Tradable Performance Standards in the Transportation Sector

Several tradable performance standard (TPS) programs have recently been implemented in the US transportation sector: regulations for greenhouse gas emissions from passenger cars and trucks (national), zero-emission vehicle programs (10 states), the Renewable Fuel Standard (national), and low-carbon fuel standards (two states). The primary motivations are to promote innovation, to address consumers’ undervaluation of efficiency, and to reduce externalities, such as air pollution and the risks of dependence on foreign oil. A TPS sets a standard of technology performance but leaves technology choice to the producers; it increases the relative costs of technologies with undesirable performance characteristics and lowers the costs of technologies with desirable characteristics. We review the TPS programs and compare TPS with carbon pricing. Whereas carbon pricing creates incentives for both output reduction and technology change, TPS programs do not fully internalize the costs of emissions, resulting in lower price effects on products and raising the total cost of emissions reductions compared with carbon pricing. However, a TPS provides stronger incentives for upstream innovation and technology transformation. We show that TPS programs are generally additive to the effects of carbon pricing, so the policies can be combined without sacrificing the efficiency properties achieved by pricing. Given that the expected carbon price may be too low to substantially affect transportation demand or technology change, combining TPS with a carbon price may be necessary to drive innovation and achieve a sustained low-carbon transformation in the sector.


Introduction
Performance standards have a long history in the US transportation sector, beginning at the national level with the Corporate Average Fuel Economy (CAFE) standards enacted by Congress in 1975 (Greene 1990), and in California with the Zero Emission Vehicle (ZEV) program adopted by the California Air Resources Board (CARB) in 1990 (Collantes and Sperling 2008).Unlike technology mandates, which prescribe specific technology (e.g., a three-way catalytic converter), performance standards set a goal (e.g., maximum emissions per unit of performance, or vehicle efficiency of at least x miles per gallon), enabling the regulated entity to choose any technology mix that achieves that outcome (Bergek and Berggren 2014).Performance standards can force technology innovation by specifying a standard that cannot be met with conventional technology (Lee et al. 2010) or that imposes higher cost (Gerard and Lave 2005).When implemented as technology volume requirements (e.g., requiring that products with a specific technology constitute x percentage of all new sales), performance standards push an identified technology into the market, usually with the intent of achieving cost reductions through market penetration.When implemented as an emissions intensity standard, performance standards provide flexibility by enabling compliance through incremental improvements without specifying technology choice.Tradability or averaging of performance across facilities introduces additional flexibility and thus improves the cost-effectiveness of the policy.
TPS programs have been prominent in the US electricity sector, exemplified in 29 states' renewable portfolio standards and in the Obama administration's proposed Clean Power Plan.Recently, China announced its intent to implement the largest ratebased emissions trading program in the world in its electricity sector (Goulder and Morgenstern 2018).Some incentive-based regulatory programs address the transportation sector.In the bonus-malus (Latin for good-bad, here meaning credittax) system in France, consumers receive a rebate for low-emission vehicles but pay a fee for higher-emission vehicles.The European Union's Renewable Energy Directive mandates that clean sources account for minimum shares of the energy consumed in road and rail transport in member countries.
Very few programs in the transportation sector, however, allow companies to trade credits for compliance. 1 The most prominent examples have been in the United States: the national regulations for greenhouse gas (GHG) emissions from passenger cars and trucks (introduced in 2009, with first trading in 2012), zero-emission vehicle programs in 10 states (first adoption in 1990, first trading in 2012), the national Renewable Fuel Standard (introduced in 2005, first trading in 2010), and California's Low Carbon Fuel Standards (adopted in 2010, first trading in 2011).
Earlier attention to performance standards focused on their ability to promote innovation (Nentjes et al. 2007;Bergek and Berggren 2014;Klier and Linn 2016) and to address imperfect competition, consumers' undervaluation of energy efficiency (Greene 2019;Fischer 2010), consumers' inelasticity to fuel price changes (Hughes et al. 2008;Lin and Prince 2013), and externalities such as air pollution and the risks of dependence on foreign sources of oil (Lin and Prince 2009;Schnepf and Yacobucci 2013).In the absence of federal leadership in implementing ambitious carbon pricing, attention has shifted to the cost-effectiveness of TPS compared with carbon pricing (Nentjes and Woerdman 2012).It is often argued that without comprehensive, global carbon pricing, TPS programs must increasingly be part of emissions reduction policies if we are to control the total carbon concentration in the atmosphere (Sperling and Nichols 2012).
Broadly speaking, pollution can be abated either through changes in technology (inputs and production technologies) or through reduced consumption of the goods that embed emissions.Abatement of emissions through changes in technology is the main objective of TPS programs.In contrast, a price on carbon emissions provides incentives to change both technology (input substitution and abatement effects) and consumption levels or technology (the output effect) (Goulder et al. 1999).Emissions pricing puts an equal value on emissions reductions throughout the value chain; this generally means that it is the most efficient policy (Sterner 2007).TPS programs are more common than carbon pricing in the US transportation sector, for several reasons.The sector has some of the highest GHG mitigation costs of developed countries' economies (IPCC 2014;Creutzig et al. 2015;Yeh et al. 2017;Gillingham and Stock 2018), and therefore either very high fuel taxes or separate policy instruments other than carbon pricing are needed to achieve GHG reduction goals.Even with high fuel taxes in Europe, for example, emissions reduction targets still face significant challenges.Fuel tax increases have provoked strong backlashes, such as the "yellow vests" movement in France.In contrast, both politically and economically, the production subsidy to desired technologies implicit in the TPS approach (explained below) drives technological change while reducing the visible costs to consumers, compared with achieving the same outcome with carbon pricing.Nonmarket-based policy instruments such as regulations, technology mandates, and standards focus directly on technology change, but they may result in substantial differences in marginal abatement costs even within the same industry.Tradable standards are more economically efficient because they allow for equalization of marginal costs across technologies and firms.But unlike emissions pricing, they do not have an output effect: consumers do not bear the full cost of the pollution and do not have incentive to reduce consumption of polluting products.
Our paper begins by evaluating TPS programs, focusing on real-world implementation of two national programs and two state-level programs as they have been implemented in California (Section 2).We then compare TPS and carbon pricing approaches in terms of their formulations and price effects (Section 3) and the mitigation costs at the program and system levels (Section 4).We look specifically at the policies' effectiveness in reducing GHG emissions but ignore other important aspects, such as equity effects, that are reviewed elsewhere.In Section 5 we summarize the lessons learned and offer recommendations.

