Zero Emission Vehicles: Forecasting Fleet Scenarios and their Emissions Implications (2019)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

30 CHAPTER 4 MODELING ASSUMPTIONS AND RESULTS With the addition of one MA3T simulation for the Base Case analysis scenario, 16 sets of simulations were conducted for analysis of emissions changes due to advanced adoption of ZEVs. Each of the simulation sets consist of three separate simulations (with “Low”, “Medium”, and “High” parameter adjustment), which allowed further analysis of the sensitivity of MA3T to changes in input parameters. All of the simulations (a total of 49), including the parameters adjusted and the magnitude of parameter adjustments, are summarized in Table 8. Information gathered in the literature review completed for this project was used to inform the selection of which MA3T parameters to adjust, and, to the extent possible, the magnitude of the parameter adjustments made in the ZEV adoption scenario simulations. For example, the availability of public charging stations was based on data from the Electrify America National ZEV Plans. The assumptions made for each set of parameter adjustments are described in the next section. In general, the parameter adjustments for the infrastructure (I) scenarios are informed by more quantitative information compared to the Incentive/Policy (P) and cost parity (C) scenarios, for which quantitative projection data are lacking. A more generalized approach was used for many of those scenario simulations. In general, large parameter adjustments were used to better understand the relative importance of each parameter within the MA3T consumer choice model. These parameter adjustments do not necessarily reflect a likely future condition, but rather a possible future condition given technology and known investments in 2019. The largest parameter adjustments were used to demonstrate the largest possible increases in ZEV adoption given technology that is currently feasible. The analysis is not intended to make a case for any particular future outcome, but to illustrate the effect of each scenario on future emissions. Infrastructure parameters in MA3T include charging availability and power for specific “year points.” There are four year points, and the first (Year #1) and last (Year #4) points correspond with the first and last years in the MA3T modeling period (2005 and 2050). Based on the calibration of the model, the first and last year points should not be adjusted and the model must be run for the entire modeling period. Data for any years within that period can be selected for analysis. Although the analysis year for this study is 2040, parameter adjustments for the infrastructure simulations are only made relative to Year #4 (2050), as seen in Table 8. Furthermore, adjustments cannot be made to default year points with default years prior to 2019 without recalibration of the model, which is outside the scope of this work.

31 Table 8. MA3T parameter adjustments for ZEV adoption scenario simulations. Simulation Set ID MA 3T Parameters Adjusted Parameter Adjustment Values B No changes to default MA3T input parameters No changes to default MA3T input parameters I1 Public Charging Availability Defaults: 0% for year points #1 and #2 (2005 and 2011); variable by area for year points #3 and #4 (2017 and 2050) Low: Only values for CA adjusted (1.5x default values for central city and suburb; 2x for rural) Medium: Midpoint between high and low adjustment ([100% plus default] / 2) for Year #4 (2050) High: 100% availability by Year #4 (2050) I2 Public Charging Power Level Defaults: 0 kW for Year #1 (2005); 3.0 kW for Year #2 (2011); 11 kW (central city and suburb) and 3.0 kW (rural) for Year #3 (2017); 11 kW for Year #4 (2050) Low: No change Medium: Midpoint between high and low adjustment (181 kW) for all areas by Year #4 (2050) High: 350 kW for all areas by Year #4 (2050) I3 Home Charging Availability Defaults: variable by area and year point Low: No change Medium: Midpoint between high and low adjustment ([100% plus default] / 2) for Year #4 (2050) High: 100% for all areas by Year #4 (2050) I4 Home Charging Power Level Defaults: 1.1 kW for Year #1 (2005); 2.0 kW for Year #2 (2011); 3.0 kW for Year #3 (2025); 6.0 kW for Year #4 (2050) Low: No change Medium: 19.2 kW for all areas by Year #4 (2050) High: Same as medium adjustment I5 Workplace Charging Availability Defaults: 5% for all areas and year points Low: No change Medium: Midpoint between high and low adjustment (53%) for Year #4 (2050) High: 100% availability by Year #4 (2050) I6 Workplace Charging Power Level Low: No change

32 Simulation Set ID MA 3T Parameters Adjusted Parameter Adjustment Values Defaults: 1.1 kW for Year #1 (2005); 2.0 kW for Year #2 (2011); 3.0 kW for Year #3 (2025); 3.0 kW for Year #4 (2050) Medium: Midpoint between high and low adjustment (177 kW) for Year #4 High: 350 kW for all areas by Year #4 (2050) P1 American Recovery and Reinvestment Act (ARRA) Max # Vehicles per original equipment manufacturer (OEM) combined with Maximum Subsidy Default: 200,000 vehicles cap, and $7,500 maximum subsidy Low: 300,000 vehicles cap and $3,500 maximum subsidy Medium: 400,000 vehicles cap and $3,500 maximum subsidy High: 600,000 vehicles cap and $3,500 maximum subsidy P2 ARRA Number of OEM Producing ZEVs Default: 8 Low: 17 Medium: 20 High: 25 P3 State Rebate Amount Defaults: variable by state Low: 1.1x default Medium: 1.25x default High: 1.5x default P4 State Rebate Duration Default: 5 year duration (2011 start year) Low: 10 years (default start year) Medium: 15 years (default start year) High: 20 years (default start year) P5 Rebate Amount and Duration Applied to Other States Defaults: no rebates Low: Median rebate ($2,500) and 10-year duration Medium: Median rebate ($2,500) and 15-year High: Median rebate ($2,500) and 20-year duration P6 HOV Lane Access Duration Default: 5 years (2014 start year) Low: 10 years (default start year) Medium: 12 years (default start year) High: 15 years (default start year) C1 Vehicle Manufacturer Cost Defaults: variable by vehicle class and powertrain, and by analysis year Low: Unsubsidized vehicle manufacturer cost parity between ZEVs with ICEVs counterparts by category in 2040

33 Simulation Set ID MA 3T Parameters Adjusted Parameter Adjustment Values Medium: Unsubsidized vehicle manufacturer cost parity between ZEVs with ICEVs counterparts by category in 2035 High: Unsubsidized vehicle manufacturer cost parity between ZEVs with ICEVs counterparts by category in 2030 C2 Gasoline Price Defaults: variable by region and analysis year Low: A minimum rate of increase ($0.05/year) slightly lower than the increase in annual average price from U.S EIA data between 2002 and 2019, applied to all regions beginning in 2019 Medium: Rate of increase ($0.07/year) for annual average price from U.S. EIA data between 2002 and 2019, applied to all regions beginning in 2019 High: A maximum rate of increase ($0.10/year) slightly higher than the increase in annual average price from U.S. EIA data between 2002 and 2019, applied to all regions beginning in 2019 C3 Diesel Fuel Price Defaults: variable by region and analysis year Low: A minimum rate of increase ($0.05/year) slightly lower than the increase in annual average price from U.S. EIA data between 2002 and 2019, applied to all regions beginning in 2019 Medium: Rate of increase ($0.08/year) for annual average price from U.S. EIA data between 2002 and 2019, applied to all regions beginning in 2019 High: A maximum rate of increase ($0.10/year) slightly higher than the increase in annual average price from U.S. EIA data between 2002 and 2019, applied to all regions beginning in 2019

34 4.1 ASSUMPTIONS FOR INFRASTRUCTURE SCENARIOS A range of infrastructure scenarios that reflect likely changes in charging availability in public, workplace, and home locations were considered. First, it was validated that baseline inputs in MA3T were similar to “current” (i.e., approximately 2018) conditions, which was generally found to be the case. However, in looking at future years, the team found that MA3T’s assumptions with regard to infrastructure availability were generally conservative. The team did not make modifications to assumptions where there was limited evidence that conditions would change in the future, but rather left those factors as controls. Ultimately, six simulation sets that focused on changes to infrastructure available for charging electric vehicles were developed. 4.1.1 Public charging availability assumptions (I1 simulations) MA3T defines “public charging availability” as the ratio of the number of charging stations to the number of gasoline stations, expressed as a percentage. Charging availability is intended to indicate that drivers of electric vehicles have approximately equal access to charging stations as drivers of gasoline powered vehicles have to gas stations. Full public charging availability is the estimated density of EVSE necessary to fully meet demand for public electric vehicle charging, (i.e., offer equal refueling opportunities to BEV owners as gasoline powered vehicle owners). Less than 100% availability is a proxy for the percentage of demand that is met The model includes availability for three area types: central city, suburb, and rural locations. Model developers were not able to share proprietary data for model inputs; in 2019, the percentage availability set within MA3T was equal to approximately 27% in both central city and suburb locations across all states. For rural locations, availability is set to 5% for all states. This approach was used to estimate the relative magnitude of planned EVSE rollout. To estimate the number of gasoline stations, the number of establishments classified under National American Industry Classification System (NAICS) sub-sector 447 (gasoline stations) by state, using the 2016 County Business Patterns (United States Census 2018) were found. The number of public charging stations by state was obtained from the U.S. DOE Alternative Fuels Data Center (United States DOE 2019c). These estimates were used (1) as a baseline from which to estimate the magnitude of planned expansions of electric charging stations across the U.S., and (2) to inform adjustments to the percentage of public charging availability in 2050. The number of electric stations (stations may include multiple chargers, although the capacity of charging stations may not equal that of a gas station) and gas stations is included in Table A-1 in Appendix A Supplemental Data. Several key initiatives informed the review of estimated increase in public charging stations: the initiatives planned by Electrify America, subsidies provided by California investor owned utilities (IOUs) and planned “electric” transportation corridors. The largest of these initiatives is Electrify America’s planned investments in electric charging stations. As of 2019, Electrify America has released Cycle 2 plans for spending at a national level (excluding California) and within the state of California that include a targeted number of operational charging stations (Table A-2). Electrify America plans to deploy approximately 3,700 stations throughout the U.S. by 2021; while these will largely be targeted to residential users, workplaces may also be included. The rollout will also include up to 155 entirely public chargers. The three IOUs in California—SDG&E, SCE, and Pacific Gas and Electric (PG&E)—are also implementing pilot programs to install infrastructure to support electric vehicle charging in workplace, home, and public destinations as of 2019; Table A-3 lists the numbers of planned charging stations for

