafted onto a statute whose authors would not have recognized it, and the graft changed how the program was debated without changing what its operative text required.

The choice of a fleet average, rather than a per-vehicle minimum, was itself a policy judgment worth understanding. A per-vehicle floor would have banned the thirstiest models outright, a politically explosive proposition in a country attached to large cars and working trucks. A fleet average let manufacturers keep selling thirsty models as long as they sold enough efficient ones to balance the math, which made the mandate politically survivable and economically flexible at once. The harmonic-mean formula then ensured that the balancing could not be done cheaply with token efficient models. Congress was, in effect, buying the maximum conservation the politics would bear, and the formula was the mechanism that kept the bargain honest.

The fuel-economy title was the act’s demand-side centerpiece. Rather than rationing gasoline or taxing fuel, the statute regulated the efficiency of the new-vehicle fleet, on the theory that the cars sold in 1978 would still be burning gasoline in 1988 and that the fleet turns over slowly enough for a standard to compound. The 18.0 mile-per-gallon figure for model year 1978 was set barely two years after enactment, which shows how much readily available efficiency the 1975 fleet contained: the industry met the early years of the ramp with changes that were available almost immediately. The 27.5 figure for 1985 was the stretch goal, and meeting it required the deeper reengineering of the early 1980s. The statute’s framers understood the fleet-turnover arithmetic that makes fuel-economy regulation slow and powerful at once: slow, because the standard touches only new vehicles and the existing fleet keeps burning; powerful, because every efficient vehicle sold displaces a less efficient one for its entire service life, and the savings accumulate across decades.

The statute wrote only the bookends of the passenger-car ramp into law and delegated the middle to the Secretary of Transportation, who was directed to prescribe the intervening model years under the same maximum-feasible standard. The resulting schedule, announced in June 1977, climbed in steady steps: 22.0 miles per gallon for model year 1981, 24.0 for 1982, 26.0 for 1983, and 27.0 for 1984, reaching the statutory 27.5 mark for 1985 and thereafter. The design tracked the engineering reality of product cycles. The early statutory years captured the gains available almost immediately in the 1975 fleet; the delegated middle years followed the industry’s retooling capacity; the 1985 figure set the stretch horizon. The delegation also supplied the program’s first lesson in administrative discretion, because the middle years’ numbers were chosen by a regulator applying the four statutory factors rather than by Congress writing a table. The Senate report that later reconstructed this history preserves the full eight-year table as the baseline the 2007 rewrite would eventually supersede.

The energy-security framing also explains features of the chapter that puzzle readers who approach it as environmental law. The domestic and import fleet separation makes sense as industrial policy for an energy emergency: Congress wanted the domestic industry’s product line transformed, not merely the national sales average improved by imports. The maximum-feasible criterion’s reference to the need of the United States to conserve energy makes sense as a mobilization standard, not as a cost-benefit test. And the penalty’s per-vehicle structure makes sense as a compliance incentive aimed at production decisions rather than as a pollution price. The chapter is a conservation statute wearing the clothes of a regulatory program, and its environmental co-benefits, real and large, are consequences of its design rather than expressions of its purpose. The Harvard environmental law backgrounder’s formulation captures it: the chapter regulates to provide for improved energy efficiency of motor vehicles, full stop.

How the CAFE average is actually calculated

The compliance obligation is not a per-vehicle minimum. No individual car has to meet the standard. The statute requires that the average of everything a manufacturer sells in a model year meet the number, and the averaging method is specified in the text with unusual precision. Section 32904(a)(1)(B) directs the Environmental Protection Agency’s Administrator to calculate a manufacturer’s average fuel economy by dividing the number of passenger automobiles the manufacturer produced in the model year by the sum of the fractions obtained by dividing, for each model, the number of automobiles of that model by the fuel economy measured for that model. In symbols, the average equals N divided by the sum over models of n-sub-i over mpg-sub-i, where N is the total count of vehicles and n-sub-i is the count of each model. That is a sales-weighted harmonic mean: sales-weighted because each model’s contribution scales with its sales volume, harmonic because the averaging happens over the reciprocals, gallons per mile, rather than over miles per gallon directly.

What does the harmonic-mean formula actually do?

It totals each model’s fuel consumption rate, in gallons per mile, weighted by sales volume, then inverts the sum. A thirsty model therefore drags the average down more than its sales share would suggest under a simple average, which is why the formula rewards eliminating the worst models more than improving the best ones.

The worked intuition is worth carrying through an example. Suppose a maker sells two models in equal numbers, one rated at 20 miles per gallon and the other at 40. A simple arithmetic mean would report 30. The statutory formula instead converts each to consumption: the 20-mpg model burns 0.05 gallons per mile, the 40-mpg model burns 0.025. The average consumption across the two equal-volume models is 0.0375 gallons per mile, and inverting gives 26.67 miles per gallon, noticeably below the simple mean. Now skew the mix toward the thirsty model, with three 20-mpg cars for every 40-mpg car, and the average falls to 22.86. Skew it the other way, three efficient cars for every thirsty one, and it rises to 32. The formula is doing exactly what its drafters intended: it measures the fleet’s actual fuel appetite, in gallons, rather than a number that flatters the lineup. A maker cannot offset a blockbuster gas-guzzler with a handful of halo hybrids, because the harmonic mean punishes concentration at the thirsty end.

That punishment has a mirror image in the program’s treatment of the best models. Improving a 40-mpg model to 50 mpg saves 0.005 gallons per mile per car. Improving a 15-mpg model to 20 saves 0.0167 gallons per mile per car, more than three times as much. The statute therefore concentrates each manufacturer’s compliance incentive where the gallons actually are, at the inefficient end of the lineup. This is the arithmetic answer to the question of why CAFE pushed the industry toward downsizing engines and shedding weight in its first decade: the formula made the thirstiest models the most expensive to keep.

The formula repays a second worked example, this time with three models, because real lineups are never as tidy as two-model illustrations. Suppose a manufacturer sells 100,000 cars at 15 miles per gallon, 200,000 at 25, and 100,000 at 40. The simple average of the three ratings is 26.67. The statutory calculation instead sums the consumption fractions: 100,000 divided by 15, plus 200,000 divided by 25, plus 100,000 divided by 40, which equals 6,666.67 plus 8,000 plus 2,500, or 17,166.67 gallon-miles of consumption weight. Dividing the 400,000 total vehicles by that sum yields 23.3 miles per gallon. The thirstiest tenth of the lineup, the 100,000 cars at 15, contributes nearly 39 percent of the consumption sum despite being only a quarter of sales. That disproportion is the point: the formula forces each manufacturer to confront its worst models first.

The compliance strategy the formula rewards follows directly. Eliminating or dramatically improving the thirstiest model in a lineup buys more compliance than incremental gains across efficient models, because gallons saved per vehicle are largest where consumption per mile is highest. A manufacturer choosing between a costly redesign of its best-selling efficient sedan and a moderate improvement to its thirsty full-size offering will find the latter buys more average, per dollar, in almost every case. The statute thus directs engineering investment toward the vehicles the energy-security purpose most needs improved, which is precisely what a gallons-based measure should do. A miles-per-gallon average would have flattered the lineup; the harmonic mean measures the fuel.

One more structural detail of the calculation matters. The EPA Administrator performs the measurement, not the transportation regulator that sets the standards. The division of labor dates to the 1975 design: the transportation safety regulator writes the numbers, and the environmental agency tests the cars and runs the math. The fuel economy figure for each model comes from the test procedures the EPA administers, and the manufacturer’s average is computed from those figures against the transportation regulator’s standard. The two agencies were therefore entangled from the beginning, long before the second regulatory regime arrived.

What the rated number measures, and what it does not

Every compliance calculation in the program rests on the rated fuel-economy figure for each model, and the rated figure comes from test procedures the EPA administers. The procedures drive vehicles through defined speed and load patterns on a dynamometer, measure the fuel consumed, and convert the result into the miles-per-gallon rating that enters the harmonic-mean formula. The rating is therefore a legal construct as much as a physical measurement: it is the number the tests produce under the procedures in effect, and manufacturers engineer to the tests because the tests are what count. Changes to the procedures change the ratings without changing the vehicles, which is why test-procedure rulemakings are among the most quietly consequential actions in the program’s history.

The well-known gap between rated mileage and the mileage drivers experience does not undermine the program’s logic, though it does limit what the program can claim. The standards are comparative instruments: they require each manufacturer’s fleet to beat a target computed the same way, from the same tests, so the playing field is level even if the absolute numbers flatter every vehicle equally. What the gap does affect is the translation from the program’s numbers to real-world fuel savings and emissions reductions, the quantities that matter for energy security and air quality. Analysts who estimate the program’s benefits apply adjustment factors to bridge the test-to-road gap, and the size of the adjustment is itself a subject of technical debate. The honest way to read a CAFE number is as a regulatory index, precise and comparable within its own system, rather than as a prediction of what any particular driver will see at the pump.

The window sticker that consumers see is derived from the same laboratory data but processed through a different methodology, one that adjusts the raw test results downward to approximate real-world driving with its air conditioning, cold starts, aggressive acceleration, and high speeds. The two numbers therefore differ by design: the sticker is the consumer-information figure, and the laboratory value is the compliance figure. Confusion between them generates a durable stream of misunderstanding, including the belief that manufacturers are somehow failing to meet standards that the sticker suggests they exceed, or exceeding standards the sticker suggests they miss. The statute cares only about the laboratory value. Section 32904(a)(1)(B) says the Administrator calculates the average from the fuel economy measured for each model, and the measurement protocol is the laboratory test.

Why the program keeps cars and trucks in separate fleets

The statute does not apply one average to everything a manufacturer sells. It applies the car standard to passenger automobiles and a separate, historically lower standard to non-passenger automobiles, the statutory term for what the regulations call light trucks. Section 32901(a)(18) defines a passenger automobile as an automobile the Secretary decides by regulation is manufactured primarily for transporting not more than ten individuals, excluding an off-highway-capable automobile with a significant off-highway feature other than four-wheel drive that is either four-wheel drive or rated above 6,000 pounds gross vehicle weight. Section 32901(a)(17) then defines a non-passenger automobile as an automobile that is neither a passenger automobile nor a work truck. The definitions look technical, but they are the doorway through which entire product lines pass into the easier fleet.

The doorway’s exact shape was drawn by regulation, not by the statute. The NHTSA rule at 49 CFR 523.5 classifies which vehicles count as non-passenger automobiles, and large sport utility vehicles pass through it principally on two prongs. The first prong covers vehicles designed for a truck-like function: carrying more than ten people, providing living quarters, carrying property on an open bed, offering more cargo than passenger volume, or, for model year 2008 and later, providing three rows of standard seating with a fold-flat load floor. The second prong covers off-highway capability: four-wheel drive or a rating above 6,000 pounds gross vehicle weight, plus at least four of five geometry characteristics, an approach angle of at least 28 degrees, a breakover angle of at least 14 degrees, a departure angle of at least 20 degrees, a running clearance of at least 20 centimeters, and front and rear axle clearances of at least 18 centimeters each. Most large SUVs enter the light-truck class through the off-highway prong or the three-row fold-flat-seat prong, and that classification, rather than anything in the statute itself, places them under the less demanding non-passenger standard. The regulation is where the program’s most debated boundary was actually drawn.