Review of Notable Programs
A performance standard can target performance (e.g., sales of electric vehicles, vehicle efficiency in miles per gallon) or emissions intensity (e.g., carbon intensity of fuels, emissions intensity of vehicles).The requirements typically become more stringent over time, often leading to higher credit prices and incentives to innovate.Compliance requirements can be placed upstream (on producers or suppliers) or downstream (on consumers).Often, the point of regulation is placed as far upstream as possible to reduce the regulatory burden for administration, reporting, and monitoring.Small producers below a certain threshold are frequently exempted.
TPS compliance options typically include (1) producing products that meet or exceed the standard, thereby generating credits that can be sold to other producers; (2) purchasing surplus credits from other producers; and (3) using banked credits or credits borrowed against future credits (if allowed).Monitoring, verification, and enforcement requirements are similar to those in other environmental markets, with the noncompliance penalty becoming the de facto ceiling for the credit price.If there is a cap on credit prices that is lower than the marginal compliance cost, then the standard will not be met (Greenstone and Nath 2019).
Most of the credit trading programs in the transportation sector are not large enough to support public trading platforms.Instead, trading takes place bilaterally, between companies.Reporting of credit prices may or may not be mandatory but reporting the number of credits sold or bought and the credit balance at the end of the compliance period is always mandatory.Credit prices that are not reported can sometimes be calculated from companies' annual financial reports.Some commodity trading companies that specialize in these markets also report credit prices as part of the market reports for current and potential clients.
Table 1 summarizes the major TPS programs in the US transportation sector.Two programs (regulations for GHG emissions from passenger cars and trucks; and zeroemission vehicles, ZEV) address vehicles.The other two (Renewable Fuel Standard, RFS; and low-carbon fuel standards, LCFS) address fuels.GHG regulations and LCFS are intensity standards, regulating the average emissions intensity of vehicles and fuels, respectively, whereas ZEV and RFS mandates are based on sales volumes, which makes them effectively intensity standards as well.ZEV, RFS, and LCFS all seek to push new, cleaner technology into the markets, whereas GHG regulations for cars and trucks aim to improve existing technology.When fuel economy or GHG standards are binding, they impose shadow prices on inefficient, higher-emitting vehicles in proportion to their deviation from the standard but subsidize fuel-efficient, low-GHG vehicles (Davis and Knittel 2019).In the case of the footprint standard, the shadow pricing is relative to the target for a vehicle of a given type and size (Liu and Greene 2014).Before trading between firms was allowed, shadow prices exhibited substantial heterogeneity across manufacturers because of differences in their market segments and their access to fuel-efficient technology (Jacobsen 2013;Anderson and Sallee 2011).This heterogeneity should incentivize between-firm trading.One analysis estimated that full trading across vehicle classes and between manufacturers could reduce the compliance costs of a 40 percent increase in fuel economy from 2012 to 2020 by more than 10 percent (Rubin et al. 2009).A small reduction in the benefits of trading, however, might result from oligopoly and oligopsony in credit markets.Another reason the full potential of trading might not be realized is that manufacturers may choose to pay fines if the cost of meeting the standards exceeds the fine.
Before footprint standards and credit trading, manufacturers of luxury imported vehicles, including Mercedes-Benz, BMW, Porsche, Volvo, Daimler Chrysler, and Jaguar Land Rover, accumulated more than $870 million in fines (nominal value) between 1983 and 2012 (Figure 1).Some companies-Fiat Chrysler, Jaguar Land Rover, and Volvo-continued to pay fines after 2012, when trading became available, but the number of companies paying fines fell significantly.The fines translate to less than $100 per vehicle except for luxury brands, such as Jaguar Land Rover ($200 per vehicle) and Ferrari Maserati, Saleen, and Spyker ($600 per vehicle).
During 2008-2015, the initial years of CAFE and GHG credit trading, nearly all firms were accumulating credits, and trades were very infrequent (Figure 2), suggesting that the standards were likely nonbinding or that firms were expecting compliance costs to rise (Bialek and Shrader 2019).Of the credits earned from 2012 to 2018, three firms (Honda, Tesla, and Toyota) accounted for 69 percent.The same three firms supplied 94 percent of the credits sold in 2018 (Figure 2).Because buyers and sellers arrange sales bilaterally, the program lacks transparency: the quantity of trades and prices, as well as possible side agreements, are not reported.Bialek and Shrader (2019) used banking behavior in CAFE to identify expectations of marginal abatement costs.Leard and McConnell (2017), using Tesla's 2020 Form 10-k filings with the Securities and Exchange Commission (SEC) for the sales of GHG credits, as well as the settlement between EPA and Hyundai and Kia for violating the standard, estimated credit prices at $36-$63/Mg (Table 2).Our own estimates using Tesla's filing data yielded lower estimates of $10-$18/Mg for 2017 and 2018 (see the following section).

