ZEV infrastructure credits are capped at 5% of the prior quarter’s deficit generation – 2.5% for hydrogen fueling and 2.5% for DC fast charging equipment. Applications for ZEV infrastructure credits are open through 2025 and are valid for 15 years in the case of hydrogen infrastructure, and 5 years in the case of DC fast charging infrastructure. On one level, the addition of infrastructure credits represents a major departure from the original design of the LCFS as it does not directly subsidize the consumption of a low carbon .fuel. Rather, the credits subsidize a fixed cost of providing network infrastructure that may encourage adoption of EVs, the technology which may in turn use a low carbon fuel. In the same way, however, the infrastructure credit can reduce the very effect that LCFS critics have focused on as the central flaw in the regulations design: the encouragement of low, but still nonzero carbon fuel. While infrastructure credits may spur vehicle adoption, their effect on expanding driving miles would be second order. At the same time, if the amount of infrastructure credits awarded through the program were significant enough to ease compliance, these credits can have the effect of lowering the overall LCFS credit price, and therefore reduce even the diluted carbon price effect on end-use fuel prices. The magnitude of any price-suppression effect would depend upon both the quantity of infrastructure credits and the slope of the LCFS compliance cost curve.Initially, there were no formal limits on how high LCFS credit prices could rise, although legal challenges to the regulation effectively delayed implementation,cannabis grow equipment freezing the standard from 2013 through 2015, and effectively limited demand for credits and their pass-through to fuel prices.
However, as the lawsuits were resolved in favor of continued implementation of the LCFS and the standard declined steadily in the last several years , credit prices have risen steadily and raised increasing concerns about the cost of the regulation.32 In its 2015 re-adoption rule, the ARB introduced the credit clearance market, which is a cost-containment mechanism that would in theory limit price increases under some scenarios.Entities in need of LCFS credits for purposes of immediate compliance can purchase credits in the credit clearance market at a price no higher than the prescribed maximum of $200 per ton in 2016 and adjusted for inflation thereafter . If these entities are unable to purchase sufficient credits in this market to reach compliance, then they may carry over their deficits to future periods. Carryover deficits grow by 5% per year, meaning that firms pay an ‘interest’ penalty for deferring compliance. However, firms that hold credits are not required to sell in the credit clearance market, and they would not do so if they believed that they be able to sell their credits at a higher price in the future. Thus, the credit clearance market provides only a soft cap. However, ARB is currently proposing to impose a hard price cap of $200 per ton in 2016 dollars for LCFS credit transactions. To help facilitate compliance under this cap, it proposes a mechanism to ‘borrow credits’ from future residential electric vehicle charging. Under this mechanism, obligated entities could use credits expected to be generated in 2026- 2030 to meet unmet annual deficit obligations in 2020 – 2025. These cost-containment mechanisms are suited for dealing with a transient disruption in clean fuel supply or some other cause of a short-term supply-demand imbalance of LCFS credits. Because of the requirement that borrowed credits be restored with interest, it will not be effective at containing costs in an environment of chronic, long-term credit supply demand imbalance.
The future prospects of the regulation are therefore linked to the potential supply and demand balance through the next 11 years of the program. A circumstance where compliance is only feasible through high cost fuels or sharp reductions in fuel consumption would push credit prices above the maximum credit price for the credit clearance market. One objective of this paper is to assess the potential likelihood of such an outcome. In 2019, ARB is proposing amendments that would backstop this cost containment mechanism, enforcing additional borrowing of future credit generation from residential electricity charging for electric vehicles at the maximum credit price, with a rolling payback schedule enforced on utilities that will borrow the credits, up to a cumulative total of 10 million borrowed credits.This section outlines data and methods used to project business-as-usual for LCFS credit and deficit generation to 2030. In this paper we use the term business-as-usual frequently, and take it to mean, regarding LCFS credit demand, the continuation of historical trends through the compliance period. For LCFS credit supply, BAU refers to a continuation of current alternative fuel mix trends to 2030. Therefore, the uncertainty in the projections stems from the estimation of BAU demand, which against an assumed steady state of supply, yields a distribution of net deficits accumulate over the period 2019 to 2030, on which we base subsequent analysis.We are interested forecasting demand for fuel and vehicle miles under BAU economic conditions. Demand for fuel and vehicle miles are highly dependent on other economic variables. Demand for both fuel and vehicle miles will be influenced by general economic activity and oil prices. In a booming economy, consumers travel more and purchase more fuel.