The definitions at section 32901 repay close reading because they nest inside one another like boxes, and the nesting determines which standard applies to every vehicle sold. The outermost box is “automobile” at section 32901(a)(3), as amended by section 103 of the 2007 act: a four-wheeled vehicle propelled by fuel or alternative fuel, manufactured primarily for use on public streets, roads, and highways, and rated at not more than 10,000 pounds gross vehicle weight. That weight cap is the program’s outer boundary; above it sit the medium- and heavy-duty vehicles that the 2007 law addressed separately. Inside the automobile box sit three mutually exclusive categories. The “passenger automobile” of section 32901(a)(18) is an automobile the Secretary decides by regulation is manufactured primarily for transporting not more than ten individuals, minus the off-highway-capable vehicles described earlier. The “non-passenger automobile” of section 32901(a)(17) is whatever is left after removing passenger automobiles and work trucks. And the “work truck” of section 32901(a)(19), added in 2007, is a vehicle rated between 8,500 and 10,000 pounds gross vehicle weight that is not a medium-duty passenger vehicle.

The structure means that classification is everything and that the statute deliberately left the hardest classification calls to regulation. The ten-individual test for passenger automobiles sounds crisp until one asks about vehicles that straddle passenger and cargo uses, which is why the Secretary’s regulatory judgment does the real work. The exclusion for off-highway-capable vehicles, with its 6,000-pound alternative and its significant-feature test, is where the sport utility vehicle enters the statute’s field of vision: a vehicle built to carry families can still escape the passenger-automobile definition if it is heavy enough or capable enough off pavement. The work-truck definition’s carve-out for medium-duty passenger vehicles, in turn, keeps large passenger vans and SUVs from migrating into the newest category. Every boundary in the section is a policy decision about which competitive segment bears which burden, and the decades of argument about the program’s fairness have mostly been arguments about these boxes.

The reason Congress created the boundary in the first place was workload, not favoritism, or at least not favoritism as the statute’s defenders told it. Trucks in 1975 meant work vehicles: pickups, vans, and utility vehicles that needed torque and payload capacity the technology of the day could not deliver at car-like efficiency. A single standard would have forced either the abandonment of working trucks or a car standard set by truck capability. The separate track let each fleet improve on its own curve. What Congress did not foresee, and what the regulation could not prevent, was the migration of family transportation into the truck class. As minivans and then SUVs absorbed the station-wagon market, the easier fleet absorbed the volume, and the program’s overall stringency became a function of a classification decision as much as of the numbers.

The light-truck track began modestly. The first truck standards were set for model year 1979 at 17.2 miles per gallon, a gentler curve reflecting the era’s assumption that trucks were work vehicles with payload duties the technology of the day could not square with car-like efficiency. The truck numbers stayed well below the car numbers through the 1980s, which was the original compromise: cars would do the heavy conservation lifting while trucks improved on a separate, slower schedule. The compromise made sense when trucks were a minority of sales. As the truck share of the market grew through the 1990s, the separate track’s leniency became the program’s central structural fact, and the 2006 reform rule’s decision to bring footprint-based targets to the truck fleet first was an attempt to make the easier fleet pull its weight.

Why do large SUVs count as trucks under the CAFE rules?

Because the classification regulation at 49 CFR 523.5 sorts vehicles by function and capability rather than by how families use them. An SUV with four-wheel drive and the ground clearance, or with three rows of fold-flat seating, meets the definition of a non-passenger automobile, placing it in the light-truck fleet under that fleet’s lower target.

Inside the car fleet there is a second wall that most summaries of the program omit. Section 32904(b)(1)(A) provides that a passenger automobile counts as manufactured domestically in a model year if at least 75 percent of the cost to the manufacturer is attributable to value added in the United States or Canada, Canada having been added to the original United States-only test by the 1993 NAFTA Implementation Act. Section 32904(b)(2)(A) then requires the EPA Administrator to make separate calculations for domestically manufactured and non-domestically manufactured passenger automobiles, treating the two groups as if they were made by separate manufacturers. A maker cannot average its imported small cars against its domestically built large ones; the domestic car fleet must comply on its own. The purpose, as the Congressional Research Service’s standard history of the program explains, was to keep manufacturers from meeting the car standard by importing small cars while continuing to build large cars at home, a design meant to protect the diversity of models manufactured in the United States. The provision’s labor dimension is well documented in the later record: the United Auto Workers consistently defended the fleet separation on domestic-jobs grounds, including in a 1994 petition proceeding in which the union invoked auto-worker job losses since the 1970s. What the record does not support is the stronger claim that the union wrote the 1975 provision; no primary source from the enactment attributes the text’s insertion to union lobbying, and the honest history credits the design to Congress’s concern about compliance gaming rather than to any single interest group’s authorship. The 2007 amendments added one refinement worth noting: the domestic and imported fleets remain walled off for compliance calculations, with the wall breached only for credit accounting under section 32903, so credits can move across a boundary that the averages themselves cannot cross.

The 75-percent test that separates domestic from imported passenger cars operates on cost to the manufacturer, not on parts counts or assembly location alone. Section 32904(b)(1)(A) asks what share of the cost is attributable to value added in the United States or Canada, and the 75-percent threshold means that a car assembled domestically from mostly imported components can still count as imported, while a car with deep North American supply chains counts as domestic wherever the final assembly occurs. The statute adds a timing wrinkle for Canadian assembly: a car assembled in Canada and imported more than 30 days after the model year’s end does not get domestic treatment, a provision aimed at model-year gaming. Canada’s inclusion in the test came from the 1993 NAFTA Implementation Act; the original 1975 text counted only United States value added. A 2006 House report loosely described the test as covering United States, Canadian, or Mexican value; this article does not resolve the discrepancy and quotes the statute, whose words are “United States or Canada.” The 1994 proceeding in which the transportation regulator denied the United Auto Workers petition, published at 59 FR 64884, remains the fullest administrative treatment of the separation’s labor rationale.

The Energy Independence and Security Act of 2007 also added a third vehicle category, the work truck, defined at section 32901(a)(19) as a vehicle rated between 8,500 and 10,000 pounds gross vehicle weight that is not a medium-duty passenger vehicle. The same act’s amendment to section 32901(a)(3) capped the statutory term “automobile” at 10,000 pounds gross vehicle weight, drawing the outer boundary of the program. Above that line sit the medium- and heavy-duty vehicles that the 2007 law brought under a new standards provision at section 32902(k), a separate regime with its own rulemaking track.

The legislative vehicle for the 2007 rewrite was the Energy Independence and Security Act, Public Law 110-140, whose fuel-economy title Congress named the Ten-in-Ten Fuel Economy Act, a reference to the decade-long climb it mandated. Title I, Subtitle A of the act rewrote section 32902, and the rewrite’s centerpiece was the attribute mandate codified at section 32902(b)(3): the Secretary must prescribe separate average fuel economy standards for passenger and non-passenger automobiles based on one or more vehicle attributes related to fuel economy, expressed as a mathematical function. The provision converted the 2006 administrative pilot into a legislative command and extended it to passenger cars, ending the era of flat fleet-wide numbers. The Senate report accompanying the legislation, S. Rept. 110-278, laid out the full 1975 phase-in schedule that the act was now superseding, a fitting archival bookend: the Congress of 2007 was consciously replacing the Congress of 1975’s chosen mechanism while keeping its energy-security purpose.

The act’s numerical core was the combined-fleet target of at least 35 miles per gallon for model year 2020, set at section 32902(b)(2)(A), with ratable annual increases across model years 2011 through 2020. “Ratable” did the quiet work: the increases had to be distributed across the decade rather than backloaded, preventing the familiar regulatory maneuver of promising stringency in the out years while doing little early. The 35-mpg figure was a statutory floor for the combined fleet, not a prediction about any single manufacturer’s obligation, because under the attribute system each maker’s number would depend on its mix. The act preserved the “maximum feasible” standard-setting criterion of section 32902(a), the longstanding instruction that the Secretary set standards at the maximum feasible level, and it added the program’s newest frontier at section 32902(k): standards for work trucks and medium- and heavy-duty vehicles, the weight classes above the 10,000-pound automobile cap, on a separate rulemaking track with its own feasibility analysis.

The recodified minimum domestic-passenger-car standard at section 32902(b)(4) deserves emphasis as the rewrite’s belt-and-suspenders provision. Under a pure attribute system, a manufacturer could lower its required average indefinitely by enlarging its vehicles, subject only to what buyers would accept. The (b)(4) floor, the greater of 27.5 miles per gallon or 92 percent of the Secretary’s projected combined domestic and import average, guarantees that the domestic car fleet meets a real efficiency level regardless of footprint drift, and the prohibition on satisfying it with traded or transferred credits guarantees that the efficiency is real rather than purchased. The provision is the clearest textual evidence that Congress understood the attribute system’s gaming margin and legislated a backstop against it.

The work-truck and heavy-duty frontier

The 2007 act’s most structurally novel addition was the program’s extension above the 10,000-pound line. Section 32901(a)(19) defined the work truck as a vehicle rated between 8,500 and 10,000 pounds gross vehicle weight that is not a medium-duty passenger vehicle, and section 32902(k) directed the Secretary to prescribe fuel-economy standards for work trucks and for medium- and heavy-duty vehicles more broadly. The provision recognized what the original 1975 design had set aside: the heaviest vehicles on the road burn a disproportionate share of transportation fuel, and a program serious about conservation could not leave them permanently outside its reach. The rulemaking track for these vehicles is separate from the light-duty program, with its own feasibility analysis reflecting the different engineering, duty cycles, and market structure of the heavy-duty world.

The work-truck definition’s boundaries show the same classificatory care as the light-duty definitions. The 8,500-pound floor separates work trucks from the light-truck fleet below, while the 10,000-pound ceiling marks the outer boundary of the statutory term “automobile.” The carve-out for medium-duty passenger vehicles keeps large passenger vans and sport utility vehicles in the light-duty program rather than letting them migrate upward into the newer, less developed regime. As with the 1975 fleet definitions, the lines are drawn where the regulatory burden changes character: below the line, the mature harmonic-mean and footprint machinery; above it, a newer standards program built for vehicles whose fuel use is measured in ton-miles rather than passenger-miles.

The heavy-duty provision also illustrates the chapter’s definitional method at its limit. Where the light-duty program sorts vehicles with the passenger, non-passenger, and work-truck definitions of section 32901, the heavy-duty world confronts vehicles assembled in stages, sold as bare chassis, and completed by body builders, so the very idea of a manufacturer with a fleet average strains at the edges. The statute answered by directing the Secretary to design a suitable program rather than by forcing heavy trucks into the light-duty mold, a rare instance of Congress acknowledging that its formula had met its boundary. The text the One Test reader must master stops at ten thousand pounds, and beyond that line a different program with different physics begins.

The maximum-feasible inquiry: how the numbers get set

Section 32902(a) contains the instruction that governs every standard-setting rulemaking in the program’s history: the Secretary shall prescribe average fuel economy standards at the maximum feasible level. The phrase is doing two jobs at once. “Maximum” pushes toward stringency; “feasible” pulls back toward reality. The statute then lists the factors the Secretary must weigh in finding the balance: technological feasibility, economic practicability, the effect of other motor vehicle standards of the Government on fuel economy, and the need of the United States to conserve energy. Those four considerations are the permanent frame of the program’s central argument, and nearly every dispute about the level of a standard is a dispute about how to weigh them.