Zero-Emission Vehicle (ZEV) Programs
The California Air Resources Board first adopted the ZEV requirement in 1990.
Currently there are nine states that have adopted California's ZEV regulations: Connecticut, Maine, Maryland, Massachusetts, New York, New Jersey, Oregon, Rhode Island and Vermont.Auto manufacturers are required to produce a certain number of ZEVs and plug-in hybrids each year, determined as a percentage of their total California sales.The percentage was 4.5 percent in 2018 and rises to 22 percent by 2025.Each vehicle receives one to four credits, based on its electric driving range.The more electric range a vehicle has, the more credits it receives.Auto manufacturers can also purchase credits to achieve compliance and bank credits for future use.
Manufacturers must comply at the end of each compliance period and can carry excess credits over to the next period.
The numerous ZEV credit categories and calculations reflect the several kinds of lowemission vehicles available: full battery-electric vehicles (BEVs), hydrogen fuel-cell vehicles, plug-in hybrid-electric transitional vehicles (TZEVs), partial ZEVs (PZEVs), and advanced technology (AT) PZEVs.Prior to 2015, the credit unit was called nonmethane organic gas mass emission (grams per mile), which was simplified to ZEV credits after 2015. Figure 3(a) shows ZEV credit trades for 2012-2018.Tesla has generated the most credits by far for sale to the other car companies (68 percent); Toyota is second (10 percent).5Toyota also bought the most credits (32 percent), followed by Fiat/Chrysler (16 percent), Honda (13 percent), and Ford (11 percent).ZEV credit sales have provided crucial financial support for Tesla, especially in the early years of the program.In general, Tesla vehicles earn about four ZEV credits per BEV sold, depending on the model (Forbes 2017).Credit sales constituted 135 percent6 of Tesla's gross profit (or about $17,000 per vehicle across all models) in 2012; the value decreased to 15 percent (or about $2,550 per vehicle across all models) in 2019 (Figure 4(b)).Overall, Tesla sold more than $1.05 billion in ZEV credits in 2009-2019, according to its SEC filings.Each ZEV credit could theoretically be worth up to $5,000 (the fine for noncompliance), although the market value is typically far less.Tesla's ZEV credit sales are estimated at around $1,000-$4,200 per ZEV credit except for 2013, when the value is estimated to have been close to $7,000 per ZEV credit.In 2013, Tesla realized approximately $28,000 in ZEV credit value on each sale of Model S (priced at $70,000 to $100,000 per vehicle).In addition to ZEV credit sales, Tesla also benefited from GHG credit sales (see Section 3.2), reported at $315 million in 2018 (Figure 4(a)).

Tesla Credit Sales Activity and Value
Notes: Shifting the overlapping years between these data will change the annual estimates but not the overall ranges nor trends.

US Renewable Fuel Standard (RFS)
The national RFS program requires that a certain volume of renewable fuels replace or reduce petroleum-based transportation fuel, heating oil, or jet fuel.The RFS was created in 2005 and expanded under the Energy Independence and Security Act of 2007, which increased the volume of renewable fuels to be blended into transportation fuel to 36 billion gallons by 2022.The four renewable fuel categories under the RFS are biomass-based diesel, cellulosic biofuel, advanced biofuel (biodiesel or sugarcane ethanol), and total renewable fuel (nonadvanced or conventional biofuel, such as corn ethanol).The maximum levels of life-cycle GHG emissions compared with baseline fuels (gasoline or diesel) are shown in Figure 5, where the height of the dots (right side) illustrates the maximum life-cycle carbon intensity of each fuel type.Obligated parties under the RFS program are refiners or importers of gasoline or diesel fuel.Compliance is achieved by blending renewable fuels into transportation fuel, or by obtaining credits to meet an EPA-specified renewable volume obligation.EPA calculates and establishes these obligations every year through rulemaking, based on the RFS volume requirements and projections of gasoline and diesel production for the coming year.The standards are converted into percentages, and obligated parties must demonstrate compliance annually.Credits are called renewable identification numbers (RINs).Each fuel type is assigned a D-code, which identifies the renewable fuel type based on the feedstock used, fuel type produced, energy inputs, and GHG reduction thresholds, among other requirements.The RFS program's four renewable fuel standards are nested within each other.That is, a fuel with a higher GHG percentage reduction can be used to meet the standards for a lower GHG percentage reduction, but not vice versa.For example, RINs for advanced biofuel (biodiesel or sugarcane ethanol) can be used to meet the total renewable fuel standards (corn ethanol).This has important implications for the price of RINs, discussed below.

Shares, Price Trends, and Emissions of Renewable Fuels
RFS has created incentives for production of corn ethanol (D6) and biodiesel or renewable diesel (D4) (Figure 5(a)).The D6 RIN price increased dramatically, from a few cents in 2012 to more than $1 in 2013, when in 2013 the gasoline fuel mix hit a "blend wall"-the maximum amount of ethanol (10 percent, or E10) that can be blended into regular gasoline without causing any risk or fear of engine damage in conventional vehicles.After the saturation of the E10 pool when the blend wall was hit, any additional volume of ethanol can only be blended as E85 (85 percent ethanol).
Because the use of E85 is limited to flexible-fuel (dual-fuel) vehicles with specialized engines and the sales volume of E85 is small, it was difficult to sell or blend more ethanol to generate D6 RINS, causing the D6 RIN price to rise in 2013 (Burkholder 2015).After 2013, RIN prices of corn ethanol (D6), biomass diesel (D4), and advanced biofuel (D5) started to converge because the nesting nature of RFS allows flexible compliance across fuel types, effectively lowering the RIN prices.Because cellulosic biofuel never materialized at scale, its RIN prices were significantly higher, but they dropped after 2018, when renewable natural gas (biogas) was included as a compliance option to be counted as cellulosic biofuel.