35 the three California utilities. An additional 12,500 electric vehicle charging stations are planned in this IOUs initiative, of which 6,250 would be located either at homes or at workplaces. Finally, there are a variety of regional and national initiatives to “electrify” important transportation corridors by providing access to electric vehicle charging stations at regular intervals. The West Coast Electric Highway Network, the Regional Electric Vehicle Plan for the West (REV West), and the National Alternative Fuel and Electric Charging Network, cumulatively plan to build electric vehicle charging infrastructure on approximately 92,800 miles of roadway. Conservatively assuming approximately 50 miles between charging stations, “electrifying” this roadway would require approximately 1,900 new charging stations. These estimates illustrate that initiatives underway at the time this report was prepared (2019) provide strong support for electric vehicle charging infrastructure relative to the availability of gasoline stations. However, reductions in the cost of installing EVSE and uncertainties in driver behavior make future availability difficult to estimate. To reflect the rapid deployment of EVSE and the high degree of uncertainty around anticipated demand, and to test the impact of charging infrastructure availability, public charging infrastructure is set to 100% availability in 2050 for the High case in this simulation set. This estimate is not intended to represent a forecast of future growth, but rather an upper limit of charging availability. For the Low case, the percentage availability in 2050 is assumed to equal the minimum of either 50% availability, or twice the default level of availability in MA3T V20190404. The default inputs resulted in a slight decrease in future public charging availability in California, where the Low case values were set to increase 1.5 times over default levels in the central city and suburbs, and double in rural areas. The Medium case was assumed to be the midpoint of the high and low cases. 4.1.2 Public charging power level (I2 simulations) In the I2 simulation set, the public charging power (expressed in kilowatts) is equal to the power available at public charging stations. Known key parameters for this simulation set are defined by the technology being deployed by Electrify America. Level 1 chargers generally describe household AC power, typically with a maximum power of 1.9 kW. Level 2 chargers typically charge up to 19.2 kW. DCFCs (also referred to as Level 3 chargers) offer the fastest charging speeds. Electrify America stations currently deploy Level 3 DCFCs with a maximum power of 350 kW. The parameter values for both public charging power in 2017 and public charging power in 2050 are set within MA3T to 11 kW for central city and suburbs and 3 kW for rural areas. Based on advances in the technology known to be underway, the High case assumes power levels of 350 kW for all area types in 2050. The Low case assumes no change from MA3T default parameters for 2050. The Medium case assumes a midpoint between the high and low case for 2050. 4.1.3 Home charging availability (I3 simulations) Home charging is most easily accomplished in a home garage or carport. Absent those features, home owners or renters face additional expenses, such as added security, weather, permitting, and metering, that must be factored into the cost of installing charging infrastructure in a residential setting. Homeowners/renters may also choose to use a portable charging station. A key parameter in these simulations is the availability of subsidies to homeowners/renters to install electric charging infrastructure in homes, and the availability of solutions geared towards homeowners/renters without carports. The High case for this parameter is assumed to equal 100% for all areas by Year #4 (2050). The Low case is assumed to equal baseline MA3T inputs, which vary by year and area type and are developed based on

36 proprietary data that was not obtained for this study. The Medium case assumes the midpoint between the high and low adjustments for 2050. 4.1.4 Home charging power (I4 simulations) Home charging power is the power available at home charging stations, expressed in kilowatts. Within MA3T, default home charging power is assumed to equal 3 kW in Year #3 (2025) and 6 kW in Year #4 (2050). L2 chargers have a maximum charge power of 19.2 kW (approximately equal to the power required to run a clothes dryer). For the Low case, no change to the MA3T baseline assumptions was assumed. For the Medium case, it was assumed that home chargers will have access to 19.2 kW in 2050. However, no technology was identified during the literature review that could be used to upgrade home infrastructure above a maximum output of 19.2 kW. The assumption for the High case was the same as in the Medium case. 4.1.5 Workplace charging availability (I5 simulations) Workplace charging availability may fill a gap for drivers who need to charge at their place of work before making a return commute trip. MA3T default values assume that workplace charging availability will be 5% for all areas and all years. For the Low case, no adjustment to MA3T default values were made. For the High case, based on known initiatives in the rollout of charging infrastructure, it was assumed that workplace charging ability will be 100% by 2050. The Medium case assumed the midpoint between the high and low adjustments for 2050. 4.1.6 Workplace charging power level (I6 simulations) Workplace charging power within MA3T is assumed to equal 3 kW in 2050. Default values were used for the Low case. Based on the availability of technology discussed above for public charging, the High case assumes charging power of 350 kW in 2050. This assumes that as this technology becomes more commonplace, it will become available in most public venues, including the workplace. The Medium case input is assumed to be the midpoint between the Low and High values. 4.2 ASSUMPTIONS FOR INCENTIVE/POLICY SCENARIOS In the Incentive/Policy simulations, all dollar values in MA3T are measured by the purchasing power of dollars in 2018. That is, the amounts are nominal dollars that do not account for inflation or deflation of other past or future years relative to 2018. 4.2.1 ARRA parameter assumptions (P1 and P2 simulations) In the P1 simulation set, the maximum cumulative number of subsidized vehicles per OEM was assumed to represent a range of values based on recent legislative proposals. For example, the proposed Driving America Forward Act (Congressional Research Service 2019) essentially raises the cap on the number of subsidized vehicles from 200,000 to 600,000. For each case of the P1 simulation set, increasing caps were assumed, and the cap was set to 600,000 for the High case. In addition to the input for the cap on vehicle sales, the P1 simulation set also includes adjustment of the maximum subsidy per eligible vehicle. Although the Driving America Forward Act proposes a credit of up to $7,000 for the number of vehicles beyond the 200,000 cap, only a single subsidy value can be used in MA3T. A more conservative value ($3,500) was used in the P1 simulation set. MA3T uses nominal 2018 dollars as input for the credit amount, but it is possible that credits would be issued in future years until a manufacturer reaches the cap

37 on vehicle sales. Test simulations where the credit amount was changed, not presented in this report, indicated that the effect of the credit amount on future ZEV population was small. For the P2 simulation set, the Low case value for number of OEMs producing eligible vehicles was assumed to equal 17, which is supported by recent data (Nanalyze 2017). For the Medium case, it was assumed that nearly all auto manufacturers (approximately 20 as of 2019) selling light-duty passenger vehicles in the U.S. produce eligible vehicles; and for the High case it was assumed that five additional manufacturers of ZEVs would enter the market. 4.2.2 State rebate assumptions (P3 through P5 simulations) Simulation sets P3 and P4 relate to inputs for state rebates, and inputs were only adjusted for the 19 states that had rebate programs beginning in 2011 (the default start year for rebates in MA3T). In fact, one additional state (Oregon) has a rebate program that went into effect in 2018 and ends in 2024, and three states with rebate inputs in MA3T actually have or had tax credits, not rebates (Colorado [effective 2017- 2026], Georgia [expired in 2015], and Maryland [effective 2017-2020]). A test MA3T simulation showed that changing rebate inputs to tax credit inputs for those three states had no effect on estimated ZEV population. Although rebates for some states actually took effect in years later than 2011, changing the start year would require recalibration of the model. Therefore, the start year (2011) for rebates in these simulations was not changed. The changes to input parameters for these simulations were set arbitrarily, assuming that advanced use of incentives would correspond to increasing the rebate amount or the duration of existing rebates. Rebate amounts were increased by 10% for the Low case, 25% for the Medium case, and 50% for the High case. The default rebate amounts are shown in Table A-4 in Appendix A (the default rebate duration is five years). Rebate amounts in MA3T do not include any offsets or negative policy impacts such as states increasing ZEV registration fees as a result of losses in gas taxes, nor are there other model inputs that account for such factors. Rebate durations were increased in five-year increments over the Low, Medium, and High cases up to 15 years beyond the default value for the High case. Simulation set P5 represents the adoption of rebates by states that did not offer rebates as of 2011 (the remaining 31 states). The inputs for this set of simulations were also set arbitrarily, using (1) the median rebate amount across states that already offer rebates, and (2) the Low, Medium, and High adjustments to rebate duration used in simulation set P4. The approach for parameter adjustments in this simulation set was selected after analysis of the population change results from sets P3 and P4. During test simulations, it was found that adjusting the amount of rebates had little impact on ZEV populations, but adjusting the duration of rebates had a moderate impact. This could be reflective of the fact that consumers are generally unaware of available financial incentives for the purchase of ZEVs (e.g., Jin and Slowik, 2017). Consumer education has been identified as an important need for increasing the effectiveness of incentives and ZEV adoption, as demonstrated by planned investments by Electrify America (Electrify America 2017 and 2018). In addition, states with large rebates as of 2019 do not necessarily have correspondingly high rates of electric vehicle adoption. For the purposes of this research project, it was decided that a simulation set resulting in a moderate impact would be more valuable than one resulting in essentially no impact.