Technological feasibility asks what engineering can do within the lead time available. The inquiry is forward-looking: not what the industry has demonstrated in production, but what it can plausibly deploy across its fleets by the model years the standard will cover. When the 1975 Congress wrote the 18.0 to 27.5 ramp, the hardware that would deliver it was either on the shelf or visible on the horizon, and the industry met the schedule. The same pattern repeated at each later increment: manufacturers argued that the next step required breakthroughs, the regulator pointed to hardware already in limited production, and the outcome usually landed between the two. The maximum-feasible inquiry thus functions as a technology-forcing mechanism with a reasonableness backstop: the law demands the best the industry can do, not the best anyone can imagine.

Economic practicability asks what the industry can afford without unreasonable disruption, considering development costs, capital investment cycles, and the effect on vehicle prices and sales. The other-standards factor recognizes that the same vehicles are simultaneously being redesigned for safety and emissions rules, and that engineering resources and vehicle attributes are shared across the regimes. The need to conserve energy is the thumb on the scale for stringency, the statutory reminder that the program exists because Congress judged the market’s outcome insufficient. The Secretary’s task is to maximize subject to all four, and the rulemaking record of every major standard is, at bottom, a few hundred pages of analysis defending a particular balancing.

The balancing is where the politics enter, because the factors point in different directions and their weights are not specified. A Secretary who emphasizes technological feasibility and the need to conserve energy will set a demanding standard; one who emphasizes economic practicability will set a lenient one. The 1986 through 1989 episode, when the regulator lowered the car standard to 26.0 miles per gallon for model years 1986 through 1988 and 26.5 for 1989 before restoring 27.5 in 1990, is the clearest historical example of the balance shifting with circumstances. The statute requires the standards to be set with enough lead time for manufacturers to plan their product cycles, which means each rulemaking is a bet about the future: future fuel prices, future technology costs, future consumer demand. What the maximum-feasible framework has never permitted is stasis without justification; every standard must be defended as the most the law’s four factors allow. The 2007 amendments addressed the timing dimension directly by requiring ratable annual increases for model years 2011 through 2020, so that manufacturers would face a published glide path rather than a series of cliff rulemakings. A maximum-feasible standard that arrives too late to engineer for is not feasible at all, and the statute says so in the form of a schedule.

The 2006 reform rule and the arrival of footprint

For three decades the program worked the way the 1975 design contemplated: flat fleet-wide numbers, revised periodically, with separate car and truck tracks. The flat structure had a known weakness, and the transportation regulator named it in the rulemaking that changed the truck system. A single fleet-wide target gives every manufacturer the same number regardless of what it sells, which means a maker of small cars complies easily while a maker of large trucks strains, and the most direct compliance path is to shift the mix toward smaller vehicles. The regulator’s April 2006 reform rule, published at 71 FR 17650, addressed the truck fleet first. For model years 2008 through 2010 it offered manufacturers a choice between the traditional flat standard and a new reformed track, then made the reformed track mandatory for model year 2011. The reformed track expressed each manufacturer’s obligation as a mathematical function of vehicle size, and the size measure it chose was footprint: the product of wheelbase and average track width, divided by 144 to express the result in square feet.

The 2006 rule matters for two reasons. First, it was the administrative pilot for everything that followed: footprint-based compliance was tested on the truck fleet before Congress mandated it for the whole program, which is why the truck footprint curves were more developed than the car curves when the joint standards for model years 2012 through 2016 were written. The regulator had been drawing truck curves for years and was drawing car curves almost from scratch. Second, it established the regulatory logic that size-based targets are fairer across manufacturers with different product mixes, because a maker that sells mostly large vehicles faces a less demanding number than a maker that sells mostly small ones. Whether that logic counts as fairness or as surrender depends on what one thinks the program is for, and the debate that the 2006 rule opened would dominate the next decade of fuel-economy policy.

The footprint definition the rule adopted came from the emissions-testing world: track width in inches multiplied by wheelbase in inches, divided by 144, yielding square feet, the formulation codified at 40 CFR section 86.1803-01. Wheelbase and track width were chosen because they are difficult to game quickly, being fundamental to a vehicle’s architecture, and because they correlate with the utility consumers buy in larger vehicles: interior space and stability. The regulator’s stated rationale was fairness across manufacturers with different product mixes, coupled with the hope that size-based targets would reduce the incentive to comply through downsizing. The irony that the new system created its own upsizing incentive, documented in the economic literature within a few years, does not retroactively make the 2006 rationale dishonest; it makes the design problem harder than the rulemaking acknowledged. Congress watched the pilot, liked the architecture, and legislated it for the whole program the following year.

What is a vehicle footprint?

The footprint is the rectangle bounded by a vehicle’s wheelbase and average track width, in square feet. Under section 32902(b)(3), each model’s economy target is a mathematical function of its footprint: larger footprints get lower targets, smaller footprints get higher ones.

The footprint is conventionally computed as track width in inches times wheelbase in inches divided by one hundred forty-four, and the fleet obligation is the sales-weighted average of the model-level targets. The function is continuous, so there is no cliff at any particular size, only a smooth curve that assigns each increment of size a slightly more lenient obligation.

What the 2007 rewrite changed

The Energy Independence and Security Act of 2007, Public Law 110-140, carried the fuel-economy title Congress called the Ten-in-Ten Fuel Economy Act, and it rewrote the program’s standard-setting machinery more thoroughly than any enactment since 1975. Four changes did the heavy lifting. First, section 32902(b)(3) required the Secretary to prescribe separate average fuel economy standards for passenger and non-passenger automobiles based on one or more vehicle attributes related to fuel economy, expressed as a mathematical function. That is the statutory basis of the footprint system: Congress legislatively mandated attribute-based standards for both fleets beginning with model year 2011, extending to passenger cars the approach the regulator had piloted for trucks. Second, section 32902(b)(2)(A) set a combined-fleet target of at least 35 miles per gallon for model year 2020, with ratable annual increases across model years 2011 through 2020. Third, the act preserved and recodified the minimum domestic-passenger-car standard, at section 32902(b)(4): the greater of 27.5 miles per gallon or 92 percent of the Secretary’s projected combined domestic and import fleet average, a floor beneath the attribute system that no credit maneuvering can satisfy. Fourth, the act added work trucks and medium- and heavy-duty vehicles to the program’s reach at section 32902(k).

What did the 2007 law change about how targets are set?

It replaced administratively chosen flat fleet numbers with legislatively mandated attribute-based targets. Section 32902(b)(3) requires standards expressed as a mathematical function of vehicle attributes for both cars and trucks from model year 2011, set a 35-mpg target for 2020, kept a minimum domestic-car floor, and extended the program to work trucks and heavy-duty vehicles.

The 2007 law also rebuilt the flexibility provisions that let manufacturers comply across model years and across fleets. Section 32903, as amended by the act’s section 104, extended credit carry-forward from three to five consecutive model years after the year the credits are earned, preserved carry-back of up to three model years before the year earned, authorized the Secretary to establish a credit-trading program among manufacturers at section 32903(f) with the statutory condition that traded credits preserve total oil savings, and authorized within-manufacturer transfers between compliance categories at section 32903(g). The carry provisions let a maker bank overcompliance from a strong year against a weak one; the trading and transfer provisions let the industry’s cheapest compliance do the work. One limitation keeps the system honest: credits earned by trading or by transfer cannot be used to satisfy the section 32902(b)(4) minimum domestic-passenger-car standard, so the floor under the domestic car fleet must be met with actual domestic-car efficiency, not purchased paper.

The footprint mechanism itself deserves a close look, because the formula is where the modern program’s behavior is determined. Each vehicle has a footprint, computed as track width in inches multiplied by wheelbase in inches, divided by 144, yielding square feet. The standard for a manufacturer’s fleet is then derived from a target curve: for each footprint value, the curve specifies a miles-per-gallon target, with larger footprints assigned lower targets. The manufacturer’s obligation is the sales-weighted harmonic average of the targets for the vehicles it actually sold, computed on the same harmonic-mean principle as the compliance average itself. Two consequences follow immediately. The compliance number moves with the product mix: sell larger vehicles and the required average falls. And no two manufacturers face the same number unless they sell the same mix: the standard is manufacturer-specific, recomputed every model year from actual sales. The flat number of the old era is gone; in its place stands a function.

The manufacturer-specific character of the standard is the attribute system’s most radical departure from the 1975 design. Under the old flat numbers, every maker faced the same 27.5; under the footprint system, a maker selling mostly compact cars faces a substantially more demanding number than a maker selling mostly large trucks, and neither number appears in the statute. The statute supplies the function; the market supplies the inputs. Defenders of the system argue that this is fairness: each maker is judged against the vehicles it chooses to sell rather than against a competitor’s lineup. Economists who prefer flat standards paired with credit trading contend that it is a concession: the program no longer asks the industry as a whole to reach a defined efficiency level, but asks each maker to improve modestly from wherever its mix puts it. Both descriptions fit the same mathematics, and the choice between them is a choice about what the program is for.

A guided tour of the chapter

Chapter 329 of Title 49 rewards a section-by-section reading, because the architecture is unusually clean. Section 32901 is the dictionary: it defines automobile, passenger automobile, non-passenger automobile, work truck, fuel economy, model year, and the other terms on which every operative provision depends. The definitions are not decorative. As the classification discussion showed, the question of which fleet a vehicle belongs to, and therefore which standard applies, is answered almost entirely within this section and the regulations it authorizes. A reader who masters 32901 has mastered the program’s vocabulary, and most misreadings of the program trace to skipping it.

Section 32902 is the engine room: the standard-setting authority. It contains the maximum-feasible mandate, the attribute-based requirement added in 2007, the 35-miles-per-gallon combined target for model year 2020, the ratable-increase command, the minimum domestic-passenger-car standard, and the newer provisions for work trucks and heavy-duty vehicles. If 32901 answers “what is covered,” 32902 answers “what is required,” and the two sections together state the program’s substantive law. The remaining sections are the machinery that makes the substantive law operable.

Section 32903 is the flexibility title: the earning, banking, borrowing, trading, and transferring of credits. Section 32904 is the measurement title: the harmonic-mean formula, the domestic and imported calculations, and the testing provisions that assign each model its rated figure. Section 32912 is the enforcement title: the civil penalty formula and the Secretary’s authority to adjust the rate, the provision that makes the rest of the chapter credible. Section 32919 is the federalism title: the preemption of state fuel-economy law and its carve-outs. The sections between these landmarks address the program’s ancillary needs: testing procedures and the EPA’s role in them, judicial review of the standards, the fuel-economy labeling program, recordkeeping and reporting duties, and the administrative details of running a fleet-average regime. The chapter is short enough to read in full, and the reader who does will find that its reputation for complexity comes less from any single provision than from the interaction of the averaging formula, the fleet definitions, and the credit system operating at once.

Two reading habits will serve anyone who opens the Code with the citations in hand. First, read the formula before the numbers, because the numbers change with each rulemaking while the harmonic mean, the fleet definitions, and the credit rules persist across decades. Second, read the exceptions before the rules, because the domestic-import separation, the classification regulation, and the waiver mechanism do more to shape outcomes than the headline standards they qualify. The program rewards the reader who starts with section 32904 and works outward, and it punishes the reader who starts with the press release.

Why Congress chose the attribute system in 2007

The 2007 decision to mandate attribute-based standards legislatively, rather than leaving the 2006 administrative pilot to run its course, reflected a congressional judgment about fairness that had been building for years. Under the flat standards, manufacturers whose historical product mix skewed toward larger vehicles faced structurally higher compliance costs than makers of small cars, and the disparity had no relationship to engineering effort or good faith. The attribute system promised to judge each manufacturer against its own product choices: the target would follow the footprint, and the engineering challenge would be comparable across the industry. The Senate report accompanying the legislation framed the reform as modernization, replacing a one-size-fits-all number with a system that recognized the diversity of the American market.

