Incentivizing Alternative Fuel Vehicle Transactions: Analysis of Cash-for-Clunkers Transactions for New Alternative Fuel Vehicles

Edmund Zolnik ^*

Schar School of Policy and Government, George Mason University, Arlington, VA 22201 USA

* Correspondence: Edmund Zolnik

Academic Editor: Islam Md Rizwanul Fattah

Special Issue: Public Policy and the Development and Adoption of Alternative Energy Technologies

Received: January 14, 2022 | Accepted: July 28, 2022 | Published: August 02, 2022

Journal of Energy and Power Technology 2022, Volume 4, Issue 3, doi:10.21926/jept.2203026

Recommended citation: Zolnik E. Incentivizing Alternative Fuel Vehicle Transactions: Analysis of Cash-for-Clunkers Transactions for New Alternative Fuel Vehicles. Journal of Energy and Power Technology 2022; 4(3): 026; doi:10.21926/jept.2203026.

© 2022 by the authors. This is an open access article distributed under the conditions of the Creative Commons by Attribution License, which permits unrestricted use, distribution, and reproduction in any medium or format, provided the original work is correctly cited.

1. Introduction

One manifestation of the macroeconomic shock from the COVID-19 pandemic is low private vehicle sales. Total sales¹ in April of 2020 (9.1 million) equal the low in the United States since January of 1976 [1] (Figure 1). Year-over-year sales in April of 2020 are also low worldwide [2]. Contraction in an industry so vital to past, present, and future economic growth is the impetus for calls to reintroduce programs to incentivize private vehicle transactions. In the United States, Senator Chuck Schumer (D-NY), Senator Debbie Stabenow (D-MI), Senator Sherrod Brown (D-OH), and Senator Jeff Merkley (D-OR) propose a program to incentivize private vehicle transactions from gasoline power to zero emission known as Clean Cars for America [3]. Such a program is reminiscent of the private vehicle retirement program in the United States known as Cash for Clunkers. In the short term, the goals of the programs are the same; that is, to incentivize transactions so as to decrease greenhouse gas (GHG) emissions from the private vehicle fleet. In the long term, however, the goal of the latter program is different than the goal of the former program; that is, to incentivize transactions so as to accelerate the transition of the private vehicle fleet from internal combustion engine power to battery power.

Figure 1 Total sales (1E+6) by month from January of 1976 to October of 2020 in the United States with lowest-sales (9.1) months (December of 1981 and April of 2020) and with highest-sales (22.1) month (October of 2001).

Support from stakeholders worldwide for Clean Cars for America speaks to the potential of battery power. However, the diffusion of new technology is typically not even spatially or temporally, especially new technology for private vehicles. Further, such programs may not accelerate the transition of the private vehicle fleet because the target of the policy is wrong. The argument relates to the paradox of a policy to economically incentivize transactions when social behavior at the point of sale renders such incentives irrelevant to consumers. Indeed, first adopters already possess the resources to absorb any potential financial losses due to the inherently uncertain viability of new technology [4]. The shift of the policy target from the vehicle to the consumer harkens to past efforts by social scientists to refocus attention on how individuals behave economically and socially [5]. Regarding the latter, such programs to accelerate the retirement of older private vehicles engender resistance from individuals who object to government efforts to enforce the obsolescence of classics worthy of preservation, not clunkers worthy of retirement [6].

To test the viability of a national retirement program to transition the private vehicle fleet from fossil fuels to alternative fuels², the study analyzes a national retirement program to incentivize private vehicle transactions from 2009. Analysis of the national retirement program highlights the transactions where the trade-in vehicle uses fossil fuels and the new vehicle uses alternative fuels. Analysis of the national retirement program also highlights who adopts the new technology and where the new technology is popular. The questions the analysis answers are as follows. First, what are the economic costs as well as the environmental benefits of the monetary incentives for new alternative fuel vehicles (AFVs) in the national retirement program? Second, how effective are vouchers to incentivize transactions for new AFVs in each state?