Low-Carbon Fuel Standards (LCFS)
California's LCFS, adopted in 2010, is the first major public initiative to codify life-cycle concepts into law (Sperling and Yeh 2009;Yeh et al. 2016).The same policy was adopted by Oregon in 2016 as the Clean Fuels Program.The carbon intensity (CI) of fuels is measured on a life-cycle basis-that is, emissions from extraction or cultivation of feedstock, production, transportation, and use of the fuel are included.
The legislation calls for at least a 10 percent reduction in life-cycle GHG emissions per unit of energy (gCO2e/MJ)) by 2020 and 20 percent by 2030.Oil refiners can sell lowcarbon fuels or buy credits generated by low-carbon fuel producers, such as biofuels producers or electric utilities that sell power to electric vehicles.
The alternative fuels for compliance are largely biofuels from corn ethanol, biodiesel, and renewable diesel (Figure 6

Observations across Programs
Our review of TPS programs indicates that the policies have succeeded in providing flexibility for compliance: companies have pursued different strategies to meet the standards, including selling or purchasing credits and banking.We found substantial amounts of early banking for some programs, since companies expected costs to increase as the standards became more stringent, consistent with studies suggesting that banking can both smooth out and lower compliance costs for companies (Rubin and Leiby 2013;Bialek and Shrader 2019).Program transparency varies.In RFS and LCFS, fuel and commodities associations publish (unofficial) weekly credit prices that inform their members or customers.In the CAFE and ZEV programs, companies are reluctant to report prices and regulators are reluctant to require reporting because it can reveal commercial information; however, the market is small, major players are visible, and in the ZEV program, direct transfer of credits between two parties is reported.
The observed credit prices of TPS programs reflect the stringency level of standards and other factors mentioned above.Credit prices, however, are not directly comparable with a price on carbon implemented through an emissions fee or cap-andtrade, or with each other, given that the primary objective of these programs is encouraging innovation and system transition.Viewed in terms of cost per ton of avoided carbon emissions (Section 4.1), observed credit prices are higher than in any carbon pricing program globally, but the levels are anticipated to be temporary, and technology benefits are expected to spill over to transportation markets in other countries.The programs simultaneously address other objectives, including energy security, local air quality, and (in the case of US ethanol) agricultural support, and they may yield savings due to the energy efficiency paradox.Nevertheless, given the importance of GHG emissions reductions, below we compare and illustrate the difference between TPS and carbon pricing at the conceptual level (Section 3), and then look at empirical studies that examine the (cost-)effectiveness of these two approaches individually and at the system level (Section 4).

Performance Standards versus Carbon Pricing
In this section, we formalize the description of the two policy instruments to show how they affect the relative prices of technologies, innovation, and efficiency.This will also allow us to compare tradable performance standards with conventional carbon pricing.
We characterize the change in relative product prices7 resulting from technology choice under regulation as driving market share change and providing an incentive to innovate.We illustrate this incentive with an example of a carbon price compared with a performance standard resembling the LCFS.Below we compare the change in fuel prices under carbon pricing versus a fuel intensity standard.Because ethanol has less energy content than gasoline, we do all the accounting for ethanol in liters of gasoline equivalent.The equivalency is established by measuring each fuel in energy content (megajoules, MJ) rather than in liters or some other measure that would not be comparable.

Effect of Carbon Pricing on Fuel Prices
The effect of a carbon price, implemented as a carbon tax or through cap-and-trade, is to add a cost component corresponding to the carbon emissions 9 (emissions multiplied by the carbon price level) to the fuel price. 10The emissions from ethanol are typically lower than emissions from gasoline, and hence the price of ethanol rises less than the price of gasoline.
Under a carbon price, described henceforth as a tax, the increment to the price per liter of gasoline depends on the price of carbon and the total amount of carbon (we can view this as the intensity of the fuel relative to zero): The change in the price of ethanol per liter of gasoline equivalent is also positive, but smaller 11 : The change in the relative prices of the fuels, given by the difference in their carbon intensities, provides an incentive for the use of ethanol.We call this the incentive margin (IM) for producing a liter of alternative fuel under a carbon tax.This margin is given by Equation ( 3): The purpose of this is to internalize the economic cost of the estimated emissions.In the capand-trade example, we assume the carbon price is applied over the full life cycle, though in real policy applications this is not the case.The LCFS accounts for emissions across the entire life cycle for each fuel.These two approaches reflect the system boundaries of the two policies and have significant implications in terms of leakage.See DeCicco (2012) for a discussion of the trade-offs. 10Here we have assumed the conventional structure of a tax: the user pays a tax on each unit of emissions.This need not be the case.A different outcome would result if the policymaker allowed a certain amount of pollution for free and charged only for the excess pollution (Pezzey and Jotzo 2013).This possibility is rarely (if ever) used, but it would make the tax similar in some respects to a tradable intensity standard.The same result as for a tax would be achieved in a carbon trading scheme where allowances were distributed using output-based allocation of emissions allowances (Fischer 2019), a practice that is observed for a portion of the market in the EU, California, and Quebec trading programs.
11 It is often the case, however, that biofuels are treated as "carbon neutral" under most capand-trade programs and therefore do not pay carbon taxes.See Searchinger et al. (2009).