The choice also reflected a deliberate move away from the downsizing incentive that the flat standards created. Under a single fleet-wide number, the cheapest compliance path is to sell smaller vehicles, and the industry’s safety advocates had long argued that weight reduction carried occupant-protection costs. The footprint system was presented as the answer: by tying each vehicle’s target to its size, the standard would no longer reward shrinking cars, and manufacturers could meet their obligations through technology rather than through mix shifts. The presentation was sincere, and the economic literature’s later finding, that the system rewards upsizing instead, does not impeach the motive. It reveals the deeper difficulty: any standard that varies with a vehicle attribute creates an incentive to adjust the attribute, and the only standard with no attribute incentive is a flat one, which creates the mix incentive instead. Congress chose which incentive to live with, and the 2007 act’s minimum domestic-car floor and credit-trading provisions were its hedges against the choice’s known risks.

That function is the source of the program’s most cited unintended consequence. A rule designed to stop manufacturers from complying by shrinking cars made every vehicle’s obligation depend on its own size, which means enlarging a vehicle lowers its target. Economists have tested whether manufacturers exploit that margin, and the findings are uncomfortable. Whitefoot and Skerlos, in a 2012 Energy Policy study using firm-level simulations of the footprint system, found that automakers respond to footprint-based targets by increasing vehicle size in 20 of 21 scenarios, with footprint growth of roughly 2 to 32 percent eroding fuel-economy gains by an estimated 1 to 4 miles per gallon and raising carbon dioxide emissions 5 to 15 percent, the largest effects falling on light trucks. Ito and Sallee, studying Japan’s weight-based standards as a natural experiment in a 2017 Review of Economics and Statistics article, found that automakers increased vehicle weight to qualify for less stringent categories, with about 10 percent of vehicles affected at an average of roughly 110 kilograms heavier, and concluded that attribute basing costs roughly twice as much as a flat standard combined with credit trading. The literature does not claim that every upsizing decision traces to the footprint curve; fuel prices and consumer taste drive size as well. But the incentive is written into the formula, and the measured response is real. The most cited unintended consequence in modern regulatory design is not an accident of implementation. It is the standard’s own arithmetic.

The economic theory behind attribute basing starts from a genuine problem. A flat standard applied to manufacturers with very different product mixes imposes very different compliance costs, and the cheapest compliance path under a flat standard is to change the mix rather than the engineering: sell smaller vehicles, or, perversely, drop the large ones. Attribute basing was supposed to neutralize the mix incentive by adjusting the target to the product, so that engineering effort rather than product planning would do the work. The theory is elegant. Its flaw is that the adjustment itself becomes a product-planning variable: if the target depends on size, then size becomes a compliance instrument, and the manufacturer choosing between a 45-square-foot and a 47-square-foot version of a vehicle is choosing between two different legal obligations.

The policy dilemma the findings pose has no clean resolution within the statute’s terms. A flat standard punishes makers of large vehicles and invites mix shifting; an attribute standard invites upsizing and costs more per gallon saved. Credit trading softens both problems by letting the cheapest compliance, wherever it occurs, cover the most expensive, which is why the economists’ preferred alternative pairs a flat standard with trading. The 2007 Congress chose attribute basing with trading, a hybrid that captures some of each approach’s virtues and some of each approach’s vices. The footprint inversion is therefore not a drafting error to be corrected but a structural trade-off to be understood: the formula that made the standard fairer across manufacturers also made it softer against the industry’s oldest compliance strategy, building bigger.

Credits, penalties, and the price of missing

The credit system at section 32903 is the program’s shock absorber, and its mechanics reward close reading. A manufacturer that beats its standard in a model year earns credits denominated in tenths of a mile per gallon times vehicles, effectively. Those credits may be applied forward for five consecutive model years or backward for three, which lets a maker smooth the lumpiness of product cycles: a redesign year that overshoots can cover the years before and after. Trading under section 32903(f) lets one manufacturer’s surplus cover another’s shortfall, subject to the Secretary’s program design and the statutory requirement that total oil savings be preserved, meaning trades cannot be structured to lose gallons. Transfers under section 32903(g) let a single manufacturer move credits between its own car and truck fleets, crossing the car-truck boundary that the compliance averages respect. And as noted, the domestic-import wall of section 32904(b) opens for credit accounting, so the two domestic and imported passenger-car groups, which must each comply separately, can share credits. The system is generous by design: Congress preferred that manufacturers bank, borrow, and trade their way to compliance rather than pay penalties or distort product plans.

The credit mechanics at 49 CFR Part 536 translate the statute’s flexibility provisions into an operating system. Credits are earned in the currency of the program, tenths of a mile per gallon multiplied by vehicles, and they are banked against the model year in which they were earned. The five-year carry-forward window means that a manufacturer coming off a strong redesign cycle can coast, in compliance terms, through several weaker years, while the three-year carry-back means that a shortfall already incurred can be cured by future overcompliance. The windows are asymmetric by design: Congress wanted to reward early action more than it wanted to forgive past failures, so the forward window is longer. Trading extends the budget across the industry, and the statutory condition that traded credits preserve total oil savings ensures that the market moves gallons rather than paper. The limitation that traded and transferred credits cannot satisfy the minimum domestic-passenger-car standard keeps the market from hollowing out the program’s floor: the domestic car fleet must achieve its efficiency in metal, not in paper.

The penalty for falling short anyway is set at section 32912(b): five dollars, multiplied by each tenth of a mile per gallon by which the standard exceeds the manufacturer’s average fuel economy, multiplied by the number of vehicles, reduced by available section 32903 credits. The statute’s base rate of five dollars dates to 1975, but the operative rate through the date wall was higher. The Secretary exercised section 32912(c) authority to raise the rate to $5.50 per tenth of a mile per gallon per vehicle beginning with model year 1998, and that $5.50 figure was the rate in effect as of November 2016. The arithmetic of a shortfall is therefore straightforward to illustrate in principle: a maker missing its standard by a full mile per gallon across a million vehicles, with no credits to apply, faces ten tenths times $5.50 times one million, or $55 million, before any settlement or enforcement discretion. A smaller illustration shows the same linear scaling: a shortfall of half a mile per gallon across one hundred thousand vehicles produces a base liability of five tenths times $5.50 times one hundred thousand, or $2.75 million, before credits. In practice the credit system’s five-year carry-forward and three-year carry-back mean that outright penalty payments are the last resort rather than the routine outcome, and the industry’s compliance history shows makers managing their credit banks across product cycles to avoid the cash penalty.

The penalty’s practical role is deterrence at the margin rather than routine punishment. The $5.50 rate per tenth of a mile per gallon per vehicle, in effect from model year 1998 under the Secretary’s section 32912(c) authority, is large enough to sting and small enough that a manufacturer facing a choice between an expensive last-minute engineering fix and a penalty payment will sometimes rationally choose the payment, particularly late in a model year when the mix is already set. The statute anticipates this calculus by making credits the first resort: the penalty is reduced by available section 32903 credits, so a maker pays only on the shortfall it cannot cover from its bank. Enforcement then becomes a year-end accounting exercise rather than a courtroom drama, with the regulator verifying the averages, applying the credits, and assessing the remainder. The penalty is the backstop that makes the whole flexibility structure credible, the price that keeps the credits honest.

The compliance calendar of a model year

The program’s unit of account is the model year, and the provisions this article has walked through come together in the annual rhythm of a manufacturer’s compliance operation. Five to seven years before the model year, product planners fix the lineup: which models will be offered, their footprints, their powertrains, and their projected sales volumes. The footprint curves for the model year, published years earlier under the ratable schedule, assign each planned model a target, and the planners compute the sales-weighted harmonic-mean target for each fleet: domestic cars, import cars, and light trucks. On the performance side, the engineers project each model’s laboratory fuel economy from the test cycles, and the same harmonic-mean arithmetic produces the projected fleet average. The gap between the two numbers, target and projection, is the compliance position, and everything the company does for the rest of the cycle is about closing it.

The tools for closing the gap are the ones the statute provides. Technology deployment raises the projected average. Pricing and mix shift move the sales weights: incentives on efficient models, allocation of scarce high-demand models, the quiet art of selling more of what helps the average. Credit banking covers the residual: surplus from a good year carried forward up to five years, a bad year’s shortfall carried back up to three, trades with competitors under section 32903(f) preserving total oil savings, transfers between the company’s own fleets under section 32903(g) except against the domestic-car minimum. The penalty at section 32912(b), five dollars and fifty cents per tenth per vehicle at the date-walled rate, is the price of failing to close the gap by any of these means, and the compliance operation is managed so that the penalty is the option of last resort rather than a cost of doing business.

The EPA Administrator’s calculation under section 32904(a)(1)(B) is the moment of truth, performed after the model year closes on the reported production and the measured laboratory values. The separate books for domestic and import passenger cars are totaled separately, the truck fleet is totaled on its own curve, the credits are applied, and the result is compliance, a banked surplus, or a penalty. The cycle then repeats, with the next model year’s curve sitting slightly higher under the ratable schedule, the banked credits aging toward expiration, and the product plan for five years hence already being drawn. The statute governs this entire rhythm through a handful of provisions: the standard-setting authority, the calculation formula, the fleet definitions, the attribute mandate, the credit rules, and the penalty. That is the operative text as the location of the real policy, the series thesis this article was written to illustrate.

Preemption and the two 2007 court decisions

Section 32919(a) provides that when an average fuel economy standard under the chapter is in effect, a state or political subdivision may not adopt or enforce a law or regulation related to fuel economy standards or average fuel economy standards for covered automobiles. The provision is express preemption, and it reads like a wall: fuel economy is federal territory. But the wall has a door, and the door is the Clean Air Act. The preemption question that reached the courts in 2007 was whether state greenhouse-gas emission standards for vehicles, adopted through the Clean Air Act’s California waiver mechanism, count as laws “related to” fuel economy standards. The automakers said yes, arguing that carbon dioxide emissions and fuel consumption are two sides of one tailpipe, so regulating one regulates the other. The states said the two regimes coexist by congressional design.

Two federal district courts answered the question the same way in 2007. In Green Mountain Chrysler Plymouth Dodge Jeep v. Crombie, 508 F. Supp. 2d 295 (D. Vt. 2007), decided September 12, 2007, Judge Sessions held that Vermont’s adoption of California’s greenhouse-gas vehicle standards was not preempted by the fuel-economy statute, assuming the federal environmental agency granted the Clean Air Act waiver, finding that Congress was aware of the fuel-economy and emissions overlap and intended the two regimes to coexist. In Central Valley Chrysler-Jeep, Inc. v. Goldstene, 529 F. Supp. 2d 1151 (E.D. Cal. 2007), as corrected in March 2008, the Eastern District of California reached the same holding for California’s own standards. The precedential weight of these decisions must be stated precisely: both are federal district-court rulings, and neither became circuit precedent, because the appeals were dismissed after the 2009 national agreement among the federal administration, the California Air Resources Board, and the automakers harmonized the federal and state programs. They stand as the reasoned view of two trial courts, not as settled appellate law, and any account that presents them as circuit authority overstates the record. The placement of the preemption provision as the chapter’s final section is itself a structural signal: Congress put the federalism boundary after the standards, the credits, and the penalties, marking where the program’s operative text ends and the states’ reserved territory begins.