The following section reviews the empirical literature on vehicle retirement programs.

2. Background

2.1 Clean Air Act Amendments of 1990

The Clean Air Act Amendments of 1990 (Public Law 101-549) identify a "program to encourage the voluntary removal from use and the marketplace of pre-1980 model year light duty vehicles and pre-1980 model light duty trucks" (p. 2466) as one of sixteen control measures to mitigate emissions. Guidance from the United States Environmental Protection Agency [7] highlights the generation of mobile-source emission reduction credits (MERCs) from a market-based, control measure to accelerate the retirement of vehicles. Such guidance encourages potential public sponsors and/or potential private sponsors to target high-emitter vehicles and to target vehicles with obsolete emission-control technology so as to maximize cost effectiveness. However, the objective is to assist program sponsors in the design of flexible retirement programs to best suit the sponsors’ goals within the context of sound policy. To that end, sponsors can design emission-limiting programs or market-response programs. In the former, a sponsor retires vehicles until the program achieves a specific emission reduction target. In the latter, a sponsor incentivizes retirements without a specific emission reduction target. Either way, public sponsors and/or private sponsors incentivize retirements so as to generate an emission credit equivalent to the emission difference between the new vehicle and the old vehicle.

2.2 Consumer Assistance to Recycle and Save Act of 2009

The Car Allowance Rebate System (CARS), also known as Cash for Clunkers, is an example of a market-response, retirement program in the United States. The enactment of the Consumer Assistance to Recycle and Save Act of 2009 (Public Law 101-549) on June 24^th, 2009 directs the Secretary of Transportation to establish and to administer CARS thru the National Highway Traffic Safety Administration (NHTSA). CARS incentivizes purchase transactions or lease transactions via vouchers dependent on the fuel economy difference from the new vehicle to the trade-in vehicle. The total appropriation for CARS transactions was three billion dollars. A total of 677,842 voucher applications were paid from July 1^st, 2009 to August 25^th, 2009. The sum of the paid voucher applications was \$2.9 billion. The mean paid voucher was \$4,209.00.

Total (purchase plus lease) transactions from CARS was 677,842:401,274 automobiles; 274,602 light trucks; and 1,966 heavy trucks. Mean combined fuel economy of new vehicles was 24.9 miles per gallon. Mean combined fuel economy of trade-in vehicles was 15.8 miles per gallon. The mean combined fuel economy difference (from the new vehicle to the trade-in vehicle) per transaction was 9.2 miles per gallon. NHTSA estimates CARS decreases annual fuel consumption by approximately thirty-three million gallons. NHTSA also estimates CARS decreases quarter-century carbon dioxide (CO₂) emissions by nine million metric tons.

2.3 Literature Review

Table 1 lists analyses on the effectiveness of vehicle retirement programs. The list is long, but not exhaustive. Rather, the list represents a sample of analyses on the effectiveness of vehicle retirement programs: prospective [8,9,10,11,12]; and retrospective [13,14,15,16,17,18]. Analyses on the optimal subsidy for vehicle retirement programs [19,20] are not on the list. A review of the analyses in Table 1 reveals the following. First, variation in program design is evident. Nonetheless, the majority are market-response programs without a specific emission reduction target. Second, the majority of programs are local in scale or regional in scale [18]. To that end, Retiring Old Cars: Programs To Save Gasoline and Reduce Emissions from the Office of Technology Assessment [8] is a rare example of a prospective analysis to estimate the effectiveness of a national program to retire one million vehicles in the United States. Third, cost-benefit analyses of vehicle retirement programs are rare [9,10,11,12,18].

Table 1 Analyses of vehicle retirement programs.