38 4.2.3 HOV lane access duration assumptions (P6 simulations) The default start year for HOV lane access policy in MA3T is 2014, and the default duration is five years. Therefore, HOV lane access for any ZEV would end in 2019 within the modeling framework. The default start year for HOV lane access in MA3T cannot be adjusted without recalibrating the model. For the P6 simulations, it was assumed that all states with HOV lane access for ZEVs would be extended by 5 (Low case), 7 (Medium case), and 10 (High case) years. These adjustments were informed by several studies and research findings, including: • In 2013, HOV lane access in California was found to be the primary motivation for the purchase of several models of PEVs (United States Energy Information Administration 2017) • A UCLA study asserted that access to HOV and high-occupancy toll (HOT) lanes is the “single biggest incentive” for Californians to purchase ZEVs (e.g., see Los Angeles Times 2018 and Digital Trends 2018) • Another UCLA study found that HOV lane access for ZEVs had no capital costs and is more cost effective than the state’s rebate program (Sheldon and DeShazo 2016) • California introduced a new 4-year HOV lane access incentive beginning in 2019 (California Department of Motor Vehicles 2019). 4.3 ASSUMPTIONS FOR COST PARITY SCENARIOS 4.3.1 Vehicle manufacturer cost assumptions (C1 simulations) In MA3T, battery electric cars with a range of 100 miles reach vehicle manufacturer cost parity with conventional spark ignition cars in 2035. No other ZEVs in other vehicle classes in MA3T reach cost parity out to 2050. In fact, MA3T embeds default values that assume the manufacturer costs of ZEVs in the other classes (i.e., car-SUVs, pickup trucks, truck-SUVs, and vans) are substantially greater than their conventional vehicle (referring to ICEV) counterparts. However, in order to assess the extent of vehicle manufacturer cost impacts on ZEV populations, the C1 simulations included aggressive adjustments to MA3T inputs. For these simulations, the vehicle manufacturer costs as assigned by MA3T were changed to achieve cost parity at a faster rate than the model’s default values. The years of cost parity in the Low, Medium, and High cases are 2040, 2035, and 2030. These adjustments reflect the literature findings that purchase cost—of which battery cost is the most important determinant—is the largest barrier to ZEV adoption; at the time of this report (2019), battery cost continues to decline at a fast rate (Beacon Economics 2018). The values used are listed in Table A-5. These costs are a manufacturer’s capital costs, and do not include maintenance costs or financial incentives such as rebates or tax credits. It is important to acknowledge auto industry reports (for example, as of 2019, reports by Ford, General Motors, and Volkswagen) announcing that automakers will release tens of ZEV models by 2025. The auto industry reports suggest that cost parity will be reached sooner than may have previously been expected, perhaps by mid-2020 or 2030 (Eisenstein 2019; Lambert 2019; Smith 2019; Hanley 2019). 4.3.2 Gasoline and diesel price assumptions (C2 and C3 simulations) Data for weekly U.S. retail gasoline (all grades and formulations) and diesel (U.S. No. 2) prices from the U.S. EIA (https://www.eia.gov/petroleum/gasdiesel/) were used to estimate rates of price increase for the Low, Medium, and High case simulations. These data are shown in Figure A-1 and Figure A-2 in

39 Appendix A. From 2002 through 2018, theses prices increased by a small constant rate (on an annual average, $0.07 for gasoline and $0.08 for diesel). The default gasoline and diesel fuel prices in MA3T vary by region of the U.S. and year. From 2019 to 2050, the forecasted prices for both gasoline and diesel increase on an annual basis by only $0.02. To capture the historical rates of increase in gasoline and diesel prices (based on the U.S. EIA data) in the C2 and C3 simulations, annual growth rates of $0.07 and $0.08 were applied for the Medium cases of the gasoline and diesel price simulations, respectively. A rate of $0.05/year for both simulation sets was used for the Low case, and a rate of $0.10/year for both simulation sets was used for the High case. Although the historical prices reflect inflation, MA3T assumes that prices in all future years are represented in 2018 dollars. The adjusted price inputs in were applied in MA3T for all regions. Increases in fuel price can affect whether a consumer will choose to buy (either an ICEV or ATV) or not buy a vehicle, as well as the mode of transportation a consumer chooses. MA3T includes a no-buy option that can be influenced by high fuel prices. 4.4 MODEL RESULTS FOR ZEV POPULATIONS The total ZEV populations estimated by the MA3T model for the year 2040, shown in Table 9, demonstrate changes from the Base Case modeled ZEV population (25.5 million ZEVs) that are generally consistent with the importance of ZEV adoption drivers and barriers identified in the literature review conducted for Task 2 of this project. By default, MA3T assumes that there are no FCEVs on the market throughout the 2005-2050 modeling period. To obtain FCEV sales and population estimates from MA3T, numerous input parameters would be required and the model would need to be recalibrated. The 2019 AEO forecasts that FCEVs will account for less than half a percent of the total light-duty vehicle fleet in 2040, and it is highly uncertain what the FCEV technology and supporting infrastructure will be 20 or 30 years from now (2019). Adjustments to ZEV costs—in terms of manufacturer cost (battery cost is not used separately from manufacturer cost in the latest version of MA3T when this report was prepared [version V20190404]) and gasoline price—in the cost parity simulations result in the largest ZEV populations (and therefore the largest increases in ZEV population) across all scenarios. Diesel price has only a small impact on ZEV adoption, possibly because of the higher cost of diesel vehicles and small number of diesel vehicle options in MA3T. Reflective of the importance of HOV lane access to ZEV adoption, the corresponding simulations result in ZEV populations that represent increases in ZEV population relative to the Base Case that are among the highest across simulations, after only consideration of (1) vehicle manufacturer costs and (2) gasoline costs. The public charging power level simulations also produce some of the highest increases in ZEV adoption across the simulations. Slow charging time was identified in the literature review as a key barrier to ZEV adoption. The modeled increases in ZEV adoption for the public charging availability simulations were less than for the charging power level simulations. Increases in ZEV adoption in the home charging availability simulations was comparable to those for the public charging availability simulations, but the change in ZEV adoption in home charging power levels was essentially zero. This could reflect the fact that home charging usually occurs overnight with long charging time, so the power level is less important than the power level of public charging stations. Interestingly, changes to workplace charging availability and power level had little impact on ZEV adoption. This could imply that ZEV owners with access to home and public charging have less need for charging at their workplaces.

40 Although ZEV purchase cost was identified as the greatest barrier to ZEV adoption in the literature review, and financial incentives such as rebates effectively reduce the cost of ZEVs, changes to the ARRA tax credit parameters and state rebate amounts had only a modest impact on the ZEV populations estimated by MA3T. Increasing the number of OEMs producing eligible ZEVs had a slightly greater impact than increasing the cap on the number of vehicles eligible for the ARRA credit. Surprisingly, in test simulations for the ARRA cap and maximum subsidy in which the maximum subsidy was set to $7,000, the increases in ZEV adoption were essentially the same when using the lower ($3,500) maximum subsidy. Detailed results from those test simulations are not presented in this report. Similarly, increasing state rebate amounts had essentially no impact on ZEV adoption. However, increasing the duration of state rebates had impacts similar to increasing the ARRA cap. The increase in ZEV adoption was substantial for the High case simulation in which the median state rebate with a 20- year duration was applied to states that have no rebates as of 2019; this increase is comparable to the increase in ZEV population in the High case simulation for public charging power level. One of the MA3T model developers provided information that vehicle manufacturer cost, rebates, and tax credits are valued differently in the model. It is possible that the model is underestimating the value of tax credits and rebates.