The preemption provision’s two companion subsections complete its picture. Section 32919(b) extends the preemption to state fuel-economy labeling requirements, ensuring that the federal label is the only one consumers see, while section 32919(c) expressly permits states to set fuel-economy standards for vehicles they purchase for their own fleets, a carve-out that lets governments lead by example without regulating the market. The structure is coherent: the market-facing rules are exclusively federal, and the state’s role as a market participant is preserved. The provision’s drafting reflects the 1975 Congress’s judgment that a patchwork of state fuel-economy standards would balkanize the national automobile market, a judgment that has never been seriously revisited. The operative phrase is “related to,” and the question the courts faced was whether a greenhouse-gas emission standard, which necessarily affects fuel economy, counts as related to fuel economy standards within the meaning of the fuel-economy chapter. The Green Mountain and Central Valley courts read it narrowly, holding that Congress legislated with knowledge of the overlap and intended coexistence, and the 2009 national agreement made the question academic for the industry’s planning.

The second regulator arrives

The third regime governing the same tailpipe did not come from Congress at all. It came from the Supreme Court’s reading of a 1970 statute. In Massachusetts v. EPA, 549 U.S. 497 (2007), decided April 2, 2007 by a five-to-four vote, the Court held that greenhouse gases are “air pollutants” under the Clean Air Act and that the Environmental Protection Agency has statutory authority to regulate greenhouse-gas emissions from new motor vehicles under section 202(a) of that act, codified at 42 U.S.C. section 7521(a), once the agency makes the required endangerment finding. The precise holding matters because it is so often overstated: the Court did not order the agency to regulate. It held that the agency could not decline to decide the endangerment question on policy grounds divorced from the statute, and it remanded for a decision grounded in the statutory text. The decision authorized regulation; it did not compel it. The agency’s subsequent vehicle greenhouse-gas standards rest on that authorized authority, and the full story of the decision belongs to the series’ case explainer, Massachusetts v. EPA.

Did the Supreme Court order EPA to regulate vehicle greenhouse gases?

No. Massachusetts v. EPA, 549 U.S. 497 (2007), held that greenhouse gases are air pollutants under the Clean Air Act and that EPA may regulate them from new vehicles under section 202(a) after an endangerment finding. The Court barred the agency from refusing to decide on non-statutory policy grounds, but the choice to regulate remained the agency’s.

Once the environmental agency held that authority, a single vehicle faced two federal masters. The transportation regulator sets fuel-economy standards under the “maximum feasible” mandate of section 32902(a). The environmental agency sets greenhouse-gas emission standards under Clean Air Act section 202(a). The two measure different things in different units, miles per gallon against grams of carbon dioxide per mile, but they constrain the same engineering decisions, because burning less fuel per mile emits less carbon dioxide per mile. The agencies coordinated their rulemakings for model years 2012 through 2016 and again for 2017 through 2025, issuing joint rules that let manufacturers plan one compliance strategy for both regimes. The joint rulemakings are the practical answer to the overlap: rather than litigating which regime governs, the two agencies wrote their rules together.

The endangerment finding is the trigger the opinion left in the agency’s hands. Section 202(a) authorizes regulation of emissions from new motor vehicles that, in the Administrator’s judgment, cause or contribute to air pollution reasonably anticipated to endanger public health or welfare. The finding is a scientific and judgmental determination, reviewable by the courts, and it is the legal foundation on which the vehicle greenhouse-gas standards rest. Once made, it activated the EPA’s authority to set fleet-average greenhouse-gas standards expressed in grams of carbon dioxide per mile, a metric that constrains the same combustion chemistry as the miles-per-gallon standard but from the emissions side of the ledger. The two agencies then did what the overlapping statutes practically required: they wrote joint rulemakings for model years 2012 through 2016 and again for 2017 through 2025, aligning their analytical baselines and compliance timelines so that manufacturers could engineer once and comply twice. The joint rules are administrative coordination achieving what the statutes, written decades apart, never expressly provided.

When the standards go to court

Fuel-economy rulemakings are among the most litigated actions in the administrative state, and the litigation follows a predictable pattern. After the transportation regulator issues a final rule setting standards for a span of model years, manufacturers, states, and advocacy organizations petition for review in the federal courts of appeals, challenging the maximum-feasible balancing that produced the numbers. The challengers’ arguments track the four statutory factors: the industry contends the standards exceed what is technologically feasible or economically practicable within the lead time, while states and environmental organizations contend the standards fall short of the maximum that feasibility allows. Both sides draw on the same rulemaking record and reach opposite conclusions about what it proves, which is why the cases turn less on the engineering than on the standard of review.

That standard is the Administrative Procedure Act’s familiar test: the court asks whether the agency’s action was arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law. In practice this means the court examines whether the regulator considered the factors Congress specified, whether its technical analysis hangs together, and whether its explanation connects the evidence to the chosen numbers. Courts do not set fuel-economy standards themselves; they remand rules whose reasoning does not survive scrutiny and let the agency try again. The dynamic gives the rulemaking record its enormous size: every technology assessment, cost estimate, and sales forecast in the record exists partly to survive this review. The program’s standards have thus been shaped as much by the anticipation of litigation as by the litigation itself, with each rulemaking building a more elaborate defense of its balancing than the last.

The joint rulemakings: how two agencies wrote one program

The practical answer to the three-regime overlap was administrative coordination on a scale the drafters of neither statute imagined. For model years 2012 through 2016, and again for 2017 through 2025, the transportation regulator and the environmental agency issued joint rulemakings that set the fuel-economy standards and the greenhouse-gas standards together, on a shared analytical foundation. The joint rules aligned the agencies’ technology assessments, cost estimates, and compliance timelines, so that a manufacturer could plan a single product strategy against both sets of numbers. The coordination was not legally required by either statute; it was the agencies’ solution to the problem that two independent rulemakings constraining the same engineering decisions would otherwise create, with the risk that a manufacturer could comply with each regime separately only by building two different fleets.

The joint rulemakings also had to reconcile two different units of account. The fuel-economy standards speak in miles per gallon, averaged harmonically across each fleet; the greenhouse-gas standards speak in grams of carbon dioxide per mile, with their own averaging and credit provisions. The coordination rests on a physical identity: a gallon of gasoline contains a fixed quantity of carbon, and burning it releases a fixed mass of carbon dioxide, so miles per gallon implies grams of carbon dioxide per mile through a fixed conversion. The two metrics are reciprocals scaled by a constant, which means a fuel-economy standard and a greenhouse-gas standard set at equivalent stringency demand exactly the same combustion efficiency. The equivalence has limits that the regulators’ separate programs reflect: the greenhouse-gas program covers gases and sources the fuel-economy program does not, including air-conditioning refrigerants and other tailpipe gases that have no fuel-economy analogue, so a manufacturer can comply with the greenhouse-gas standard through measures that do nothing for its harmonic-mean fuel economy. That is why the joint rulemakings calibrate the headline numbers to converge but maintain separate compliance books: same physics, different law.

The 2009 national agreement that made the coordination possible deserves its place in the story. With the two 2007 district-court preemption rulings on appeal and the waiver requests pending, the federal administration, the California Air Resources Board, and the automakers negotiated a harmonized national program: the federal agencies would set coordinated standards, California would accept compliance with the federal program as compliance with its own, and the appeals would be dismissed. The agreement is why the preemption appeals never produced circuit precedent and why the state and federal programs, which the courts had held could coexist, were in practice merged into a single compliance exercise. It was an executive and negotiated solution to a problem the statutes had created and the courts had only partially addressed, and it held through the rulemakings that followed.

California, the waiver, and the fifty-state question

The third layer of the overlap is the oldest in one sense and the newest in application. Clean Air Act section 209(a), codified at 42 U.S.C. section 7543(a), preempts states from adopting their own new-motor-vehicle emission standards. Section 209(b), at 42 U.S.C. section 7543(b), then requires the federal environmental agency to waive that preemption for California, which had standards in place before March 30, 1966, unless the agency finds that the state’s determination is arbitrary and capricious, that the state has no compelling and extraordinary need, or that the standards are inconsistent with section 202(a). California’s 2004 greenhouse-gas vehicle rules, phased in over eight model years beginning with 2009, were the first application of the waiver mechanism to climate pollutants, and the waiver decisions that followed are chronicled in the series’ rulemaking explainer, EPA clean air act rulemaking.

Section 177 of the Clean Air Act, at 42 U.S.C. section 7507, extends the mechanism beyond California. Any other state with a federally approved implementation plan may adopt and enforce California’s waived standards, provided the state standards are identical to California’s and provided at least two model years of lead time. No separate federal approval is required. The legal operation is therefore precise: California innovates, the federal agency waives preemption unless one of three statutory findings defeats the waiver, and other states may opt into the identical California program. The arrangement is neither a state veto of federal policy nor a federal rubber stamp of state policy; it is a statutory procedure with defined findings, and the politics that surround it do not change the procedure’s terms. Note the citation discipline here: section 177 is a Clean Air Act provision, and it should never be written as a Title 49 section.

The waiver provision’s three findings structure the entire state-federal negotiation. Under section 209(b), the federal environmental agency must grant California’s waiver request unless it finds that the state’s determination is arbitrary and capricious, that the state does not need the standards to meet compelling and extraordinary conditions, or that the standards are inconsistent with section 202(a) of the Clean Air Act. The findings set a high bar for denial: California’s historic air-quality problems satisfy the compelling-conditions prong as a matter of course, and inconsistency with section 202(a) requires a showing that the state program would undermine the federal one. The provision thus creates a presumption in favor of the waiver, rebuttable only on the three specified grounds, which is why California has historically received its waivers and why denials are the rare, contested exceptions. The identicality requirement for section 177 adoption prevents a patchwork of fifty different programs while allowing a bloc of states to follow California’s lead, and the lead-time requirement gives manufacturers planning certainty.

The section 177 mechanism is deliberately lightweight. A qualifying state adopts California’s waived standards identically, with at least two model years of lead time, and no separate federal approval is required; the state’s Clean Air Act implementation plan supplies the authority. The design balances uniformity against state initiative: one engineering target for the industry, developed through California’s waiver, with other states permitted to join the same greenhouse-gas numbers without multiplying compliance obligations. A bloc of states thus operates under California’s program while the two federal programs run alongside it, and manufacturers plan for the strictest applicable number in each state where they sell.

Readers encountering the program for the first time usually ask who is actually in charge, and the honest answer is that the question has no single answer. The transportation regulator writes the fuel-economy numbers under a 1975 energy statute. The environmental agency writes the greenhouse-gas numbers under a 1970 clean-air statute as interpreted by a 2007 Supreme Court decision. California writes a third set of greenhouse-gas numbers under a waiver provision of that same clean-air statute, and other states may adopt California’s program identically. The regimes overlap because they govern the same physical object, the burning of fuel in an engine, measured in different units and justified by different statutory purposes. The table below maps the overlap as a working reference: each row is a regulator, its authority, what it constrains, how the constraint is expressed, and how it bears on a single vehicle leaving the factory.

How an outsider can check the math

The program’s arithmetic is public in a way that invites independent verification, and the verification exercise is worth describing because it shows how the statute’s transparency works. The inputs to any manufacturer’s compliance calculation are, in principle, observable: the rated fuel-economy figure for each model, from the EPA’s test procedures, and the number of each model sold in the model year, from industry sales data. The formula is statutory text, not agency discretion: the sales-weighted harmonic mean of section 32904(a)(1)(B), applied separately to each fleet. The standards are published rulemakings, and the credit provisions are codified at section 32903 and detailed at 49 CFR Part 536. A researcher with the sales data and the ratings can therefore replicate the compliance calculation for any manufacturer and any model year, up to the limits of the credit bank, whose contents are less fully public.