In light of the above on the effectiveness of vehicle retirement programs, Lessons Learned Cash for Clunkers Program [21] and Subsidizing Replacement of Motor Vehicles: An Analysis of Cash for Clunkers Program [22] audit the performance of a national vehicle retirement program in the United States. Explicit program objectives are not evident in the Consumer Assistance to Recycle and Save Act of 2009, but implicit program objectives are: to stimulate private vehicle sales; and to increase private-vehicle fuel economy. The performance audits suggest Cash for Clunkers was somewhat successful with regard to the former implicit program objective. Higher monthly, private vehicle sales in July of 2009 and in August of 2009 than in January of 2009, in February of 2009, in March of 2009, in April of 2009, in May of 2009, or in June of 2009 are testament to the economic benefit of the incentive program. The program audit also suggests Cash for Clunkers was successful with regard to the latter implicit program objective. The mean combined fuel economy difference per transaction (9.2 miles per gallon) is testament to the environmental benefit of the incentive program.

3. Methodology and Data

3.1 Methodology

The methodology to explore CO₂ emission reduction and CO₂ emission reduction cost from CARS transactions for new AFVs is known as a multilevel model. The advantages of a multilevel methodology for such an exploration are as follows. First, a multilevel model nests micro-level events within macro-level units of analysis. Second, a two-level model, such as in the study, estimates two sets of parameters. The first set of parameters summarizes the average relationship between CO₂ emission reduction or CO₂ emission reduction cost and the state-level independent variables known to affect CARS transactions for new AFVs. The second set of parameters summarize the variation in the average relationship between CO₂ emission reduction or CO₂ emission reduction cost and the state-level independent variables known to affect CARS transactions for new AFVs.

The multilevel models in the study nest micro-level transactions (t) within macro-level states (s) [23]. In each macro-level unit of analysis, the micro-level dependent variable (Y_ts) is a function of micro-level independent variables as well as a micro-level error term (r_ts)

\[ Y_{t s}=\beta_{0 s}+\beta_{1 s} X_{1 t s}+\beta_{2 s} X_{2 t s}+\cdots+\beta_{\mathrm{Ps}} X_{\mathrm{Pts}}+r_{t s} \tag{1} \]

where: Y_ts is CO₂ emission reduction or CO₂ emission reduction cost for transaction t in state s; β_Ps are (p = 0, 1, 2, …, P) transaction-level coefficients; X_Pts is the transaction-level independent variable P for transaction t in state s; and r_ts is the transaction-level error term. The variance of the transaction-level error term (r_ts) is σ². In a two-level model, such as in the study, the y-intercept and the coefficients at the micro-level are fixed (invariant at the macro-level) or random (variant at the macro-level). The two-level model in the study is known as a random-intercept model—the y-intercept is random, but the coefficients are fixed. The model for variation in CO₂ emission reduction or in CO₂ emission reduction cost between states is as follows

\[ \beta_{0 s}=\gamma_{00}+\gamma_{01} W_{1 s}+\gamma_{02} W_{2 s}+\cdots+\gamma_{0 \mathrm{Q}} W_{\mathrm{Q} s}+u_{0 s} \tag{2} \]

where: γ₀₀ is the y-intercept for the transaction effect β_0s; γ_0Q are (q = 1, 2, 3, …, Q) state-level coefficients; W_Qs is the state-level independent variable Q in state s; and u_0s is the state-level error term. The variance of the state-level error term (u_0s) is $\tau$₀₀.

The multilevel models in the study are fit with HLM software, Version 8.2.1.13 of HLM for Windows [24].

3.2 Data

The transaction-level dependent variables and the transaction-level independent variables in the multilevel models are from CARS.