Effect of Tradable Intensity Standards
Tradable intensity standards and taxes have different effects.A tax raises the prices of both fuels, but it does so more for gasoline than for ethanol.An intensity standard requires that fuels have an average emissions intensity of CS.Producers of fuel with higher emissions intensity need to buy credits, and producers of fuel with lower emissions intensity earn credits that can be sold.In effect, credit payments by gasoline serve as a production incentive for ethanol.The average price of all fuels is lower under an intensity standard than with revenue-raising carbon taxation because payments are made and received within the sector; no revenue leaves the sector to go to government.This is good news for motorists (and may explain why this instrument is often favored by policymakers).However, it also means that the incentive to economize on miles driven will be weaker-this is the inefficiency of the system.
The price change for gasoline under the performance standard is given by Equation ( 4): The price change for ethanol is given by Equation ( 5), where we see clearly that the price of ethanol falls because Ce < Cs: We calculate the incentive margin for producing a liter of alternative fuel under a tradable intensity standard as the difference in prices for gasoline and ethanol: Note that IMS and IMT are of the same functional form and would actually be identical for the case that PS = PT.It is also interesting that CS is absent in Equation ( 6); that is, the incentive margin of the tradable intensity standard IMS is only indirectly related to the level of standard through the credit prices.The more stringent the intensity standard, the higher the credit price, and therefore the higher the incentive margin.
We illustrate in Figure 8 that the incentive margins of the two programs IMS and IMT are identical when PS = PT.Under a carbon tax, however, both high-carbon fuel (dirty technology, DT, such as gasoline) and low-carbon fuel (clean technology, CT, such as ethanol) are penalized by higher prices, whereas under the tradable intensity standard, the low-carbon fuel receives a "reward" instead.Importantly, when PS = PT, the levels of changes in fuel prices (both positive and negative) are much smaller under an intensity standard, since the price change is only the credit price multiplied by the difference between the standard and the carbon intensity of the fuel (Cg -CS and Ce -CS ), whereas the price change under a carbon tax is the credit price multiplied by the full carbon intensity of the fuel or technology (Cg and Ce).

Price Change of Dirty and Clean Technology under Carbon Tax and Tradable Intensity Standard
Notes: Price change of two products (a dirty technology, DT, and a clean technology, CT, with equivalent energy content) under a carbon price (T) or credit price under an intensity standard (S), and the incentive margins of a carbon tax (IMT) vs. a performance standard (IMS), when: (a) for an equivalent carbon and standard price level, PT = PS = P; and (b) for a range of P = $0-$200.A negative value means a technology receives subsidies under the program, and a positive value means the program increases the price of a technology.For reference, the carbon price in California was around $17/tCO2e and LCFS credit price was around $200/tCO2e in the first half of 2020.The barely visible red line at $17 and the dashed red line at $200 correspond to the IMT and IMs given the observed carbon and LCFS credit prices in California.Though the figures are intended to be illustrative (therefore the vertical axes are not labeled), the figures are scaled realistically, using realistic assumptions of gasoline CI = 100 gCO2e, ethanol CI = 60 gCO2e with the standard set at 90 gCO2e (10 percent below gasoline CI).
As the carbon tax and credit price go up, the change in cost also goes up linearly, and IMS = IMT is still the same at any equivalent carbon and credit price combination.As we discuss in Section 4, both programs are in fact implemented simultaneously in California, where the credit price and the incentive to switch from dirty to clean fuel is about 10 times greater under the tradable performance standard than under the carbon price (PS = 10×PT), but the change in the product price viewed by consumers attributable to each program is about equal (around 20 cents per gallon) and additive.

Policy Guidance
Here, we make two important observations.First, the cross subsidy expressed as IM (included implicitly in a carbon tax program or explicitly in a performance standard program) is a function solely of the carbon tax or credit price across both programs.Empirically, we observe that carbon tax programs tend to have lower values per ton carbon, and therefore weaker incentives for technology switching, than TPS programs.
Second, because emissions pricing puts an equal value on emissions reductions through changes in technology or through reduced consumption of goods that embed emissions, it is equivalent to a coupled performance standard program and consumption tax, where the credit and tax prices have a fixed relationship with each another.Because a performance standard focuses on producers and a consumption tax affects consumers, policymakers wishing to increase the effect of technology switching by producers could consider decoupling these two aspects of a carbon tax.Although any deviation away from a pure carbon tax would be less efficient economically, a decoupling may foster greater technology innovation by producers without transmitting an equivalent change in product prices to consumers.This may help maintain the industry's overall market share amid international competition during a transition until technical, political, and societal factors allow for greater levels of a consumption tax.

Effects on GHG Emissions Reductions
As emphasized previously, the most important objective of TPS is not GHG emissions reductions but other societal benefits-energy transition, energy efficiency, clean air, energy security, agricultural jobs-that lower the cost of clean(er) technology over time.The cost estimates also differ, depending on a static versus dynamic view of the program and the system-level interactions that are taken into account.For example, the GHG reduction from one ZEV credit depends on what vehicle the alternative vehicle replaces, the system-level effects of the program such as emission leakages and rebound, and the effects of complementary or overlapping policies (Mansur et al. 2016).In this section, we evaluate the cost-effectiveness of individual TPS programs in terms of the empirically observed carbon abatement cost ($/ton CO2e abated) (Section 4.1).We review studies that examine the effects of the programs in isolation (Section 4.2), the interactions of TPS programs (Section 4.3), and the effects of a TPS program within a cap-and-trade program (Section 4.3).

Carbon Mitigation Costs
Given the various policy targets, the units of TPS credits vary.For example, GHG regulations for vehicles have a credit price of $63/MgCO2e12 , the LCFS credit price is around $200 per credit (tCO2e), RFS RIN credits range from $0.02 to $2.24 per RIN (equivalent value of biofuel gallon), and ZEV credits range from $1,000 to $7,000 per credit (Table 1).The GHG regulations for vehicles credit price of $63/Mg ($63/tCO2e) and the LCFS credit price of $200/mtCO2e can both be taken directly as the marginal mitigation cost of GHG abatement.At $1,000-$4,000 per ZEV credit, the marginal mitigation cost of ZEV is $82-$320/tCO2e. 13A RIN price of 50 cents per gallon implies that the mitigation cost of biodiesel is $74/tCO2e.14These illustrative estimates are consistent with the static cost estimates reviewed in Gillingham and Stock (2018).