41 Table 9. Model results of total ZEV population (millions of vehicles) in 2040 under all scenarios, as estimated by MA3T.a Simulation Set ID Scenario Description Low Medium High B Base Case 25.5 C1 Vehicle Manufacturer Cost 30.9 69.3 113.5 C2 Gasoline Prices 28.3 32.5 38.5 I2 Public Charging Power Level NAb 34.7 37.3 P5 Instant Rebate Amount and Duration Applied to Other States 25.5 26.8 36.6 P6 HOV Lane Access Duration 26.3 28.0 31.6 I1 Public Charging Availability 25.5 27.7 30.0 P2 ARRA Number of OEM Producers 28.1 29.9 29.9 P4 Instant Rebate Duration 25.5 26.5 29.5 P1 ARRA Vehicle Cap and Maximum Subsidy 26.8 27.9 29.4 I3 Home Charging Availability NA 27.3 29.2 I5 Workplace Charging Availability NA 25.7 25.9 C3 Diesel Prices 25.6 25.8 25.9 P3 Instant Rebate Amount 25.5 25.5 25.5 I6 Workplace Charging Power Level NA 25.5 25.5 I4 Home Charging Power Level NA 25.5 25.5 a Order of data is based on “High” simulation results. b NA indicates Not Applicable; no changes were made to default parameter values. Table 10 summarizes the number of ZEVs relative to the total light-duty vehicle population estimated by MA3T for each analysis scenario across the Low, Medium, and High cases for the year 2040. The percentage of ZEVs in the Base Case scenario is 9%. Although the percentage of ZEVs does not increase from the Base Case for several of the simulations, it does increase by 2% on average and by as much as almost 30% across all simulations. The median increase in percentage of ZEVs across all scenario

42 simulations is 0.5%. For comparison, the market share of BEVs, FCEVs, and PHEVs in California in 2018 was 8%. The results in Table 4 indicate that—based on the MA3T simulations—by 2040, the nationwide percentage of ZEVs could surpass the 2018 market share of ZEVs in California, where sales of BEVs and FCEVs were a factor of ten greater than sales in the state with the next highest sales in 2018. Table 10. Model results of total ZEV population in 2040 (as a percentage of the total light- duty vehicle population) under all scenarios, as estimated by MA3T.a Scenario ID Scenario Description Low Medium High B Base Case 9% C1 Vehicle Manufacturer Cost 11% 24% 38% C2 Gasoline Prices 10% 12% 15% I2 Public Charging Power Level NAb 13% 13% P5 Rebate Applied to Other States 9% 10% 13% P6 HOV Lane Access Duration 10% 10% 11% I1 Public Charging Availability 9% 10% 11% P2 ARRA Number of OEM Producers 10% 11% 11% P4 Instant Rebate Duration 10% 10% 11% P1 ARRA Vehicle Cap and Maximum Subsidy 10% 10% 11% I3 Home Charging Availability NA 10% 11% C3 Diesel Prices 9% 9% 9% I5 Workplace Charging Availability NA 9% 9% P3 Instant Rebate Amount 9% 9% 9% I6 Workplace Charging Power Level NA 9% 9% I4 Home Charging Power Level NA 9% 9% a Order of data is based on “High” simulation results. b NA indicates Not Applicable; no changes were made to default parameter values. 4.5 MODEL RESULTS FOR EMISSIONS REDUCTIONS The MOVES2014b model was used to calculate light-duty vehicle exhaust emissions for all analysis scenarios. The population of the conventional vehicles modeled with MA3T was combined with the emission factors modeled with MOVES2014b to calculate the exhaust emissions. While other emission processes (e.g., start emissions, evaporative emissions, and refueling emissions) were not modeled, the

43 increased adoption of ZEVs would generally reduce emissions; an example of this would be a reduction of evaporative HC from the light-duty fleet. MOVES2014b provides emission factors in grams per vehicle mile traveled (VMT). Therefore, to calculate emissions using MOVES emission factors, VMT information is required. However, MA3T provides annual population and sales for light-duty vehicles; it does not provide estimates of VMT. The VMT per vehicle estimates from MOVES2014b were used to calculate the total VMT for each vehicle type from MA3T’s population estimate. This approach assumes that the VMT for each vehicle type would not change because of the increase or decrease of electric vehicle population. MOVES2014b was run in emission mode using the national default activity (i.e., age distribution, source type population, and VMT) for the calendar year 2040. The activity output (i.e., population and VMT) and the emission estimate (in grams) were used to back calculate the emission factors in g/mile. The population and VMT estimates were also used to calculate VMT per vehicle. The pollutants included in the analysis were criteria pollutants (nitrogen oxides [NOx], carbon monoxide [CO], particulate matter [PM2.5 and PM10]), total gaseous hydrocarbons (HCs), MSATs (1,3-butadiene, acetaldehyde, acrolein, benzene, ethylbenzene, formaldehyde, naphthalene [gaseous and particulate]), and GHGs (carbon dioxide [CO2], methane [CH4], and nitrous oxide [N2O]). Since the emissions focus of this study was vehicle exhaust, the modeling did not include emissions of PM from tire wear, brake wear, and re-entrained road dust, which are independent of vehicle fuel type. The emission factors and VMT per vehicle estimates were separated by model year to more accurately calculate the emissions. MA3T estimated annual population and sales for light-duty vehicles, including cars, car sport utility vehicles (CSUV), pickup trucks, truck sport utility vehicles (TSUV), and vans. These vehicles were mapped to the MOVES vehicle types as shown in Table 11. Annual sales from each year were used as the population from each model year. To calculate the emissions, the population was then multiplied by the VMT per vehicle and the emission factor for the corresponding model year. The total emissions for the calendar year 2040 was then calculated by summing up the emissions from different model years. Table 11. Vehicle type mapping between MA3T and MOVES2014b. MA3T Vehicle Class MOVES Vehicle Type MOVES Vehicle Type ID Car Passenger cars 21 Car SUV (CSUV) Passenger cars 21 Pickup truck Light-duty truck 31 Truck SUV (TSUV) Light-duty truck 31 Van Light-duty truck 31

44 The emissions reductions for each simulation and pollutant modeled are summarized in Table 12 (also in Figure A-3 through Figure A-5), Table 13 (also in Figure A-6 through Figure A-8), Table 14 (also in Figure A-9 through Figure A-11), and Table 15 (also in Figure A-12 though Figure A-14). The tables are separated by pollutant type (i.e., criteria pollutants and HCs, MSATs, and GHGs) and also include the percentage change in emissions for each simulation and pollutant. GHG emissions reduction results are presented in Table 15, which also includes the GHG equivalence in terms of the number of passenger vehicles driven for one year, which was obtained from the Environmental Protection Agency’s (EPA) Greenhouse Gas Equivalencies Calculator (available at https://www.epa.gov/energy/greenhouse- gas-equivalencies-calculator). These emissions reductions are vehicle exhaust emissions only, and do not account for “wheel-to-well” emissions over the life of vehicles. The cost parity simulations for vehicle manufacturer cost result in reductions of (1) conventional vehicles equivalent to as much as 11% of the on-road light-duty vehicle fleet in 2017,iv and (2) GHG emissions equivalent to as much as 2% of total U.S. GHG emissions (CO2 equivalent) in 2017.v For the High and Medium cases of the gasoline price simulations, the GHG emissions reductions are 1% relative to total U.S. GHG emissions in 2017. The amount relative to total U.S. GHG emissions for all other simulations is much less than 1%. The reductions in GHG emissions for each simulation are generally greater than reductions in other pollutant emissions. The GHG equivalence values (equivalent to the number of passenger vehicles driven in one year) are roughly one million vehicles per 1% decrease in CO2 equivalent GHG emissions. Emissions reductions for CO2 equivalent (CO2e) are not presented with the GHG results because the differences between CO2e and CO2 emissions were 1% or less across all simulations. As would be expected, emissions reductions increase from the Low to Medium to High case simulations. Across all simulations, the maximum emissions reductions, corresponding with the High case for each simulation set, are generally 2% for all pollutants. This includes criteria pollutants, total HCs, and MSATs (the maximum reductions are slightly higher for GHGs, up to 3%). The maximum reductions are substantially greater than 2% in a few of the simulations as summarized in the lists below, which include results and observations based on the literature review. The categories are defined by a range of percentage reductions. The infrastructure simulations for charging availability and charging power are shown together in a single category, even though one of the two (availability or power) fall into another category (specifically, the less than 1% category). Simulations with emissions reductions greater than 3% • The greatest emissions reductions across all simulations are for the High cases of the vehicle manufacturer cost (a 21% reduction) and gasoline price (14%) simulations. In the Medium cases for those two simulation sets, the emissions reductions are 10% and 9%. The Low case of the gasoline price simulation set produced a 4% reduction in emissions. ‒ Initial ZEV purchase cost, driven by battery cost, is the greatest barrier to ZEV adoption. ‒ One study found that gasoline price is one of the most significant factors affecting ZEV adoption (Babaee et al., 2014). ‒ Another study found a correlation between BEV sales and gasoline prices (Transportation Research Board and National Research Council, 2015).