Our aim is to fit an econometric model that characterizes past trends in key credit demand variables such as fuel consumption and key input prices for the gasoline and diesel fuel “pools,” namely oil price and soybean prices, vehicle miles traveled, and an indicator of the state economy.The estimates from that model are then used to simulate relationships moving forward to project potential credit demand.We use data available from 1987 to 2018 for the six dependent variables to fit the VEC model. Because our data are measured at the quarterly level, we have a total of 124 observations for each variable.California GSP was collected from the Bureau of Economic Analysis .Theoil prices used in our model are Europe Brent spot prices FOB collected from the Energy Information Administration at the monthly level and aggregated to quarterly averages.We chose to use Brent oil prices rather than West Texas Intermediate prices because Brent prices are more relevant to California markets. Historical vehicle miles traveled on California highways are reported by the California Department of Transportation, CalTrans, at the monthly level.On-highway VMT data are reported in the aggregate, and not divided into gasoline and diesel vehicles.Our model also requires soybean prices, which we collect from the Agricultural Marketing Service at the United States Department of Agriculture .We aggregate monthly spot prices in Central Illinois to quarterly averages to be used in the model. The main variables of interest in our model are gasoline and diesel consumption and VMT in California as we need to forecast BAU fuel demand in order to construct a distribution of LCFS deficits. We collect monthly prime supplier sales volumes for California reformulated gasoline from the EIA.This measure captures all finished gasoline that is consumed in California, including imports to the state. We assume all gasoline is consumed in the transportation sector. Measuring diesel fuel consumption is more nuanced. The EIA reports monthly sales volumes for refiners at each step in the supply chain. We aggregate wholesale and retail sales volumes for No.2 distillate to construct a measure of consumption of No.2 distillate. According to data from the EIA, 99 percent of No.2 distillate is used for diesel fuel in California. Therefore we calculate sales volumes of CARB diesel, which is ultra-low sulfur diesel sold in California, as 99 percent of No.2 distillate sales. The diesel pool, however, comprises biomass based diesel ,vertical grow rack which includes bio-diesel and renewable diesel, as well as petroleum diesel. BBD demand was negligible prior to 2011, but has been increasing in the years since. Therefore, we construct the measure for diesel fuel consumption as the sum of BBD and ULSD. The EIA does not report sales of BBD, so we use volumes reported by CARB in the LCFS quarterly summary, since the years of substantial BBD demand occur in that time period. We aggregate monthly CARB diesel sales from the EIA to quarterly totals and add quarterly volumes of BBD from CARB. The LCFS regulates fuel used in the California transportation sector. Therefore, to accurately estimate the number of deficits generated from CARB diesel using our data, we need to measure the amount of diesel fuel consumed in California that is allocated to the transportation sector. Since 1992, approximately 70% of distillate consumed in California has been used on highway in the transportation sector.We therefore assume, in accordance with our definition of BAU, that 70% of all CARB diesel will be consumed in the transportation sector in each year over the 2019-2030 compliance period.
We are unaware of information that would lead us to believe a divergence from this long term could occur and we don’t consider altering this assumption in this study. Importantly, scaling diesel by a constant has no effect on the coefficient estimates in the VEC model that we use to generate our BAU simulations.The long-run coefficient estimates from the VEC cointegration model appear in Table 9. The three columns in Table 9 correspond to the three cointegrating equations specified in and and the rows to their long-run relationships with GSP, VMT, and the oil price.In the first two equations of Table 9, gasoline and diesel demand in California, the coefficients on the oil price capture the price responsiveness of demand for each fuel.The elasticity for gasoline, on the other hand, is positive but qualitatively small, and statistically insignificant at the 5% level. This may reflect fact that gasoline demand is very inelastic. The coefficients on GSP reflect the income effect. Gasoline and diesel fuel are normal goods and thus should be expected to be positively correlated with income in the state. The coefficient on VMT captures fuel economy improvements as more VMT per gallon implies fewer gallons. Because the VMT measure is not reported by vehicle type, implied fuel efficiency gains in each of the two fuel pools are not discernible. In the next section, we use the long-run coefficient estimates from Table 9, along with the short-run estimates located in the appendix in The estimates of the and Γ matrices from the VEC model in appear in Table A-4.We use the coefficient estimates from the VEC model to predict the distribution for each variable through the compliance period, 2019-2030. Specifically, we simulate 1000 potential values for each variable in each quarter during the compliance period. To this end, we assume that the potential shocks ?! that may occur in the compliance period have the same distribution as the shocks during our estimation sample period, 1987-2018. Using this assumption, we simulate potential future shocks by sampling randomly with replacement from the 1987-2018 shocks. For each random draw, we use the VEC model to generate a hypothetical path for the six variables. We repeat this exercise 1000 times to give us a distribution of potential paths.The hypothetical paths for blended gasoline, diesel, and VMT, simulated using , are described in Figure 17 with the median draw from each year and a 90% point wise confidence interval .In addition to those variables, we calculate the fuel economy of gasoline vehicles that is implied under BAU conditions. To do so, we multiply each VMT projection by the percent estimated in ARB’s EMFAC model to come from gasoline powered vehicles .Then we can express the average fuel economy, measured in miles per gallon , for gasoline vehicles by dividing gasoline VMT in each draw by the number of gallons of CaRFG. The implied fuel economy shown in Figure 17d highlights the range of efficiency gains considered in our simulations over the compliance period. This implied gasoline vehicle economy, derived from EMFAC percentages combined with our projections, is a fleet-wide average for gasoline powered vehicles only, and does not explicitly build in the recent California vehicle efficiency agreement with major automakers to reduce GHG emissions per mile for model years 2022 through 2026.For each variable in our VEC model, the level of future uncertainty grows as we move further into the future. In Figure 17a, 90% of the draws from our sample fall between 14 and 17 billion gallons of CaRFG being consumed in 2030 – a 12 percent increase and 13 percent decrease, respectively, from current levels.