Effects of TPS Programs in Isolation
Several studies find that CAFE has altered manufacturers' vehicle offerings and consumers' purchase decisions (Greene 1998;Michalek et al. 2005;Fischer 2010).Various studies identify an emissions rebound from increased consumption because efficiency improvements lower the cost of travel (Ross Morrow et al. 2010), though the size of rebound is debated in the literature (Small and Van Dender 2007;Hymel et al. 2010;Greene 2012;Hymel and Small 2015;Dimitropoulos et al. 2018).Nevertheless, the fuel economy standards' reduction in GHG emissions has been substantial.Greene et al. (2020) estimated that efficiency improvements to US light-duty vehicles reduced GHG emissions by 17 billion metric tons; the standards were identified as responsible for more than 80 percent of the improvements, and fuel prices changes, less than 20 percent (partly because of low fuel prices in the United States).However, the new footprint-based vehicle GHG standard could trigger a rebound effect stemming from the distorted incentives toward larger vehicles for manufacturers and consumers (Ito and Sallee 2017).
Similarly, studies find that the ZEV program has altered manufacturers' ZEV production (Collantes and Sperling 2008;Wesseling et al. 2014Wesseling et al. , 2015;;Jenn et al. 2019).Using data on patents, sales, and political activity, studies found that manufacturers also chose political strategies that evolved over time from value maintenance (minimize research and development, oppose the policy) to value creation (invest heavily in research and development to occupy the new market, proactively attempt to influence the policy) (Wesseling et al. 2014(Wesseling et al. , 2015)).
Many studies have looked at the effects of RFS (de Gorter and Just 2010; Schnepf and Yacobucci 2013;NRC 2011;Farzad and Tyner 2014) and LCFS (Yeh et al. 2009;Holland et al. 2009;Yeh and Sperling 2010;Huang et al. 2013).The policies incentivized large amounts of grain-based biofuels from corn and soybean, but both fell short of incentivizing very low carbon biofuels, such as biofuels from cellulosic biomass or advanced technologies like "drop-in" biofuels.However, consistent with the LCFS incentives, we observe under the LCFS compliance pathways an evolution toward reducing the life-cycle emissions of biofuels and increasing the contribution from nonbio-based alternative fuels (Figure 6).The more complicated questions are the net GHG emissions reductions and the interactions of the two policies.Two controversies surround the effects GHG reductions: (1) increased fuel consumption and incomplete petroleum displacement due to the (global) fuel market rebound effect (Rajagopal et al. 2011;Hill et al. 2016), and (2) indirect land-use change due to leakage when increased demand for crop-based biofuels leads to cropland expansion and forest clearance that increase GHG emissions (Tilman et al. 2009;NRC 2011).

Interactive Effects: Complementary, Overlapping, or Sequencing?
The success of performance standards critically depends on their interaction with other policy packages.This interaction is complementary in addressing long-term, large-scale energy transitions, in several ways.For instance, a ZEV policy does not stand on its own in promoting clean transportation technology: other policies provide subsidies directly to consumers (Sen et al. 2017;Münzel et al. 2019), subsidize investment in charging stations (Peterson and Michalek 2013) and hydrogen refueling stations for electric vehicles (Ogden 1999), encourage changes to laws and regulations on station design and siting (among other institutional changes), inform potential vehicle purchasers and reduce risk aversion to new technologies, subsidize R&D, and seek to lower electricity emissions.Additionally, the more successful ZEV and RFS are, the easier it is to achieve the LCFS because the adoption of low-carbon vehicles and fuels expands and enhances the compliance options.All these programs promote the adoption of clean vehicles and fuels, but LCFS provides additional incentives to lower the emissions intensity of alternative fuels.
However, a combination of policies can also lead to undesirable consequences.For example, Jenn et al. (2019) simulate the interactive effect of CAFE and ZEV policies and find that the combined policies produce higher GHG emissions than either policy alone.This is because the state mandates increased ZEV-like sales (battery electric, plug-in hybrid, and flex-fuel vehicles) in the presence of federal incentives that relax the fleet GHG standard when ZEV-like vehicles are sold.Complementarity requires, preferably, automatic program adjustments and reviews of policy targets given interactions with other policies.