45 • The High case simulation for rebates applied to other states (states that have not had rebates since as early as 2011) produced a 9% reduction in emissions. ‒ Similar to the result for rebate amount simulations for states that already have rebates, emissions were not reduced by implementing and increasing the amount of a rebate for other states across the Low, Medium, and High cases. ‒ The increasing duration of a rebate with amount equal to the median amount across states already having rebates drove the effect on the emissions reductions in this simulation set. ‒ As is the case for the state rebate duration simulations for states already having a rebate, rebate durations may be more important because late majority and laggard consumers will purchase ZEVs in the future. ‒ This simulation set had some of the most aggressive assumptions across simulations (it is unlikely that all states will implement ZEV rebates), and thus resulted in some of the greatest emissions reductions. Simulations with emissions reductions between 2% and 3% • An increase in the duration of HOV/HOT lane access results in emissions reductions up to 3%. ‒ In 2013, HOV lane access was the primary motivation for purchasing ZEVs in California (United States Energy Information Administration, 2017); this has continued to be true considering other more recent reports cited earlier in this report. ‒ The relatively small impact on emissions reductions in this simulation set is likely because HOV/HOT lane access is not a primary driver for markets without large metropolitan areas. • An increase in public charging power results in greater reduction of emissions (up to 2%) than public charging availability (up to 1%). ‒ Slow charging time is a barrier to ZEV adoption. ‒ Public charging availability is not a significant concern for drivers with access to home charging. • The ARRA simulation sets (the vehicle cap and maximum subsidy set and the number of OEM producers set) result in about the same reduction in emissions (up to 2%). ‒ It is possible that the low emissions reductions related to ARRA factors is a result of a lack of consumer and auto dealer awareness. ‒ Several studies discuss the need for increasing consumer and auto dealer knowledge of ZEVs and available incentives. • An increase in home charging availability results in greater reduction of emissions (up to 2%) than home charging power (0%). ‒ Most ZEV drivers do more than 80% of their vehicle charging at home and overnight (U.S. DOE 2019d and 2019e). ‒ Home charging power is less of a concern than availability, because long charging time can be used overnight.

46 • Emissions reductions for the state rebate amount simulation set are 0%, but up to 2% for the rebate duration set. ‒ MA3T values rebates, tax credits, and vehicle manufacturer cost differently; although rebates and tax credits effectively reduce the purchase cost of ZEVs, MA3T did not produce substantial changes in estimated ZEV populations when rebate amounts were increased. ‒ Although recent efforts—particularly in California—are better incentivizing ZEVs for low- to mid-income drivers, ZEVs have generally been purchased by higher income drivers; higher income drivers may be less affected by rebate amounts. ‒ Rebate durations may be more important because late majority and laggard consumers will purchase ZEVs in the future. • The emissions reductions for the High case of the diesel price simulation set is 2%, while the Medium and Low cases produced emissions reductions less than 2%. The majority of conventional vehicle types in MA3T have spark ignition technology. Simulations with emissions reductions of less than 2% • An increase in work charging availability results in greater reduction of emissions (up to 1%) than work charging power (0%). ‒ Workplace charging was not identified as a driver or barrier in the Task 2 literature review. ‒ One assumption is that drivers with access to home charging will not depend on workplace charging for their daily commuting. It is important to consider that the modeled reductions in GHG emissions, presented in Table 15 and in Figure A-12 through Figure A-14, do not account for expected increases in GHG emissions as a result of the phenomenon termed “leakage.” This is related to the way in which vehicle manufacturers meet the CAFE standards. Not all vehicles produced by a manufacturer are required to meet the standard; rather, the overall standard applies as an average of the fuel economies across the mix of vehicles produced by a manufacturer. Therefore, as more ZEVs are produced, less fuel-efficient vehicles can also be produced, and the manufacturer can still meet the overall fuel economy standard. Jenn et al. (2016) calculated increases in CO2 emissions and gasoline consumption each time an alternative fuel vehicle (AFV) is sold, assuming (1) no policy changes between 2016 and 2025, (2) no effect of AFV sales on a manufacturer’s total vehicle sales, and (3) that a manufacturer complies with future GHG emissions regulations. Using projections of vehicle sales from 2012-2015 AEO reports, they estimated a net increase of 30 to 70 million metric tons of CO2 over the lifetimes of vehicles sold from 2012 to 2025 as a result of the CAFE standards and regulation of vehicle fleet average GHG emissions. However, the advancement of ZEV technology and growth in ZEV sales between 2016 and 2018 could indicate a changing future vehicle fleet mix. The rapid growth in sales could also have an influence on how future standards are set and lead to stricter standards, as suggested in a report from the Congressional Budget Office (2012), resulting in a greater overall fuel efficiency of the future vehicle fleet. Furthermore, well-to-wheel emissions associated with the sources of electricity used to charge ZEVs are not modeled by MA3T, and modeling those was outside the scope of this study. However, information is available that can provide insight on how different electricity sources could contribute to ZEV emissions.

47 The DOE Alternative Fuels Data Center provides summaries of state-level electricity sources and annual emissions per vehicle for EVs, PHEVs, HEVs, and gas-powered vehicles (see https://afdc.energy.gov/vehicles/electric_emissions.html). Figure 5 shows these data for two states with different profiles of electricity sources and emissions associated with ZEVs. Washington state, where hydroelectric sources provide a majority of the power, has relatively low ZEV emissions derived from the electricity source; on the other hand, coal provides a large fraction of electricity in Colorado, where ZEVs have relatively large emissions associated with electricity sources. Further analysis of this type of data would provide a more complete picture of well-to-wheel mobile source emissions impacts due to the adoption of ZEVs. Figure 5. State average electricity sources and annual emissions per EVs and PHEVs for Washington state and Colorado. Image source: U.S. DOE (https://afdc.energy.gov/vehicles/electric_emissions.html)

48 Table 12. Reduction in modeled light-duty passenger vehicle emissions (in tons) of criteria pollutants and total hydrocarbons (HCs) for calendar year 2040. Simulation Set ID Simulation Description NOx (%) CO (%) PM10 (%) PM2.5 (%) Total HCs (%) Averagea (%) C1 Vehicle Manufacturer Cost (High) 21,233 (16) 411,466 (16) 1,236 (16) 1,095 (16) 15,739 (15) (16) C2 Gasoline Prices (High) 15,225 (11) 304,390 (12) 848 (11) 751 (11) 12,160 (12) (11) P5 Other States Rebates (High) 11,326 (8) 220,487 (9) 648 (9) 573 (9) 8,540 (8) (9) C2 Gasoline Prices (Medium) 9,200 (7) 182,173 (7) 517 (7) 458 (7) 7,280 (7) (7) C1 Vehicle Manufacturer Cost (Medium) 8,409 (6) 161,684 (6) 513 (7) 454 (7) 6,491 (6) (7) C2 Gasoline Prices (Low) 4,182 (3) 82,099 (3) 238 (3) 211 (3) 3,289 (3) (3) P6 HOV Access Duration (High) 4,131 (3) 79,696 (3) 235 (3) 208 (3) 3,163 (3) (3) I2 Public Charging Power (High) 3,110 (2) 60,322 (2) 180 (2) 159 (2) 2,337 (2) (2) P1 ARRA Max # Vehicles and Subsidy (High) 2,993 (2) 58,850 (2) 172 (2) 153 (2) 2,258 (2) (2)

49 Simulation Set ID Simulation Description NOx (%) CO (%) PM10 (%) PM2.5 (%) Total HCs (%) Averagea (%) P2 ARRA Number of OEM Producers (High) 2,835 (2) 54,380 (2) 165 (2) 146 (2) 2,177 (2) (2) P2 ARRA Number of OEM Producers (Medium) 2,835 (2) 54,380 (2) 165 (2) 146 (2) 2,177 (2) (2) P4 State Rebate Duration (High) 2,844 (2) 55,265 (2) 161 (2) 143 (2) 2,144 (2) (2) I2 Public Charging Power (Medium) 2,641 (2) 51,291 (2) 153 (2) 136 (2) 1,988 (2) (2) I3 Home Charging Availability (High) 2,573 (2) 49,679 (2) 152 (2) 135 (2) 1,982 (2) (2) P1 ARRA Max # Vehicles and Subsidy (Medium) 2,017 (2) 39,792 (2) 118 (2) 104 (2) 1,523 (2) (2) P6 HOV Access Duration (Medium) 1,836 (1) 35,048 (1) 105 (1) 93 (1) 1,423 (1) (1) I1 Public Charging Availability (High) 1,820 (1) 35,458 (1) 106 (1) 94 (1) 1,376 (1) (1) P2 ARRA Number of OEM Producers (Low) 1,743 (1) 33,252 (1) 99 (1) 88 (1) 1,347 (1) (1) C3 Diesel Fuel Prices (High) 1,709 (1) 29,515 (1) 114 (2) 101 (2) 1,049 (1) (1) C3 Diesel Fuel Prices (Medium) 1,485 25,676 100 88 910

50 Simulation Set ID Simulation Description NOx (%) CO (%) PM10 (%) PM2.5 (%) Total HCs (%) Averagea (%) (1) (1) (1) (1) (1) (1) I1 Public Charging Availability (Medium) 1,368 (1) 26,660 (1) 80 (1) 71 (1) 1,037 (1) (1) P1 ARRA Max # Vehicles and Subsidy (Low) 1,261 (1) 25,007 (1) 75 (1) 66 (1) 950 (1) (1) I3 Home Charging Availability (Medium) 1,265 (1) 24,420 (1) 75 (1) 66 (1) 974 (1) (1) P5 Other States Rebates (Medium) 1,171 (1) 22,663 (1) 67 (1) 60 (1) 902 (1) (1) I5 Work Charging Availability (High) 1,137 (1) 21,987 (1) 67 (1) 59 (1) 870 (1) (1) C1 Vehicle Manufacturer Cost (Low) 904 (1) 16,349 (1) 56 (1) 50 (1) 711 (1) (1) P4 State Rebate Duration (Medium) 722 (1) 13,910 (1) 41 (1) 37 (1) 557 (1) (1) P6 HOV Access Duration (Low) 674 (1) 12,636 (1) 36 (1) 32 (1) 532 (1) (1) C3 Diesel Fuel Prices (Low) 586 (0) 10,102 (0) 40 (1) 35 (1) 359 (0) (0)