The transaction-level dependent variables are as follows. CO₂ Emission Reduction and CO₂ Emission Reduction Cost represent the CO₂ emissions decrease and the cost of the CO₂ emissions decrease from the trade-in vehicle to the new vehicle where the trade-in vehicle uses fossil fuels and the new vehicle uses alternative fuels.³

The transaction-level independent variables are as follows. Hybrid is 1 if the new vehicle is a Hybrid Electric Vehicle (HEV), 0 otherwise. Invoice Amount is a voucher of \$3,500.00 or a voucher of \$4,500.00. New Vehicle MSRP (Manufacturer Suggested Retail Price) is the base MSRP for the new vehicle up to the price limit of \$45,000.00. New Vehicle MSRP tests if price is a barrier to emission reduction—if price is higher, then emission reduction is greater. Trade-in Vehicle Category-New Vehicle Category are retirement-eligible, vehicle-category pairs. Vehicles (trade-in or new) are Passenger Cars, Category 1 Trucks, or Category 2 Trucks.⁴ Passenger Cars excludes vehicles whose manufacture is not primarily for the transport of persons and vehicles capable of off-highway operation. Category 1 Trucks includes sport utility vehicles, small pickup trucks, medium pickup trucks, small passenger vans, medium passenger vans, small cargo vans, and medium cargo vans. Category 2 Trucks includes large pickup trucks and large vans.

The state-level independent variables in the multilevel model are as follows. CO₂ Emissions are total CO₂ emissions from the transportation sector in 2009 [29].⁵ CO₂ Emissions test if CARS is more effective in higher-emission states where the marginal benefits of retirement are presumably greater [10]. Gasoline Price is the mean of the retail prices of regular gasoline in June of 2009 [30], in July of 2009 [31], and in August of 2009 [32]. Gasoline Tax is the mean of the tax on gasoline in June of 2009 [30], in July of 2009 [31], and in August of 2009 [32].⁶ Gasoline Price and Gasoline Tax test if CARS is less effective in lower-gasoline price states where the benefits of retirement are presumably lower [33]. Vehicle Registrations is the sum of the number of private vehicle registrations in 2009 plus the number of commercial vehicle registrations in 2009 [34].⁷ Vehicle Registrations controls for the number of retirement-eligible vehicles in each state. Vehicle Type is the location quotient for Passenger Cars, for Category 1 Trucks, and for Category 2 Trucks.⁸ If the location quotient equals 1.00, then the incidence of a vehicle type is the same in the state as in the United States. Vehicle Type, therefore, controls for the incidence of different types of retirement-eligible vehicles in each state.

The code for the variables in the study is in SAS software, Version 9.4 of the SAS System for Windows.⁹

4. Results

4.1 CO₂ Emission Reduction versus CO₂ Emission Reduction Cost

Table 2 ranks mean CO₂ emission reduction in each state (from highest to lowest) and ranks mean CO₂ emission reduction cost in each state (from lowest to highest). Table 2 also highlights the top ten states as well as the bottom ten states. The five left columns in Table 2 rank mean CO₂ emission reduction in each state and the five right columns in Table 2 rank mean CO₂ emission reduction cost in each state. The last row provides column sums (Pounds (1E+3) in the second column, Transactions in the third column, Dollars in the seventh column, and Transactions in the eighth column) or column means (Pounds (1E+3) per Transactions in the fourth column and Dollars per Transaction in the ninth column) where appropriate.

Table 2 Ranks of CO₂ emission reduction in each state (highest to lowest) and ranks of CO₂ emission reduction cost in each state (lowest to highest).

The geographic distributions of top-rank states versus the geographic distribution of bottom-rank states on CO₂ emission reduction and on CO₂ emission reduction cost are clearly different. The top-rank states are mostly on the periphery (Hawaii, Connecticut, and Oregon) and more populous (California and New York), while the bottom-rank states are mostly in the core (Nebraska and North Dakota) and less populous (Vermont and South Dakota). The top-rank state (Hawaii) is also clearly different from the bottom-rank states (Nebraska and North Dakota). First, the majority of new vehicles in Hawaii transactions are HEVs (52.63%), while the minority of new vehicles in Nebraska transactions are HEVs (2.63%) and the minority of new vehicles in North Dakota transactions are HEVs (2.72%). Second, most of the Hawaii transactions (31.58%) are Category 1 Truck-Passenger Car, while most of the Nebraska transactions (64.55%) and most of the North Dakota transactions (62.93%) are Category 2 Truck-Category 2 Truck. Third, invoice amounts are higher for Hawaii transactions (\$4,342.66) than for Nebraska transactions (\$4,189.28) or for North Dakota transactions (\$4,180.27) which follows from the fact that the majority of new vehicles in Hawaii transactions are Passenger Cars.