The exercise reveals both the program’s discipline and its soft edges. The discipline is the formula: given the sales and the ratings, the average is determined, and there is no room for negotiation about what the numbers say. The soft edges are the credits and the attribute targets: the credit bank’s history across the five-year forward and three-year backward windows requires assumptions, and the footprint-derived target requires the footprint of every vehicle sold. Independent estimates of compliance margins therefore carry uncertainty bands, and different analysts’ bands differ. But the structure of the uncertainty is itself informative: it concentrates in the flexibility provisions, not in the measurement. The statute measures rigorously and forgives flexibly, and any honest audit of the program must keep those two operations distinct.

The number on the window

Most Americans never encounter the harmonic-mean formula or the footprint curve. They encounter the program through the fuel-economy label on the window of a new car, the miles-per-gallon figures that shape purchase decisions and dinner-table arguments. The labeling program is the statute’s consumer face, and its relationship to the compliance program is worth stating precisely: the label reports the rated fuel economy of the individual vehicle, derived from the same EPA test procedures that feed the compliance calculation, but the compliance obligation attaches to the fleet average, not to the labeled figure. A vehicle labeled at 18 miles per gallon is perfectly legal to sell; it simply drags the manufacturer’s average down in proportion to its sales, and the manufacturer must offset it elsewhere in the lineup.

The distinction matters because public debate routinely confuses the two. Headlines about a thirsty new model “violating” fuel-economy standards misunderstand the program’s unit of account: no individual vehicle violates anything, and the only violation the statute recognizes is a fleet average below the standard after credits. The label informs the buyer; the average constrains the maker. Section 32919(b)’s preemption of state labeling requirements ensures that the federal label is the only one, so that the consumer-facing number is consistent nationwide even where the underlying regulatory regimes differ by state. The window sticker is thus the program’s most democratic artifact: the same test data that moves billions of dollars in compliance strategy, translated into the single number by which most citizens judge the law.

The three-regime table

Regulator	Statutory authority	What it regulates	How the obligation is expressed	How it bears on one vehicle
National Highway Traffic Safety Administration, Department of Transportation	Energy Policy and Conservation Act, codified at 49 U.S.C. chapter 329, under the “maximum feasible” mandate of section 32902(a)	Average fuel economy of each manufacturer’s passenger-car, light-truck, domestic-car, and imported-car fleets	Miles per gallon, as a sales-weighted harmonic mean under section 32904(a)(1)(B), against attribute-based target curves under section 32902(b)(3), with credit banking, borrowing, trading, and transfer under section 32903 and civil penalties under section 32912	The vehicle counts toward its fleet’s harmonic-mean average and toward the manufacturer’s footprint-derived target
Environmental Protection Agency	Clean Air Act section 202(a), 42 U.S.C. section 7521(a), following Massachusetts v. EPA, 549 U.S. 497 (2007)	Greenhouse-gas emissions from new motor vehicles, including carbon dioxide	Grams of carbon dioxide per mile, as a fleet average with its own credit system, coordinated with the fuel-economy rules through joint rulemakings	The same vehicle must satisfy both the miles-per-gallon average and the grams-per-mile average, constraining the same fuel-burn engineering
California Air Resources Board, with other states opting in	Clean Air Act section 209(b) waiver, 42 U.S.C. section 7543(b), and section 177 state adoption, 42 U.S.C. section 7507	New-motor-vehicle greenhouse-gas emission standards stricter than the federal baseline, where the federal agency grants the waiver	Grams per mile under California’s program, identical in any section 177 opt-in state, with at least two model years of lead time	A vehicle sold in California or an opt-in state must meet the California program’s terms in addition to the two federal regimes

The table is the article’s findable artifact because it answers the question the three-regime structure always raises: who is actually in charge. The answer is that no one is, exclusively. The transportation regulator owns fuel economy, the environmental agency owns greenhouse gases, and California owns the waiver procedure that can add a third binding program in the states that choose it. A single vehicle leaving a factory in Ohio and sold in Sacramento carries all three obligations, and the compliance departments of the major manufacturers plan for all three at once. The overlap is not an accident of drafting; it is the product of two statutes written decades apart, a Supreme Court decision that connected them, and a waiver provision that predates both.

A fourth federal regime touches the same product for entirely different reasons. The motor vehicle safety act of 1966 created the federal safety standards that govern crashworthiness, and the same transportation safety regulator that writes the fuel-economy numbers writes the safety rules. The safety regime constrains weight reduction, the oldest fuel-economy strategy, because lighter vehicles historically raised occupant-safety questions, and the interaction between the safety and efficiency rules is one of the program’s permanent tensions. The regimes do not formally trade off against each other in either statute, but every engineer working under both knows the negotiation by heart.

Why the fuel-economy chapter outlasted its 1975 siblings

The Energy Policy and Conservation Act of 1975 built an entire architecture of energy-security instruments, and most of them are gone. Price controls were dismantled, the export ban survived until its December 2015 repeal, and the crisis agencies were reorganized out of existence. The fuel-economy chapter endured, and the reasons are structural rather than sentimental. A fleet-average standard is self-updating in a way that a price control is not: the regulator can revise the numbers through rulemaking as technology and circumstances change, without returning to Congress for new legislation. The 2007 rewrite demonstrated the point at the statutory level, converting the flat standards to an attribute system by act of Congress, but the decades of administrative standard-setting between 1975 and 2007 demonstrated it first, as the regulator adjusted the program within the maximum-feasible framework through successive rulemakings.

The program also survived because its purpose proved renewable. Energy security meant one thing in 1975, when the embargo was fresh, and another in the 2000s, when import dependence was framed around price volatility and geopolitics rather than physical shortage; the statute’s conservation purpose accommodated both framings without amendment. When the climate purpose arrived through the second regulator, the program absorbed that too, not by changing its own text but by operating alongside the greenhouse-gas regime in the joint rulemakings. Few federal regulatory programs have served three successive policy masters, energy security, economic competitiveness, and climate, with the same operative text. The chapter’s durability is a testament to the 1975 design choice that this article has emphasized throughout: Congress wrote a formula and a procedure rather than a fixed outcome, and formulas and procedures outlast the crises that produce them.

The honest account of what the program achieved

The fairest criticism of the program is also the most common: fleet fuel economy improved far less than the per-vehicle numbers suggest it should have. The criticism is worth stating carefully, because it contains both a true observation and a misattribution. New-vehicle efficiency improved substantially across the program’s life; the engineering response to the harmonic-mean incentive was real, and the reengineering of the late 1970s and early 1980s permanently changed what a car could do with a gallon. But the fleet average, the number the public sees, is sales-weighted, and the sales mix shifted hard toward light trucks across the 1990s and 2000s. The truck fleet’s lower targets, combined with the classification rules that welcomed SUVs into the truck class, meant that the mix effect absorbed much of the per-vehicle gain. The honest account separates the two: per-vehicle efficiency rose markedly, while the composition of what Americans bought moved in the opposite direction.

The mix shift itself has multiple causes, and the program is only one of them. Fuel prices collapsed in real terms in the mid-1980s and stayed low through the 1990s, which dulled the consumer incentive to buy efficient cars regardless of what the standards required. Consumer preference moved toward roomier, higher-riding vehicles for reasons of comfort, perceived safety, and utility that no regulation created. Demographics and land use played their part, as suburban households with long commutes made the sport utility vehicle the default family car. And the footprint system, once in place, added its own measured incentive toward larger vehicles, as the Whitefoot and Skerlos and Ito and Sallee findings document. Attributing the entire truck boom to the classification rules overstates the program’s causal role; ignoring the classification rules understates it. The program shaped the menu, but buyers chose from it under price signals the program did not control.

The safety argument about fuel economy standards deserves the same evenhanded statement. It runs as follows: raising miles per gallon historically meant reducing vehicle weight, lighter vehicles fare worse in collisions with heavier ones, and therefore the standards cost lives. The counter runs as follows: weight is not the only path to efficiency, the footprint system was designed precisely to remove the incentive to downsize, and the net safety outcome depends on empirical questions about fleet composition that no provision of the chapter answers by itself. The statute’s own answer to the dispute was the attribute system, chosen in part to make the standard neutral with respect to vehicle size. This article reports the dispute; it does not referee it.

The measured record on per-vehicle improvement is the program’s strongest card. The Energy Information Administration’s retrospective analysis found that fuel-economy standards affected vehicle efficiency across the program’s life, and the engineering history bears it out: the downsizing wave of the late 1970s and early 1980s and the steady refinement of engines, transmissions, and vehicle bodies moved the per-vehicle numbers in the direction the statute demanded. A new car in the mid-1980s did dramatically more with a gallon than its mid-1970s predecessor, and the harmonic-mean formula deserves substantial credit for concentrating the industry’s effort where the gallons were. The program’s defenders are entitled to this part of the record, and the critics who call the program a failure must contend with it.

The fleet-average record is where the criticism bites, and the mechanism is the sales mix. Total vehicle miles traveled grew relentlessly across the program’s life, as the population grew, suburbs spread, and driving got cheaper in real terms. The light-truck share of sales climbed through the 1990s and 2000s, carrying the lower truck targets into the weighted average. The two effects compounded: more miles driven in a mix tilted toward thirstier vehicles meant that total fuel consumption kept rising even as each vehicle burned less per mile. None of this was within the statute’s power to prevent directly. The program regulates the efficiency of new vehicles offered for sale; it does not regulate how many miles Americans drive, what fuel costs, or which vehicles they prefer. The honest measurement therefore holds two truths at once: the program substantially improved the efficiency of the vehicles it regulated, and the transportation sector’s total fuel use followed a trajectory shaped by forces the program never touched.

The compliance-cost and consumer-benefit arguments deserve the same evenhanded treatment. Manufacturers and their trade associations argued that tighter standards raise vehicle prices, because the engineering, materials, and testing required to meet the numbers cost money that flows through to buyers, and that the cost falls hardest on buyers of affordable vehicles. Consumer and environmental advocates argued that the fuel savings over a vehicle’s lifetime exceed the upfront price increase, so the standards pay for themselves at the pump, and that the savings accrue disproportionately to lower-income drivers who spend a larger share of income on fuel. Both arguments describe real mechanisms. The price effect is a cost borne at purchase; the fuel effect is a benefit realized over years of driving, discounted by each buyer’s time horizon and sensitive to fuel prices neither side controls. The program’s defenders could point to the decades of engineering improvement the standards forced; its detractors could point to the compliance strategies, including credit banking and mix management, that let makers meet the letter while the fleet average lagged the engineering. A neutral account reports both ledgers without declaring one the whole story.

Whether the program saved oil is the question the 1975 Congress actually asked, and it is the subject of the series’ impact analysis, energy independence policy impact. The program’s defenders credit it with holding American oil consumption far below the trajectory the pre-1975 fleet implied; its skeptics note that total consumption kept rising with vehicle miles traveled and that the standards were only one of several forces, alongside prices and technology, shaping demand. Measuring oil savings requires a counterfactual, and the choice of counterfactual determines the answer. The standard approach compares actual petroleum consumption against the trajectory implied by the pre-1975 fleet: the 12.9-miles-per-gallon fleet of 1974, extended forward with the growth in vehicle miles traveled that actually occurred. The skeptics’ counterfactual is different: they ask what the market would have delivered on its own, noting that fuel prices themselves drove efficiency investment and that the standards sometimes ratified improvements the industry was making anyway. The truth apportions the credit: the standards forced the pace in the early years when prices were high and the political will was strong, and they set a floor under efficiency in the later years when cheap fuel would otherwise have let it slide.