51 Simulation Set ID Simulation Description NOx (%) CO (%) PM10 (%) PM2.5 (%) Total HCs (%) Averagea (%) I5 Work Charging Availability (Medium) 567 (0) 10,962 (0) 34 (0) 30 (0) 434 (0) (0) P5 Other States Rebates (Low) 64 (0) 1,145 (0) 3 (0) 3 (0) 53 (0) (0) P4 State Rebate Duration (Low) 55 (0) 984 (0) 3 (0) 2 (0) 46 (0) (0) P3 State Rebate Amount (High) 34 (0) 626 (0) 2 (0) 2 (0) 28 (0) (0) P3 State Rebate Amount (Medium) 15 (0) 281 (0) 1 (0) 1 (0) 13 (0) (0) I1 Public Charging Availability (Low) 8 (0) 157 (0) 1 (0) 0 (0) 6 (0) (0) P3 State Rebate Amount (Low) 6 (0) 106 (0) 0 (0) 0 (0) 5 (0) (0) I6 Work Charging Power (High) 0 (0) 7 (0) 0 (0) 0 (0) 0 (0) (0) I6 Work Charging Power (Medium) 0 (0) 7 (0) 0 (0) 0 (0) 0 (0) (0) I4 Home Charging Power (High) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) (0)

52 Simulation Set ID Simulation Description NOx (%) CO (%) PM10 (%) PM2.5 (%) Total HCs (%) Averagea (%) I4 Home Charging Power (Medium) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) (0) I1 Public Charging Power (Low) NAb NA NA NA NA NA I3 Home Charging Availability (Low) NA NA NA NA NA NA I4 Home Charging Power (Low) NA NA NA NA NA NA I5 Work Charging Availability (Low) NA NA NA NA NA NA a Order of data is based on average percentage reduction in emissions across pollutants. b NA indicates Not Applicable; no changes were made to default parameter values.

53 Table 13. Reduction in modeled light-duty passenger vehicle emissions (in tons) of four MSATs for calendar year 2040.a Simulation Set ID Simulation Description 1,3-Butadiene (%) Acetaldehyde (%) Acrolein (%) Benzene (%) Average (%) C1 Vehicle Manufacturer Cost (High) 103 (15) 351 (15) 20 (15) 585 (15) (15) C2 Gasoline Prices (High) 82 (12) 273 (11) 16 (11) 464 (11) (11) P5 Other States Rebates (High) 56 (8) 190 (8) 11 (8) 329 (8) (8) C2 Gasoline Prices (Medium) 49 (7) 158 (7) 9 (7) 276 (7) (7) C1 Vehicle Manufacturer Cost (Medium) 42 (6) 144 (6) 8 (6) 235 (6) (6) C2 Gasoline Prices (Low) 22 (3) 72 (3) 4 (3) 48 (3) (3) P6 HOV Access Duration (High) 21 (3) 71 (3) 4 (3) 122 (3) (3) I2 Public Charging Power (High) 15 (2) 52 (2) 3 (2) 88 (2) (2) P1 ARRA Max # Vehicles and Subsidy (High) 15 (2) 50 (2) 3 (2) 87 (2) (2) P2 ARRA Number of OEM Producers (High) 14 (2) 49 (2) 3 (2) 84 (2) (2)

54 Simulation Set ID Simulation Description 1,3-Butadiene (%) Acetaldehyde (%) Acrolein (%) Benzene (%) Average (%) P2 ARRA Number of OEM Producers (Medium) 14 (2) 49 (2) 3 (2) 84 (2) (2) P4 State Rebate Duration (High) 14 (2) 48 (2) 3 (2) 82 (2) (2) I2 Public Charging Power (Medium) 13 (2) 44 (2) 3 (2) 75 (2) (2) I3 Home Charging Availability (High) 13 (2) 44 (2) 3 (2) 73 (2) (2) P1 ARRA Max # Vehicles and Subsidy (Medium) 10 (2) 34 (1) 2 (2) 58 (1) (2) P6 HOV Access Duration (Medium) 9 (1) 32 (1) 2 (1) 55 (1) (1) I1 Public Charging Availability (High) 9 (1) 31 (1) 2 (1) 51 (1) (1) P2 ARRA Number of OEM Producers (Low) 9 (1) 30 (1) 2 (1) 52 (1) (1) C3 Diesel Fuel Prices (High) 6 (1) 27 (1) 1 (1) 38 (1) (1) I1 Public Charging Availability (Medium) 7 (1) 23 (1) 1 (1) 39 (1) (1)

55 Simulation Set ID Simulation Description 1,3-Butadiene (%) Acetaldehyde (%) Acrolein (%) Benzene (%) Average (%) I3 Home Charging Availability (Medium) 6 (1) 22 (1) 1 (1) 36 (1) (1) P1 ARRA Max # Vehicles and Subsidy (Low) 6 (1) 21 (1) 1 (1) 36 (1) (1) C3 Diesel Fuel Prices (Medium) 5 (1) 23 (1) 1 (1) 33 (1) (1) P5 Other States Rebates (Medium) 6 (1) 20 (1) 1 (1) 35 (1) (1) I5 Work Charging Availability (High) 6 (1) 19 (1) 1 (1) 32 (1) (1) C1 Vehicle Manufacturer Cost (Low) 5 (1) 16 (1) 1 (1) 25 (1) (1) P4 State Rebate Duration (Medium) 4 (1) 12 (1) 1 (1) 22 (1) (1) P6 HOV Access Duration (Low) 4 (1) 12 (1) 1 (1) 21 (1) (1) I5 Work Charging Availability (Medium) 3 (0) 10 (0) 1 (0) 16 (0) (0) C3 Diesel Fuel Prices (Low) 2 (0) 9 (0) 1 (0) 13 (0) (0) P5 Other States Rebates (Low) 0 1 0 2

56 Simulation Set ID Simulation Description 1,3-Butadiene (%) Acetaldehyde (%) Acrolein (%) Benzene (%) Average (%) (0) (0) (0) (0) (0) P4 State Rebate Duration (Low) 0 (0) 1 (0) 0 (0) 2 (0) (0) P3 State Rebate Amount (High) 0 (0) 1 (0) 0 (0) 1 (0) (0) P3 State Rebate Amount (Medium) 0 (0) 0 (0) 0 (0) 1 (0) (0) I1 Public Charging Availability (Low) 0 (0) 0 (0) 0 (0) 0 (0) (0) P3 State Rebate Amount (Low) 0 (0) 0 (0) 0 (0) 0 (0) (0) I4 Home Charging Power (High) 0 (0) 0 (0) 0 (0) 0 (0) (0) I4 Home Charging Power (Medium) 0 (0) 0 (0) 0 (0) 0 (0) (0) I6 Work Charging Power (High) 0 (0) 0 (0) 0 (0) 0 (0) (0) I6 Work Charging Power (Medium) 0 (0) 0 (0) 0 (0) 0 (0) (0) I1 Public Charging Power (Low) NAb NA NA NA NA

57 Simulation Set ID Simulation Description 1,3-Butadiene (%) Acetaldehyde (%) Acrolein (%) Benzene (%) Average (%) I3 Home Charging Availability (Low) NA NA NA NA NA I4 Home Charging Power (Low) NA NA NA NA NA I5 Work Charging Availability (Low) NA NA NA NA NA I6 Work Charging Power (Low) NA NA NA NA NA a Order of data is based on average percentage reduction in emissions across pollutants. b NA indicates Not Applicable; no changes were made to default parameter values.