4.2 Multilevel Models

Table 3 presents descriptive statistics for the transaction-level dependent variables and the transaction-level independent variables as well as the state-level independent variables. Table 4 presents results from a multilevel model with CO₂ Emission Reduction as the dependent variable (left column) and a multilevel model with CO₂ Emission Reduction Cost as the dependent variable (right column). The left column of Table 4 lists the coefficient estimates and the y-intercept estimate for the CO₂ Emission Reduction multilevel model. The right column of Table 4 lists the coefficient estimates and the y-intercept estimate for the CO₂ Emission Reduction Cost multilevel model. At the transaction level, if the new vehicle is a HEV, then CO₂ emissions decrease by 1,249.30 thousand pounds or by 6.90%¹⁰. A one standard deviation increase in Invoice Amount (\$449.81) decreases CO₂ emissions by 805.16 thousand pounds. Transactions in the Category 2 Truck-Passenger Car category decrease CO₂ emissions by 3,386.49 thousand more pounds and cost \$2.60 less per thousand pounds than transactions in the Category 2 Truck-Category 2 Truck category. Transactions in the Category 1 Truck-Passenger Car category decrease CO₂ emissions by 3,091.98 thousand more pounds and cost \$2.59 less per thousand pounds than transactions in the Category 2 Truck-Category 2 Truck category. Transactions in the Passenger Car-Passenger Car category decrease CO₂ emissions by 1,153.44 thousand more pounds and cost \$2.08 less per thousand pounds than transactions in the Category 2 Truck-Category 2 Truck category. At the state-level, a one standard deviation increase in CO₂ Emissions (88.29 billion pounds) decreases CO₂ emissions by 17.66 thousand pounds. CO₂ emissions decrease by 5,105.22 thousand pounds in states if the location quotient for passenger cars increases by one unit.

Table 3 Descriptive statistics for the transaction-level variables and for the state-level variables.

Table 4 Results from multilevel models for CO₂ emission reduction and for CO₂ emission reduction cost.

Choropleth maps of residuals from the multilevel model with CO₂ Emission Reduction as the dependent variable (Figure 2) and from the multilevel model with CO₂ Emission Reduction Cost as the dependent variable (Figure 3) test if spatial autocorrelation is evident in the respective model specifications.¹¹ The maps display standard deviation classification schemes where observed thousands of pounds (Figure 2) or observed dollars per thousand pounds (Figure 3) are lower than expected (blue) or higher than expected (red). Results suggest that clusters of residuals are evident. Moran’s I for the residuals from the thousands-of-pounds model is +0.18 (Z = +2.28, p = 0.02), while Moran’s I for the residuals from the dollars-per-thousand-pounds model is +0.24 (Z = +3.00; p = 0.003).

Figure 2 Choropleth map of residuals from multilevel model with CO₂ emission reduction as the dependent variable.

Figure 3 Choropleth map of residuals from multilevel model with CO₂ emission reduction cost as the dependent variable.