TPS Programs Combined with Cap-and-Trade
In California, cap-and-trade (CAT) is implemented alongside other regulations and policies to reduce overall emissions (Sperling and Nichols 2012).California's CAT Phase I compliance period (effective January 1, 2013) placed a cap on GHG emissions associated with electricity consumption (for electricity both generated in the state and imported) and large industrial sources in the state, including refineries.Starting January 1, 2015, the CAT policy was expanded to include GHG emissions from on-road transportation fuels, covering gasoline, diesel, and natural gas but exempting carbon emissions from biofuels.The CAT and the TPS programs are designed to address separate challenges in achieving the state's comprehensive climate goal.One challenge is the CAT's carbon price, which is insufficiently high to achieve the kind of rapid technological innovation necessary for energy transformation and deep GHG emissions reductions in an ambitious timeline.
The policies could also have other interactions in terms of compliance and effects on both regulated parties and consumers.California's regulated parties have several compliance obligations (Table 1) in addition to CAT.For vehicle manufacturers and importers, regulations for GHG emissions from passenger cars and trucks and ZEV are additive ("stackable"), and meeting the ZEV program will help with meeting the regulations for GHG emissions, but not vice versa (Sen et al. 2017).As shown in Figure 4, Tesla profited from selling regulatory credits, including both ZEV credits and GHG credits.Similarly, for fuel producers and importers, meeting the RFS will help with meeting the LCFS targets.Renewable fuel producers and providers will receive credits from both RFS and LCFS, and compliance will reduce the obligations under the CAT if the obligated party is a fossil fuel provider.
As an example, Figure 9 uses realistic assumptions in California to illustrate the credit values per gallon for two representative fuel types, gasoline and very low carbon biofuel, and the additive effects of LCFS, CAT, and RFS (the example works similarly for diesel fuel and its substitutes).The incentive margin of the fuel market is the sum of the penalties on fossil fuels plus incentives for biofuel.The additive effect remains large, given the rising penalties on gasoline (and diesel) despite the shrinking values of the incentives for biofuel over time.The combined effect is substantially greater than the incentive margin provided by carbon pricing from the CAT alone.For example, in 2020 biofuel received a net subsidy of $3.0/GGE, and gasoline incurred a penalty of $0.37 per gallon.Under the three programs, the net cost per average gallon of gasoline (E10) bought by consumers was a net subsidy of $0.1 per gallon in 2020.In 2030 the subsidy to biofuel of the same carbon intensity falls to $2.7/GGE (assuming the same credit price) and the penalty on gasoline increases to $0.93 per gallon, for a net cost of $0.2 per gallon of blended gasoline to consumers.Many observers argue for complementary CAT and regulatory policies in the presence of imperfect markets (Bird et al. 2011) or learning spillovers (Fischer and Preonas 2012;Lehmann 2013).For example, energy-efficient technologies, a critical component of the transition to sustainable energy (GEA 2012), are often hindered by the energy efficiency paradox, a behavioral issue.Whether consumers undervalue the future savings from energy efficiency improvements (termed "internalities" by Allcott et al. 2014) and if so, to what extent, is a subject of ongoing debate and research (Gillingham and Palmer 2014;Gerarden et al. 2015).If behavioral issues are prevalent, then pricing inefficiency through a combination of energy efficiency incentives, along with a direct tax or shadow price induced by an equivalent TPS, may be a necessary component of an economically efficient policy solution (Allcott et al. 2014;Heutel 2015).

Additive Effects of Policies in
The variation in marginal costs and other incentives across these policies, however, creates inefficiency and, inevitably, forgone opportunities for emissions reductions that could be remedied through carbon pricing.In the short run, technology forcing programs can have the effect of lowering the effectiveness of carbon pricing under CAT programs through the "waterbed effect" by lowering the demand for allowances, thereby lowering their market price and the induced incentive for innovation in the carbon market (Abrell and Weigt 2011;Tsao et al. 2011;Fischer and Preonas 2012;Nelson et al. 2015).In the California CAT program, the minimum auction price provides an effective price floor that converts reduced demand for allowances into a reduced supply of allowances and reduced emissions.In the long run, innovation-driven policies in the transportation sector can be viewed as a policy sequence that provides an on-ramp to greater stringency and efficiency through expanded carbon pricing (Meckling et al. 2017;Pahle et al. 2018).

Conclusions
Tradable performance standards are technology requirements or emissions intensity standards that allow the trading of compliance credits across companies.Unlike pollution pricing, they do not fully internalize the costs of emissions and thus they raise the total cost of emissions reductions compared with pollution pricing.However, they provide incentives for upstream innovation and technology transformation that are greater per dollar change in product prices than carbon pricing and are generally additive to the effects of carbon pricing.That is, the policies can be combined without sacrificing the efficiency properties achieved by pricing.
TPS policies have special appeal in the context of achieving deep decarbonization because of the crucial role of innovation.Reducing GHG emissions enough to limit global warming to 1.5° or 2.0°C requires a global transition from fossil energy.This transition could in principle be achieved with carbon pricing, but the complexity, inherent uncertainty, and systemic nature of the climate challenge would necessitate unrealistically high carbon prices (e.g., Rosenbloom et al. 2020).Comprehensive strategies addressing institutional change, network effects, tipping points, and behavioral issues also appear unreachable except with very high carbon prices.The transition will take decades, so it requires policies that will enjoy sustained public support.TPS policies thus present an attractive alternative, particularly in jurisdictions with strong resistance to fuel taxes (e.g., Heutel 2020).Moreover, the carbon pricing and TPS strategies are not mutually incompatible.Two observations provide guidance for policy: First, the incentive margin-whether included implicitly in a carbon tax program or explicitly as a cross-subsidy among technologies in a TPS program-is a function solely of the carbon and credit prices across both policies.We observe empirically that carbon pricing programs tend to have lower prices, and therefore weaker incentives for technology switching, than existing TPS programs.Even with explicitly high fuel taxes in Europe and Japan, the incentives for technology change are still insufficient.In California, where carbon pricing may have approximately the same effect on consumer prices as a TPS, the incentive margin for technology innovation under the TPS is 10-fold greater.
Second, when combined with a carbon tax, TPS both achieves a high cross-subsidy to incentivize innovation and provides moderate output effects.TPS policies are intended to function as part of a comprehensive strategy to address long-term, largescale energy transitions.Their success therefore critically depends on the other, complementary policies, including carbon pricing, and care must be taken to ensure that overlapping policies work as intended without creating unintended consequences.With these conditions firmly in place, TPS policies will be, and are now, an important contributor to global decarbonization.
, Deficits, Purchases, and Sales, by Manufacturer, 2009-2018 Notes: (a) Total EPA GHG credits (+) and deficits (-) by company, 2009-2018.2009-2011 are early crediting years.For example, Fiat Chrysler (FCA) accumulated deficits and also purchased credits for compliance whereas Toyota accumulated net credits and sold some credits.Smaller companies are omitted.Source: US EPA (2020).(b) Light-duty vehicle GHG emission credit sales (<0) and purchases (>0) by manufacturer by model year.Manufacturers not shown sold or purchased few or no credits during the 2012-2018 period.