58 Table 14. Reduction in modeled light-duty passenger vehicle emissions (in tons) of remaining MSATs for calendar year 2040.a Simulation Set ID Simulation Description Ethylbenzene (%) Formaldehyde (%) Naphthalene gas (%) Naphthalene particle (%) Average (%) C1 Vehicle Manufacturer Cost (High) 226 (15) 206 (15) 34 (15) 0 (16) (15) C2 Gasoline Prices (High) 179 (12) 147 (11) 26 (11) 0 (11) (11) P5 Other States Rebates (High) 127 (8) 110 (8) 19 (8) 0 (9) (8) C2 Gasoline Prices (Medium) 107 (7) 89 (7) 16 (7) 0 (7) (7) C1 Vehicle Manufacturer Cost (Medium) 91 (6) 85 (6) 14 (6) 0 (7) (6) C2 Gasoline Prices (Low) 48 (3) 41 (3) 7 (3) 0 (3) (3) P6 HOV Access Duration (High) 47 (3) 40 (3) 7 (3) 0 (3) (3) I2 Public Charging Power (High) 34 (2) 30 (2) 5 (2) 0 (2) (2) P1 ARRA Max # Vehicles and Subsidy (High) 33 (2) 29 (2) 5 (2) 0 (2) (2) P2 ARRA Number of OEM Producers (High) 33 (2) 28 (2) 5 (2) 0 (2) (2)

59 Simulation Set ID Simulation Description Ethylbenzene (%) Formaldehyde (%) Naphthalene gas (%) Naphthalene particle (%) Average (%) P2 ARRA Number of OEM Producers (Medium) 33 (2) 28 (2) 5 (2) 0 (2) (2) P4 State Rebate Duration (High) 32 (2) 28 (2) 5 (2) 0 (2) (2) I2 Public Charging Power (Medium) 29 (2) 26 (2) 4 (2) 0 (2) (2) I3 Home Charging Availability (High) 28 (2) 25 (2) 4 (2) 0 (2) (2) P1 ARRA Max # Vehicles and Subsidy (Medium) 23 (1) 20 (2) 3 (2) 0 (2) (2) P6 HOV Access Duration (Medium) 21 (1) 18 (1) 3 (1) 0 (1) (1) I1 Public Charging Availability (High) 20 (1) 18 (1) 3 (1) 0 (1) (1) P2 ARRA Number of OEM Producers (Low) 20 (1) 17 (1) 3 (1) 0 (1) (1) C3 Diesel Fuel Prices (High) 15 (1) 17 (1) 2 (1) 0 (1) (1) C3 Diesel Fuel Prices (Medium) 13 (1) 15 (1) 2 (1) 0 (1) (1)

60 Simulation Set ID Simulation Description Ethylbenzene (%) Formaldehyde (%) Naphthalene gas (%) Naphthalene particle (%) Average (%) I1 Public Charging Availability (Medium) 15 (1) 13 (1) 2 (1) 0 (1) (1) I3 Home Charging Availability (Medium) 14 (1) 13 (1) 2 (1) 0 (1) (1) P1 ARRA Max # Vehicles and Subsidy (Low) 14 (1) 12 (1) 2 (1) 0 (1) (1) P5 Other States Rebates (Medium) 14 (1) 11 (1) 2 (1) 0 (1) (1) I5 Work Charging Availability (High) 12 (1) 11 (1) 2 (1) 0 (1) (1) C1 Vehicle Manufacturer Cost (Low) 10 (1) 9 (1) 2 (1) 0 (1) (1) P4 State Rebate Duration (Medium) 8 (1) 7 (1) 1 (1) 0 (1) (1) P6 HOV Access Duration (Low) 8 (1) 7 (1) 1 (1) 0 (1) (1) I5 Work Charging Availability (Medium) 6 (0) 6 (0) 1 (0) 0 (0) (0) C3 Diesel Fuel Prices (Low) 5 (0) 6 (0) 1 (0) 0 (1) (0)

61 Simulation Set ID Simulation Description Ethylbenzene (%) Formaldehyde (%) Naphthalene gas (%) Naphthalene particle (%) Average (%) P5 Other States Rebates (Low) 1 (0) 1 (0) 0 (0) 0 (0) (0) P4 State Rebate Duration (Low) 1 (0) 1 (0) 0 (0) 0 (0) (0) P3 State Rebate Amount (High) 1 (0) 0 (0) 0 (0) 0 (0) (0) P3 State Rebate Amount (Medium) 0 (0) 0 (0) 0 (0) 0 (0) (0) I1 Public Charging Availability (Low) 0 (0) 0 (0) 0 (0) 0 (0) (0) P3 State Rebate Amount (Low) 0 (0) 0 (0) 0 (0) 0 (0) (0) I4 Home Charging Power (High) 0 (0) 0 (0) 0 (0) 0 (0) (0) I4 Home Charging Power (Medium) 0 (0) 0 (0) 0 (0) 0 (0) (0) I6 Work Charging Power (High) 0 (0) 0 (0) 0 (0) 0 (0) (0) I6 Work Charging Power (Medium) 0 (0) 0 (0) 0 (0) 0 (0) (0) I1 Public Charging Power (Low) NAb NA NA NA NA

62 Simulation Set ID Simulation Description Ethylbenzene (%) Formaldehyde (%) Naphthalene gas (%) Naphthalene particle (%) Average (%) I3 Home Charging Availability (Low) NA NA NA NA NA I4 Home Charging Power (Low) NA NA NA NA NA I5 Work Charging Availability (Low) NA NA NA NA NA I6 Work Charging Power (Low) NA NA NA NA NA a Order of data is based on average percentage reduction in emissions across pollutants. b NA indicates Not Applicable; no changes were made to default parameter values.

63 Table 15. Reduction in modeled light-duty passenger vehicle emissions of GHGs (million metric tons CO2; metric tons CH4 and N2O) for calendar year 2040.a Simulation Set ID Simulation Description CO2 (%) CH4 (%) N2O (%) Average (%) Equivalent Passenger Vehicles’ Annual GHG Emissions (# vehicles)b C1 Vehicle Manufacturer Cost (High) 99 (23) 1,581 (21) 2,291 (20) (21) 21,255,112 C2 Gasoline Prices (High) 64 (15) 1,018 (14) 1,551 (14) (14) 13,779,446 C1 Vehicle Manufacturer Cost (Medium) 49 (11) 784 (11) 1,081 (10) (10) 10,431,866 C2 Gasoline Prices (Medium) 40 (9) 638 (9) 955 (8) (9) 8,578,846 P5 Other States Rebates (High) 39 (9) 610 (8) 995 (9) (9) 8,335,632 C2 Gasoline Prices (Low) 19 (4) 305 (4) 449 (4) (4) 4,081,797 P6 HOV Access Duration (High) 14 (3) 222 (3) 358 (3) (3) 2,943,064 I3 Home Charging Availability (High) 14 (3) 218 (3) 306 (3) (3) 2,897,350 I2 Public Charging Power (High) 13 (3) 210 (3) 314 (3) (3) 2,822,314 I2 Public Charging Power (Medium) 12 (3) 183 (3) 272 (2) (3) 2,460,547 P1 ARRA Max # Vehicles and Subsidy (High) 11 (3) 171 (2) 278 (3) (2) 2,357,565 P4 State Rebate Duration (High) 10 (2) 160 (2) 255 (2) (2) 2,173,409

64 Simulation Set ID Simulation Description CO2 (%) CH4 (%) N2O (%) Average (%) Equivalent Passenger Vehicles’ Annual GHG Emissions (# vehicles)b P2 ARRA Number of OEM Producers (High) 9 (2) 146 (2) 237 (2) (2) 1,902,441 P2 ARRA Number of OEM Producers (Medium) 9 (2) 146 (2) 237 (2) (2) 1,902,441 I1 Public Charging Availability (High) 8 (2) 133 (2) 195 (2) (2) 1,792,708 P1 ARRA Max # Vehicles and Subsidy (Medium) 8 (2) 116 (2) 190 (2) (2) 1,601,618 C3 Diesel Fuel Prices (High) 7 (2) 114 (2) 167 (1) (2) 1,468,296 I3 Home Charging Availability (Medium) 7 (2) 107 (1) 150 (1) (1) 1,424,457 I1 Public Charging Availability (Medium) 7 (2) 103 (1) 150 (1) (1) 1,383,304 C3 Diesel Fuel Prices (Medium) 6 (1) 101 (1) 147 (1) (1) 1,304,147 P6 HOV Access Duration (Medium) 6 (1) 96 (1) 151 (1) (1) 1,219,680 I5 Work Charging Availability (High) 6 (1) 95 (1) 133 (1) (1) 1,260,626 C1 Vehicle Manufacturer Cost (Low) 6 (1) 97 (1) 126 (1) (1) 1,261,335

65 Simulation Set ID Simulation Description CO2 (%) CH4 (%) N2O (%) Average (%) Equivalent Passenger Vehicles’ Annual GHG Emissions (# vehicles)b P2 ARRA Number of OEM Producers (Low) 5 (1) 90 (1) 143 (1) (1) 1,156,679 P1 ARRA Max # Vehicles and Subsidy (Low) 5 (1) 73 (1) 120 (1) (1) 1,011,936 P5 Other States Rebates (Medium) 4 (1) 64 (1) 102 (1) (1) 836,727 I5 Work Charging Availability (Medium) 3 (1) 47 (1) 67 (1) (1) 629,334 C3 Diesel Fuel Prices (Low) 3 (1) 43 (1) 61 (1) (1) 549,532 P4 State Rebate Duration (Medium) 2 (1) 39 (1) 62 (1) (1) 506,842 P6 HOV Access Duration (Low) 2 (1) 35 (1) 53 (1) (1) 427,714 P5 Other States Rebates (Low) 0 (0) 3 (0) 4 (0) (0) 34,192 P4 State Rebate Duration (Low) 0 (0) 3 (0) 4 (0) (0) 28,752 P3 State Rebate Amount (High) 0 (0) 2 (0) 3 (0) (0) 18,655 I1 Public Charging Availability (Low) 0 (0) 1 (0) 1 (0) (0) 9,003 P3 State Rebate Amount (Medium) 0 (0) 1 (0) 1 (0) (0) 8,335 P3 State Rebate Amount (Low) 0 (0) 0 (0) 0 (0) (0) 3,130