Local Indicators of Spatial Association (LISA) maps [35] of residuals from the multilevel model with CO₂ Emission Reduction as the dependent variable (Figure 4) and from the multilevel model with CO₂ Emission Reduction Cost as the dependent variable (Figure 5) test if local spatial association is evident in the respective model specifications.¹² The maps display High-High Cluster states and High-Low Outlier states (red) as well as Low-Low Cluster states and Low-High Outlier states (blue). Results suggest that clusters with similar values (High-High or Low-Low) and outliers with dissimilar High-Low values are evident. However, outliers with dissimilar Low-High values are not evident. In the thousands-of-pounds LISA map, High-High Cluster states are evident in the New England region (New Hampshire, Rhode Island, and Vermont) and in the South region (Arkansas and Louisiana). In the dollars-per-thousand-pounds LISA map, Low-Low Clusters are evident in the Midwest region (Illinois, Iowa, Michigan, and Wisconsin).

Figure 4 LISA (Local Indicators of Spatial Association) map of residuals from multilevel model with CO₂ emission reduction as the dependent variable.

Figure 5 LISA (Local Indicators of Spatial Association) map of residuals from multilevel model with CO₂ emission reduction cost as the dependent variable.

The Intraclass Correlation (ICC) apportions variation in the dependent variable of a random-intercept, multilevel model to the respective levels of analysis [23]. In the study, the ICC apportions variation in the dependent variables in the random-intercept, two-level models to the micro-level of analysis (transactions) versus the macro-level of analysis (states). Put succinctly, the ICC apportions variation in CO₂ emission reduction or variation in CO₂ emission reduction cost to the transaction-level independent variables versus the state-level independent variables in the respective random-intercept models. In two-level models, such as in the study, the ICC (ρ) is

\[ \rho=\frac{\tau_{00}}{\tau_{00}+\sigma^{2}} \tag{3} \]

where: $\tau$₀₀ is the macro-level variance; and σ² is the micro-level variance. The ICC for the random-intercept model for CO₂ emission reduction suggests that essentially all of the variation in CO₂ emission reduction is attributable to the transaction-level of analysis (99.82%). Likewise, the ICC for the random-intercept model for CO₂ emission reduction cost suggests that essentially all of the variation in CO₂ emission reduction cost is attributable to the transaction-level of analysis (99.67%).

5. Discussion

Analysis of a subsample of transactions for new AFVs from a past, national retirement program to increase private vehicle sales tests the viability of a future, national retirement program to decrease GHG emissions from the private vehicle fleet of the United States. Results suggest that incentives for HEV transactions decrease CO₂ emissions by a respectable 6.90%. Higher vouchers decrease CO₂ emissions more. Price point in the form of Base MSRP is not a barrier to emission reduction. Incentives for transactions when new vehicles are Passenger Cars reduce emissions most; particularly when trade-in vehicles are Category 2 Trucks. Results support the empirical literature on the marginal benefits of retirement [10] given that incentives for new AFV transactions decrease GHG emissions more in states where total GHG emissions from the transportation sector are higher. Interestingly, results contradict the empirical literature on the effects of gasoline price on the effectiveness of incentives for new AFV transactions to decrease GHG emissions [36].

Exploratory spatial data analysis of the residuals from the thousands-of-pounds multilevel model and the dollars-per-thousand-pound multilevel model with choropleth maps and with LISA maps as well as calculation of the ICC for the respective models suggest the following. First, the residuals from the respective multilevel models cluster from state to state in three regions. GHG emission reduction is higher than expected in the New England region and in the South Region, while GHG emission reduction cost is lower than expected in the Midwest region. Second, the overwhelming majority of the variation is attributable to variation between transactions within states, not variation between states. Therefore, state-level, nonmonetary incentives such as emission test exemptions, High Occupancy Toll (HOT) lane exemptions, High Occupancy Vehicle (HOV) lane exemptions, vehicle inspection exemptions, and parking fee exemptions may not account for much, if any, of the variation in the effect of monetary incentives on new AFV transactions between states in a future, national retirement program.