The purpose against which to judge the measurement is the one Congress stated. The 1975 Congress legislated against imported-oil dependence, and by that standard the program’s four decades are best read as a partial but real success: the American fleet uses far less imported oil per mile than the pre-embargo trajectory implied, even as total consumption reflects the miles Americans chose to drive. Judging the program against the climate goals that later attached to it asks the statute to answer a question it was never written to address, and the mismatch explains much of the talking past that characterizes the debate. The operative text is an energy-security instrument that acquired an environmental shadow; the honest account keeps the two purposes distinct and measures each against its own baseline.

One contextual note closes the energy-security frame. The crude-oil export ban, the other great 1975-era instrument of American energy policy, was repealed in December 2015 by section 101 of Title I of Division O of the Consolidated Appropriations Act, 2016, ending a restriction that had stood for roughly forty years. The repeal belongs to the program’s sibling article on export policy; it matters here only as a reminder that the energy-security architecture of 1975 did not survive intact, while the fuel-economy chapter did.

That fraction governed American automobiles for four decades, through the 1975 ramp, the long flat middle, the 2006 truck pilot, the 2007 rewrite, the joint rules, and the national program. It survived because it is flexible where flexibility matters, manufacturers choose their own mix, and rigid where rigidity matters, the average must be met. A reader who can compute the fraction, name the fleets, trace the footprint, and place the three regulators has the whole program in hand, which was the One Test this article set itself at the outset.

Students of the overlapping regimes often benefit from structured study aids. A legislation study notebook can help organize the three-regime overlap and the attribute formula for coursework in environmental and administrative law, and a civics study companion may assist with the statutory citations and the waiver mechanics. The operative text rewards the kind of close reading that study tools are built to support: the policy is in the formula, and the formula is in the statute.

Terms of art: the vocabulary that does the legal work

A handful of terms carry the program’s entire legal weight, and precision about them prevents most misreadings. “Fleet average” is the sales-weighted harmonic mean of a defined group of vehicles for one model year, computed under section 32904; it is not the average of the models’ ratings and not the average of anything else. “Footprint” is wheelbase times average track width divided by 144, in square feet; it is the attribute to which the target curve is keyed, and it has nothing to do with carbon footprints or any other use of the word. “Credit” is a unit of overcompliance, denominated in tenths of a mile per gallon times vehicles, that can be banked, borrowed, traded, or transferred under section 32903; it is not a subsidy and not a permission to pollute, but a quantified gallon-saving moved across time or parties. “Model year” is the program’s unit of account, the twelve-month production period to which each standard, average, and credit attaches; the statute’s deadlines, windows, and lead times all run in model years rather than calendar years.

Two more terms deserve care because they are routinely conflated. “Standard” is the legal obligation, the number the fleet average must meet; “target” is the footprint-derived figure for an individual vehicle or the curve-derived figure for a manufacturer, the input from which the standard is computed under the attribute system. Under the pre-2007 flat regime the two coincided, which is why older writing uses them interchangeably; under the attribute system they differ, and the difference is the point. Finally, “maximum feasible” is not a synonym for “as strict as possible.” It is the statutory instruction of section 32902(a) to set the most demanding standard that the four factors, technological feasibility, economic practicability, other government standards, and the need to conserve energy, will support. Every number in the program’s history is an application of those words, and every argument about the numbers is an argument about what they mean.

Three more entries complete the set. “Ratable” is the 2007 act’s instruction that annual standard increases be distributed across the model-year span rather than backloaded into the out years; it is an anti-procrastination rule for rulemaking, not a technical term, and it is why the 2011-through-2020 trajectory was a glide path rather than a cliff. “Domestic fleet” is the compliance group of passenger automobiles meeting the 75-percent North American value-added test of section 32904(b)(1)(A), calculated separately from the imported fleet as though made by a different manufacturer; the term matters because the minimum domestic-car floor of section 32902(b)(4) applies to this group alone and cannot be satisfied with traded or transferred credits. “Attribute system” is the post-2011 method in which each manufacturer’s standard is a mathematical function of vehicle attributes rather than a single flat number; footprint is the attribute Congress chose, though the statute permits others, and the system’s defining feature is that the obligation is recomputed every model year from the manufacturer’s actual sales mix.

The program as of November 2016: a snapshot

As of November 2016, the fuel-economy program stood as a mature regulatory system operating on its third structural design. The flat fleet-wide numbers of the 1975 era were history; the attribute-based system mandated by the 2007 act had governed both the passenger-car and light-truck fleets since model year 2011, with each manufacturer’s obligation computed as the sales-weighted harmonic mean of footprint-derived targets. The combined-fleet trajectory Congress set in 2007, at least 35 miles per gallon for model year 2020 with ratable annual increases, was being implemented through the joint NHTSA and EPA rulemakings that also carried the greenhouse-gas standards for model years 2017 through 2025. The civil penalty for noncompliance stood at $5.50 per tenth of a mile per gallon per vehicle, the rate in effect since model year 1998. The credit system, with its five-year carry-forward, three-year carry-back, inter-manufacturer trading, and intra-manufacturer transfers, gave the industry a full toolkit for managing compliance across product cycles.

The three-regime structure was fully operational. The transportation regulator set the fuel-economy numbers under the maximum-feasible mandate of 49 U.S.C. section 32902(a). The environmental agency set the greenhouse-gas numbers under Clean Air Act section 202(a), on the authority the Supreme Court recognized in Massachusetts v. EPA. California operated its waiver program under section 209(b), with other states following through section 177. The two 2007 district-court decisions holding the state greenhouse-gas programs not preempted by the fuel-economy statute remained the last judicial word on the overlap, their appeals long since dismissed after the 2009 harmonization agreement. The express preemption of section 32919 continued to bar state fuel-economy standards, while the waiver mechanism continued to permit state greenhouse-gas standards through the Clean Air Act’s separate procedure.

The policy debates of the program’s first decade had not been resolved so much as transformed. The question was no longer whether the government could require more efficient cars; four decades of the program had settled that. The questions were finer: whether the footprint curve’s upsizing incentive outweighed its fairness gains, whether the credit market’s flexibility came at the cost of real gallons, whether the truck classification still described the vehicles Americans actually drove, and whether the joint federal program could hold together the three regimes without new legislation. The economic literature had quantified the footprint incentive; the engineering record had vindicated the harmonic mean; the legal structure had survived direct challenge. What remained was the permanent work of a regulatory program: setting the next round of numbers under the maximum-feasible standard, defending the balancing of the four statutory factors, and measuring, against the 1975 purpose of reduced oil dependence, whether the formula was still doing the job Congress wrote it to do.

Frequently Asked Questions

Q: How do CAFE standards actually work?

The program sets a fleet-wide average fuel-economy obligation for each automaker under 49 U.S.C. chapter 329. The transportation safety regulator sets the numbers; the environmental agency measures each model’s fuel economy and computes the manufacturer’s average with a sales-weighted harmonic mean, so thirsty high-volume models drag the average down more than a simple average would. Separate averages apply to passenger cars and light trucks, and the car fleet is further divided into domestic and imported groups that must each comply on their own. Since model year 2011 the targets have been attribute-based: each vehicle’s footprint fixes its individual target, and the manufacturer’s required average is the harmonic mean of those targets across actual sales. Makers may bank credits for five model years, borrow against three prior years, trade surplus with other makers, or move credits between their own fleets. A shortfall remaining after credits draws a civil penalty, and two further regimes, the federal greenhouse-gas rules and California’s waiver program, constrain the same vehicles.

Q: Why do CAFE standards have separate car and truck fleets?

The split dates to 1975, when trucks meant work vehicles: pickups, vans, and utility models that needed torque and payload capacity the technology of the day could not deliver at car-like efficiency. One standard for everything would have forced either the abandonment of working trucks or a car target set by truck capability, so Congress let each fleet improve on its own curve. The statutory definitions live at 49 U.S.C. section 32901, with passenger automobiles and non-passenger automobiles distinguished by primary use and off-highway capability, and the classification regulation at 49 CFR 523.5 draws the exact boundary. What the drafters did not foresee was the migration of family transportation into the truck class. As minivans and then sport utility vehicles absorbed the station-wagon market, the easier fleet absorbed the volume, and the program’s real-world stringency became a function of a classification decision as much as of the numbers themselves.

Q: What is the footprint rule in CAFE standards?

The footprint rule makes each vehicle’s fuel-economy target depend on its size. Footprint is wheelbase multiplied by average track width, divided by 144, expressed in square feet. A target curve assigns a miles-per-gallon figure to every footprint value, with larger footprints receiving lower targets, and the manufacturer’s required fleet average is the sales-weighted harmonic mean of the targets for the vehicles it actually sold. The approach began as an administrative pilot: the 2006 reform rule offered light-truck makers a footprint track for model years 2008 through 2010 and made it mandatory for 2011. The Energy Independence and Security Act of 2007 then legislatively mandated attribute-based standards for both passenger cars and trucks from model year 2011, at 49 U.S.C. section 32902(b)(3). The intent was fairness across makers with different product mixes; the much-discussed side effect is that enlarging a vehicle lowers its obligation.

Q: Did CAFE standards encourage bigger vehicles?

The footprint formula creates a measurable incentive to build larger vehicles, and economists have documented the response. Whitefoot and Skerlos, in a 2012 Energy Policy study using firm-level simulations, found automakers increasing vehicle size in 20 of 21 scenarios, with footprint growth of roughly 2 to 32 percent eroding fuel-economy gains by an estimated 1 to 4 miles per gallon and raising carbon dioxide 5 to 15 percent, with the largest effects on light trucks. Ito and Sallee, studying Japan’s weight-based standards as a natural experiment in a 2017 Review of Economics and Statistics article, found about 10 percent of vehicles made heavier by roughly 110 kilograms to reach less stringent categories. The literature does not claim every upsizing decision traces to the curve: low fuel prices and consumer taste for roomier vehicles are independent drivers. But the incentive is written into the standard’s own arithmetic, and the measured response is real.

Q: Which agency sets CAFE standards?

The National Highway Traffic Safety Administration, the transportation safety regulator within the Department of Transportation, sets CAFE standards under the maximum-feasible mandate of 49 U.S.C. section 32902(a). The Environmental Protection Agency does not set the fuel-economy numbers, but it plays two essential roles: it measures each model’s fuel economy under its test procedures, and its Administrator computes each manufacturer’s fleet average under the harmonic-mean formula of section 32904(a)(1)(B). The division of labor dates to the 1975 design. A third actor enters through the Clean Air Act: the EPA sets separate greenhouse-gas emission standards for the same vehicles under section 202(a), following the Supreme Court’s 2007 Massachusetts v. EPA decision, and the two agencies coordinated their rulemakings for model years 2012 through 2025. California can add a further layer through the waiver mechanism, but the CAFE numbers themselves are the transportation regulator’s.

Q: Can California set stricter rules than CAFE standards?