66 Simulation Set ID Simulation Description CO2 (%) CH4 (%) N2O (%) Average (%) Equivalent Passenger Vehicles’ Annual GHG Emissions (# vehicles)b I6 Work Charging Power (High) 0 (0) 0 (0) 0 (0) (0) 325 I6 Work Charging Power (Medium) 0 (0) 0 (0) 0 (0) (0) 325 I4 Home Charging Power (High) 0 (0) 0 (0) 0 (0) (0) 0 I4 Home Charging Power (Medium) 0 (0) 0 (0) 0 (0) (0) 0 I1 Public Charging Power (Low) NAc NA NA NA NA I3 Home Charging Availability (Low) NA NA NA NA NA I4 Home Charging Power (Low) NA NA NA NA NA I5 Work Charging Availability (Low) NA NA NA NA NA I6 Work Charging Power (Low) NA NA NA NA NA a Order of data is based on average percentage reduction in emissions across pollutants. b Total GHG emissions are equivalent to the amount from the listed number of passenger vehicles driven for one year (https://www.epa.gov/energy/greenhouse-gas-equivalencies-calculator). c NA indicates Not Applicable; no changes were made to default parameter values. 4.6 SUMMARY OF MODELING RESULTS, MAJOR LIMITATIONS, AND IMPLICATIONS A total of 49 simulations were performed; one simulation for the Base Case scenario 1, and three sensitivity simulations for scenarios 2 through 4. Nationwide ZEV populations used in the MOVES2014b modeling were estimated by the MA3T model V20190404. The relative increases in ZEV populations modeled by MA3T for calendar year 2040 are generally consistent with findings in the literature review. The literature shows that purchase cost is the greatest barrier to ZEV adoption, and the single most

67 important determinant of purchase cost is battery cost; MA3T modeled consumer preference consistently with these findings. The largest ZEV population increases modeled with MA3T (and corresponding reductions in emissions) are for the cost parity scenarios in which (1) ZEV manufacturer costs reach parity with conventional vehicles in 2030, 2035, and 2040; and (2) gasoline prices increase at accelerated rates between 2019 and 2040. However, while estimated emissions decreased with increased ZEV population compared to the Base Case, the level of decrease was not as great as the change in total ZEV population. For example, a large decrease in the vehicle manufacturer cost of ZEVs (the High case) caused the modeled population of ZEVs to increase from 9% in the Base Case to 38% of the total light-duty vehicle population in 2040. This change in the overall light-duty vehicle fleet composition resulted in a 23% reduction in modeled CO2 emissions. That change in emissions is roughly equal to the increase in the share of ZEVs in the modeled vehicle fleet (29%). The difference between the two changes could be attributed to fleet turnover effects, such as consumers choosing not to buy a new car and keeping an older higher-emitting car. Although financial incentives like tax credits and rebates contribute to the reduction of overall ZEV cost, the MA3T model generally estimated moderate increases in the ZEV population as a result of changes to those inputs. The corresponding MOVES-estimated emissions reductions for the incentives/policy scenarios were also moderate. The models showed larger increases in ZEV population and reductions in emissions for the scenarios in which rebates with long duration were applied to states without rebates as of 2019. Findings in the literature review suggest that consumer education plays an important role in making rebates and tax credits effective. It is possible that, the longer a credit or rebate applies, the greater its effectiveness will be, as consumers have a longer period to learn about rebate and tax credit programs and consider how those will affect their net cost for ZEV vehicle ownership relative to a conventional vehicle. The literature review also showed that HOV lane access has a strong influence on ZEV adoption. The findings in this analysis suggest that, for longer periods of HOV lane access, there may be a moderate increase in ZEV adoption and corresponding reduction in emissions. Driving range and related infrastructure were identified as key factors of ZEV adoption in the literature review. Findings from the infrastructure scenarios confirmed these factors, with moderate modeled increases in ZEV populations and reductions in emissions. The relative magnitude and direction of modeled changes in ZEV populations and light-duty vehicle exhaust emissions presented in this report can provide insight into the factors that affect ZEV adoption. The qualitative impacts on emissions across all simulation sets are presented in Table 16. However, the results for ZEV population growth and emissions reduction should be interpreted with caution. There is substantial uncertainty surrounding the modeled outcomes presented here. The main objective of this work was not so much to forecast a precise expected future fraction of the vehicle fleet that will become ZEVs, as to enable readers to understand what key factors influence the degree to which ZEVs will penetrate the vehicle fleet. Although MA3T is a highly detailed consumer choice model, it does have limitations. As described in Lin and Greene 2009, which first introduced the model formulation, MA3T uses some data that may not yet exist, and represents consumer behavior that may not be fully understood. Plausible assumptions are made for data that are lacking or uncertain. Also, some attributes that vary by market segment and over time are not represented as such.

68 Another example limitation is the estimation of the distribution of daily vehicle use for which longitudinal travel data are not available. The current version of MA3T (V20190404) also assumes that no FCEVs are on the market within its modeling period (2005-2050). However, FCEVs are forecast to make up only 2% of the ATV market and less than half a percent of the total light-duty market in 2040 (U.S. EIA 2019a). The model developers have updated the model over time and acknowledge the need for further research and data to continue model updates. Finally, these are vehicle exhaust emissions only, and do not include wheel-to-well emissions associated with the source of power used to charge the vehicle. This should be considered when assessing the modeled emission reductions in this study. Table 16. Qualitative impacts of modeled cost parity, policy, and infrastructure changes in the scenario simulations. Scenario Description Qualitative Emissions Impact Vehicle Manufacturer Cost Parity (Cost Parity) High Gasoline Price Increase (Cost Parity) High Instant Rebate with Long Duration Applied to Other States (Policy) High HOV Lane Access with Long Duration (Policy) Moderate Home Charging Availability Increase (Infrastructure) Moderate Public Charging Power Level Increase (Infrastructure) Moderate ARRA Vehicle Cap Increase and Maximum Subsidy Adjustment (Policy) Moderate State Rebate Duration Increase (Policy) Moderate ARRA Number of OEM Producers Increase (Policy) Moderate Public Charging Availability Increase (Infrastructure) Low Diesel Price Increase (Cost Parity) Low Workplace Charging Availability Increase (Infrastructure) Low Instant Rebate Amount Increase (Policy) Negligible Home Charging Power Level Increase (Infrastructure) Negligible Workplace Charging Power Level Increase (Infrastructure) Negligible

69 Combinations of parameter adjustments in MA3T that lead to increased ZEV adoption could lead to greater ZEV adoption outcomes than could be achieved by adjusting a single parameter, depending on the magnitude of increased ZEV adoption from the individual parameters. The results of combined factors are not necessarily additive, since MA3T includes important feedback mechanisms that result in a synergistic effect on ZEV adoption. Modeling combined scenarios was beyond the scope of this study but should be considered for future research. The sensitivity testing completed here identified the relative importance of the various factors represented in the modeling scenarios. The findings suggest that, to support efforts to promote adoption of ZEVs, state DOTs and MPOs should consider the following (listed in order of importance): • Cost Parity simulations, specifically the gasoline price and the High and Medium cases of the vehicle manufacturer cost simulations, resulted in the greatest reductions of light-duty vehicle emissions. • Long-duration rebates in states that do not have a rebate (as of 2019) resulted in the next highest reductions of light-duty vehicle emissions. • The greatest change in HOV lane access (i.e., extended access to HOV lanes from 2014‒2030) had the next highest emissions reductions. In order to act on these findings, in some instances DOTs and MPOs may act independently. Expanding infrastructure necessary for ZEVs is one such example; for example, as of 2019, the California Department of Transportation (Caltrans) was in the process of building electric vehicle charging infrastructure (California Department of Transportation, 2019). DOTs and MPOs may also consider expanding consumer and dealer education programs to increase awareness of rebates and incentives, which was found to be an important driver of ZEV adoption in the literature review although unquantified through MA3T. In fact, some state DOTs are integrating consumer education into their plans, including the North Carolina Department of Transportation (NCDOT), which highlights opportunities to promote public awareness and education about electric vehicles in their recent draft Zero-Emission Vehicle Plan (North Carolina Department of Transportation, 2019). Establishing a policy of allowing ZEVs free or discounted access to HOV lanes is another mechanism for encouraging customers to consider purchasing ZEVs, particularly when that policy offers ZEVs free or discounted HOV access for a long period of time. However, for many of these findings, DOTs and MPOs may need to work collaboratively with other agencies and organizations (e.g., state policymakers, state environmental protection agencies, and state resource agencies). For example, DOTs, through coordinated work with sister agencies such as state departments of motor vehicles, may work to effectively increase the cost of purchasing an ICEV relative to a ZEV by increasing registration fees on vehicles with higher tailpipe emissions. To promote cost parity between ZEVs and their ICEV counterparts, planning agencies and their partners could also implement incentives and rebates that effectively lower both production costs for manufacturers and purchase costs for consumers. Findings in this research also show that increasing the price of fuel for ICEVs also makes ZEVs more attractive to consumers; therefore, increasing the gas tax may be one approach to encouraging ZEV adoption. By jointly setting emission and air quality goals with other state and local agencies, DOTs and MPOs may put to use a wide range of tools to effect a meaningful reduction in harmful emissions.