California
ZEV Credits, Deficits, and Credit Balances, by ManufacturerNotes: (a) Total ZEV credits bought (-) and sold (+) by manufacturer, 2012-2018.Credits from 2012-2014 were divided by 0.035 to convert grams/mile nonmethane organic gas credits to ZEV credits.(b) ZEV credit balances as of August 31, 2018.Smaller companies are omitted.Source: CARB (2020b).Like other transportation sector TPS programs, the market for the ZEV program is closed (i.e., trading is allowed only among regulated parties), and credits are transferred directly between two companies or through other arrangements, without going through a public marketplace.Credit transactions must be reported, but not credit prices.In 2016 Tesla sold ZEV credits to Toyota and in 2017 to Toyota, Fiat Chrysler, and Subaru, and in 2017 and 2018 Fiat Chrysler sold ZEV credits to Honda.Companies also bank a significant amount of credits for future use or sales.On August 31, 2018, for example, Toyota held about as many ZEV credits in the bank as the total credits sold by Tesla in all previous years (Figure3(b)).
(a) Tesla's annual regulatory credit sales, 2009-2019.Credit sales data for 2019 (blue bar) were not reported separately as ZEV vs. GHG credit.(b) The annual gross profit from credit sales in percentage and in dollars per vehicle ($/veh) (right vertical axis), 2012-2018.Sources: Tesla's annual SEC Form 10-k filings and investor reports; authors' own estimates.The estimates are approximate because the cutoff dates are different for sales ($ millions, January 1 to December 31) and ZEV credit generation (September 1 to August 31 the following year).For example, no ZEV credit sales were reported to CARB in 2018 (September 1, 2018, to August 31, 2019) but $103.4 million was reported for January 1-December 31, 2018) to the SEC.
(a) Renewable identification number (RIN, or credit) generated by fuel type, 2010-2019, measured as equivalence value (EV).The EV of a renewable fuel represents the number of gallons that can be claimed for compliance purposes for every physical gallon of renewable fuel used, and it is generally the ratio of the energy content of a gallon of the fuel to a gallon of ethanol.Biodiesel EV = 1.5.(b) RIN prices by fuel type, 2010-2019, and maximum life-cycle GHG emissions relative to gasoline or diesel for each fuel category.Source: US EPA (2019).
(a)) because these fuels have the lowest compliance costs and are compatible with existing vehicle technologies.In contrast with the RFS, however, LCFS allows other fuel types, including electricity, and has created strong incentives for fuel producers to lower the CI values.As ethanol hit the blend wall in 2013 (Section 2.3), ethanol's volume stayed flat (Figure6(a)) but continued generating more LCFS credits (Figure6(b)) because of higher production efficiency (e.g., more output per biomass, new technologies like corn oil extraction), the use of lower-carbon energy sources as inputs in the production processes, and a switch to biomass feedstock with lower carbon emissions, such as crop residues, used cooking oil, and wastes from food processing.As a result, the CI of ethanol and biodiesel across their life cycle has decreased by 33 and 41 percent, respectively (Figure6(b), right vertical axis).Also shown are the volume-weighted average life-cycle CI of fuels and total alternative fuels in California (Figure 6(b), right vertical axis).The CI values of ethanol and biodiesel have drastically decreased over time (shown as arrows) while the CI of other fuels has remained mostly unchanged (shown as dots with their 2019Q4 values).Alternative Fuel Volumes and California LCFS Credits, by Fuel TypeNotes: "Other" includes hydrogen, renewable naphtha, and propane.Ethanol, electricity, and other fuel volumes are in gasoline gallon equivalent (GGE); natural gas, biomethane, biodiesel and renewable diesel are in diesel gallon equivalent (DGE).Sources: CARB (2020a); pers.commun., J. Witcover, Institute of Transportation Studies, University of California, Davis, May 23 2020.The price fluctuations earlier in the program were due in large part to policy uncertainty, including legal challenges to the program(Tracy 2010).Initially, regulated parties earned more credits than deficits, creating a huge surplus of banked credits, because compliance could be largely achieved with existing fuel technologies (Figure7(a)).As the standard became increasingly stringent, regulated parties started generating more deficits than credits and in the second half of 2017 began drawing down the credit bank.Also in that year, expectations firmed up regarding the extension of the LCFS through 2030, signaling the program's durability and increasing stringency.These factors resulted in higher credit prices after 2017 (Figure7(b)).California LCFS Credits, Deficits, Sales, and Price Trends, 2011-2019 (a) Total LCFS credits and deficits reported by year and cumulative credit bank, 2011Q1-2019Q1.Sources: Witcover (2018); CARB (2020a).(b) Volumes of LCFS credit transactions (left axis) and LCFS credit price (right axis), 2013-2019.Source: CARB (2020a).
California's Fuel Market, 2017-2030 Notes: The effects on gasoline are shown in black (CAT) and gray (LCFS), and the effects on low-carbon biofuels are shown in dark green (RFS) and light green (LCFS).Positive values imply that a product receives credits, whereas negative values imply that a product incurs penalties.We use the realistic assumptions of credit prices described in Section 4: LCFS $200/credit, CAT carbon price from $15/tCO2e in 2017 to $50/tCO2e in 2030, cellulosic biofuels RIN historical values (2017-2020) and $1.8/gal after 2020.The carbon intensity of biofuel = 40 gCO2e/MJ.GGE means gasoline gallon equivalent.Gasoline has on average 119.5 MJ/gallon, and ethanol = 81.5 MJ/gallon.