California cannot set its own fuel-economy standards: 49 U.S.C. section 32919(a) expressly preempts state laws related to fuel economy for covered vehicles. What California can do, through a different statute, is set stricter greenhouse-gas emission standards for vehicles. Clean Air Act section 209(a) generally preempts state emission standards, but section 209(b) requires the federal environmental agency to waive that preemption for California, which had standards before March 30, 1966, unless the agency makes one of three adverse findings. California’s 2004 greenhouse-gas vehicle rules were the first climate application of the waiver. Other states may then adopt California’s waived standards identically, with at least two model years of lead time, under section 177. Two federal district courts held in 2007 that the state greenhouse-gas programs were not preempted by the fuel-economy law, though those rulings never became appellate precedent.

Q: What penalty applies if a maker misses CAFE standards?

A manufacturer that falls short after applying its credits faces a civil penalty under 49 U.S.C. section 32912(b): five dollars multiplied by each tenth of a mile per gallon by which the standard exceeds the manufacturer’s average, multiplied by the number of vehicles, reduced by available credits. The statute’s five-dollar base rate dates to 1975, but the operative rate through the article’s date wall was higher: the Secretary raised it to $5.50 per tenth of a mile per gallon per vehicle beginning with model year 1998 under section 32912(c) authority. In principle, a maker missing its standard by a full mile per gallon across a million vehicles, with no credits, would owe ten tenths times $5.50 times one million, or $55 million. In practice the credit system’s five-year carry-forward and three-year carry-back make outright penalty payments the last resort, and makers manage credit banks across product cycles to avoid them.

Q: Can automakers trade CAFE standards credits?

Yes. The Energy Independence and Security Act of 2007 rebuilt the flexibility provisions at 49 U.S.C. section 32903. A manufacturer that beats its standard earns credits it may carry forward for five consecutive model years or carry back for up to three model years, smoothing the lumpiness of product redesign cycles. Section 32903(f) authorizes the Secretary to establish credit trading among manufacturers, with the statutory condition that traded credits preserve total oil savings, so trades cannot be structured to lose gallons. Section 32903(g) authorizes transfers within a manufacturer’s own fleets, letting car-fleet surplus cover truck-fleet shortfalls. The domestic and imported passenger-car groups, which must each comply separately under section 32904(b), may share credits for accounting purposes. One hard limit applies: traded or transferred credits may not satisfy the minimum domestic-passenger-car standard of section 32902(b)(4), which must be met with actual domestic-car efficiency.

Q: How is the harmonic-mean fleet average calculated?

Section 32904(a)(1)(B) gives the formula in statutory prose: the EPA Administrator divides the number of passenger automobiles a manufacturer produced in the model year by the sum of the fractions obtained by dividing, for each model, the number of automobiles of that model by that model’s measured fuel economy. In symbols, the average equals N divided by the sum of n-sub-i over mpg-sub-i. It is sales-weighted because each model’s fraction scales with its sales volume, and harmonic because the averaging happens over the reciprocals, gallons per mile, rather than over miles per gallon. The effect is that a thirsty model drags the average down more than its sales share would suggest: two equal-volume models at 20 and 40 miles per gallon average 26.67, not 30. The formula concentrates each maker’s compliance incentive where the gallons are, at the inefficient end of the lineup.

Q: What were the original 1975 CAFE targets by model year?

For passenger cars, the 1975 statute set eighteen miles per gallon for model year 1978, nineteen for 1979, twenty for 1980, and twenty-seven and one-half for model year 1985 and thereafter. Congress directed the Secretary of Transportation to set the interim years, announced in June 1977: twenty-two for model year 1981, twenty-four for 1982, twenty-six for 1983, and twenty-seven for 1984. Light trucks started at seventeen and two-tenths miles per gallon in model year 1979. The regulator later lowered the car figure to twenty-six for model years 1986 through 1988 and twenty-six and one-half for 1989, returning to twenty-seven and one-half in 1990, where it sat until the footprint system. The eighteen to twenty-seven-and-one-half climb remains the steepest the program has demanded.

Q: What is the domestic versus imported fleet rule?

A passenger automobile counts as domestically manufactured when at least seventy-five percent of the cost to the maker is attributable to value added in the United States or Canada, under 49 U.S.C. section 32904(b)(1)(A), with an exception for Canadian-assembled cars imported more than thirty days after the model year ends. Section 32904(b)(2)(A) requires the EPA Administrator to calculate the two groups separately, as if made by different manufacturers, so a maker cannot average imported small cars with domestic large cars. The 2007 amendments added an exception for credit accounting under section 32903: the wall opens for credits but never for the compliance calculation itself. The rule was designed to keep the domestic product line improving on its own rather than free-riding on imports.

Q: Which vehicles count as light trucks under the classification rule?

The statute calls them non-passenger automobiles, defined at 49 U.S.C. section 32901(a)(17) as automobiles that are neither passenger automobiles nor work trucks, and the regulation at 49 CFR 523.5 decides which vehicles qualify. Large sport utility vehicles enter mainly through two prongs. Paragraph (a) covers truck-like function, including three rows of standard seating with a fold-flat load floor for model year 2008 and later. Paragraph (b) covers off-highway capability: four-wheel drive or a rating above six thousand pounds gross vehicle weight, plus at least four of five geometry measures, covering approach, breakover, and departure angles and running and axle clearances. Most large sport utility vehicles qualify on four-wheel drive or weight plus clearance, or on the seating prong.

Q: When did footprint-based CAFE standards begin?

The regulator introduced footprint-based standards for light trucks administratively in April 2006, at 71 Federal Register 17650, with an optional reformed track for model years 2008 through 2010 and a mandatory track for model year 2011. The Energy Independence and Security Act of 2007 then legislatively mandated attribute-based standards for both passenger cars and non-passenger automobiles at 49 U.S.C. section 32902(b)(3), effective with model year 2011. The statute did not invent the footprint from scratch; it codified and widened an administrative pilot. The joint federal rules for model years 2012 through 2016 and 2017 through 2025, coordinated with the environmental regulator’s greenhouse gas standards, are the mature expression of the footprint era.

Q: How did the 2007 energy law rewrite the fuel-economy program?

The Energy Independence and Security Act of 2007, Public Law 110-140, carried the fuel-economy title Congress called the Ten-in-Ten Fuel Economy Act. Section 32902(b)(3) mandated attribute-based standards for both passenger cars and non-passenger automobiles from model year 2011, expressed as a mathematical function of vehicle attributes, extending to cars the footprint approach the regulator had piloted for trucks in 2006. Section 32902(b)(2)(A) set a combined-fleet target of at least 35 miles per gallon for model year 2020, with ratable annual increases from 2011 through 2020. The act preserved the minimum domestic-passenger-car standard at section 32902(b)(4), added work trucks and medium- and heavy-duty vehicles at section 32902(k), extended credit carry-forward from three to five years, authorized credit trading and transfers, and kept the domestic-import compliance wall while opening it for credit accounting.

Q: How far forward and backward can CAFE credits be used?

Credits earned by exceeding a standard in a model year may be carried forward for up to five consecutive model years after the year earned, a window the 2007 law extended from the original three, and carried back to cover shortfalls in up to three model years before the year earned. The carry provisions let a manufacturer smooth the lumpiness of product cycles: a redesign year that overshoots its target can bank the surplus against weaker years ahead or cure deficits already incurred. Beyond banking and borrowing across time, the 2007 amendments authorized two ways to move credits across parties and fleets: trading with other manufacturers under section 32903(f), conditioned on preserving total oil savings, and transfers between a manufacturer’s own compliance categories under section 32903(g). Credits from trading or transfer cannot, however, satisfy the minimum domestic-passenger-car standard, which demands real domestic-car efficiency.

Q: What is the minimum domestic car standard?

Section 32902(b)(4) sets a separate floor for domestically manufactured passenger cars: the greater of twenty-seven and one-half miles per gallon or ninety-two percent of the Secretary’s projected combined domestic-plus-import fleet average for the model year. The twenty-seven and one-half prong preserves the 1975 number as an absolute backstop; the ninety-two percent prong ties the floor to the footprint era’s rising curves. The floor’s purpose descends from the two-fleet rule: the domestic car line must improve on its own and cannot free-ride on imported small cars. Neither traded nor transferred credits may satisfy it, making it the one obligation in the chapter that flexibility cannot buy down, enforceable only through measured performance.

Q: What does the fuel-economy preemption provision do?

Section 32919(a) provides that when an average fuel-economy standard under the chapter is in effect, no state or political subdivision may adopt or enforce a law or regulation related to fuel-economy standards or average fuel-economy standards for covered automobiles. It is express preemption of state fuel-economy law, with a carve-out at section 32919(c) letting states set standards for their own government fleets. The provision’s reach was tested in 2007 against state greenhouse-gas vehicle standards adopted through the Clean Air Act’s California waiver mechanism. In Green Mountain Chrysler v. Crombie, 508 F. Supp. 2d 295 (D. Vt. 2007), and Central Valley Chrysler-Jeep v. Goldstene, 529 F. Supp. 2d 1151 (E.D. Cal. 2007), two federal district courts held the state greenhouse-gas programs were not preempted, finding Congress intended the fuel-economy and emissions regimes to coexist. Both are trial-court rulings; the appeals were dismissed after a 2009 national harmonization agreement.

Q: What did the 2007 district-court preemption rulings decide?

Both 2007 rulings addressed whether state greenhouse-gas vehicle standards, adopted through the Clean Air Act’s California waiver mechanism, were preempted by the federal fuel-economy statute’s express preemption at 49 U.S.C. section 32919. In Green Mountain Chrysler Plymouth Dodge Jeep v. Crombie, 508 F. Supp. 2d 295 (D. Vt. 2007), Judge Sessions held Vermont’s adoption of California’s standards was not preempted, assuming the federal waiver issued, finding Congress aware of the fuel-economy and emissions overlap and intending coexistence. In Central Valley Chrysler-Jeep, Inc. v. Goldstene, 529 F. Supp. 2d 1151 (E.D. Cal. 2007), the Eastern District of California reached the same result for California’s own standards. Their precedential weight must be stated precisely: both are federal district-court decisions, and the appeals were dismissed after the 2009 national agreement harmonizing the federal and state programs, so neither became circuit precedent.

Q: Did new-vehicle efficiency improve under CAFE?

Per-vehicle efficiency improved substantially from the twelve and nine-tenths mile-per-gallon new-car fleet of 1974, as manufacturers reengineered engines, transmissions, and vehicle structures under the harmonic-mean incentive that concentrated compliance pressure on the thirstiest models. But the sales-weighted fleet average also reflects the mix shift toward light trucks, sport utility vehicles, and larger, heavier, more powerful vehicles across the 1990s and 2000s, which absorbed much of the per-vehicle gain. The mix shift has multiple causes, fuel prices, consumer preference, demographics, and the program’s own structure, so the program did not single-handedly cause it, though its gentler truck standard and classification rules accommodated it. The honest account separates the per-vehicle improvement, which was large, from the mix effect, which offset part of it.

Q: How do the fuel economy and greenhouse gas programs fit together?

They regulate the same tailpipe for different statutory purposes: the transportation safety regulator sets miles-per-gallon averages under 49 U.S.C. Chapter 329 to conserve energy, while the environmental regulator sets grams-of-carbon-dioxide-per-mile averages under Clean Air Act section 202(a), authority confirmed in Massachusetts v. EPA. Because a gallon of gasoline holds a fixed quantity of carbon, the two metrics are reciprocal descriptions of the same combustion, and the joint rules for model years 2012 through 2016 and 2017 through 2025 calibrated them to converge. The programs remain legally distinct, with different flexibilities, and the greenhouse gas side covers items like air-conditioning refrigerants that never enter the fuel economy math.