Critical analysis of the results from a retrospective study engender questions on generalizability and on relevancy. First, how generalizable are the results from an analysis of a past vehicle retirement program to a future national program? Indeed, the size of the national subsample of new AFV transactions is important to generalize the results to a future program. Unfortunately, the subsample of fossil fuel-alternative fuel transactions represents only ((34,305/677,842) × 100% = ) 5.06% of total (purchase plus lease) transactions. The subsample, nonetheless, includes a sufficient number of transactions in each state/state equivalent (except in the District of Columbia) to represent the total population of macro-level units of analysis. Second, how relevant are the results from an analysis of past transactions for new AFVs when new AFVs are presently so different? Indeed, Figure 6 shows the dramatic increase in the number of AFV models by technology/fuel from 2009 to 2020 [37]. For example, the number of HEV models in 2009 is 19, while the number of HEV models in 2020 is 81. The dramatic increase in the number of HEV models and in the number of EV models from 2009 to 2020 is evidence that new AFVs are less dependent on fuel to power an internal combustion engine and are more dependent on technology to power a battery. The decrease in the number of E85 models is attributable to the elimination of the Corporate Average Fuel Economy (CAFE) credit for FFVs, however the number of E85 models in 2020 is evidence that FFVs are presently popular with consumers.

Figure 6 Total Alternative Fuel Vehicle (AFV) models by fuel/technology from 2009 to 2020. E85 = Ethanol (85%) + Gasoline (15%); HEV = Hybrid Electric Vehicle; EV = Electric Vehicle; and CNG = Compressed Natural Gas.

6. Conclusions

Retrospective analysis of a subsample of fossil fuel-alternative fuel transactions from a past vehicle retirement program provides policy stakeholders with valuable empirical evidence on the viability of a future national program to incentivize new AFV transactions. Overall, the empirical evidence from the inferential models in the study suggests that incentivizing new AFV transactions benefits the environment by effectively reducing GHG emissions nationally even if such transactions are costlier economically. Likewise, incentivizing new AFV transactions reduces GHG emissions more in states where GHG emissions are higher and in states where the incidence of passenger cars is greater than in the United States.

The empirical evidence from the present study suggests two important tracks for ongoing research on new AFV transaction incentives. The first track for ongoing research emanates from the results on the environmental benefits of new AFV transactions; particularly new HEV transactions. The respectable 6.90% decrease in CO₂ emissions from such transactions and a recommendation on the basis of retrospective analyses of vehicle retirement programs worldwide to limit incentives to only transactions for new battery EVs (BEVs) [38] is justification for ongoing research to extrapolate the results from 2009 to the present. The past-present mismatch in the fuel/technology of AFV models from 2009 to 2020 highlights the importance of ongoing research to supplement the results of the present study with an analysis of a state (California) vehicle rebate program known as the Clean Vehicle Rebate Program (CVRP) on the effectiveness of incentives for new EVs and for new HEVs from 2010 onwards [22]. The second track for ongoing research emanates from the results on the dramatic differences in the economic costs of new AFV transaction incentives given the regional differences in the vouchers for such transactions. Static voucher amounts (\$3,500.00 or \$4,500.00) may ease implementation of a national program, but dynamic voucher amounts which vary by vehicle category and by the fuel economy difference from the new vehicle to the trade-in vehicle may help to harmonize GHG emission reduction and GHG emission reduction cost from region to region. Indeed, regional differences in GHG emission reduction and regional differences in GHG emission reduction cost justify ongoing research to infer the cause of such regional differences. A defensible approach, akin to a difference-in-differences (DID) approach [39], is to match fossil fuel-fossil fuel transactions with comparable fossil fuel-alternative fuel transactions. In such an approach, the fossil fuel-fossil fuel transaction is the control and the fossil fuel-alternative fuel transaction is the experiment so as to infer the cause of the difference in the CO₂ emission reduction between the two groups as well as to infer the cause of the difference in the cost of the CO₂ emission reduction between the two groups.

Author Contributions

The author did all of the research work for this study.

Competing Interests

The author has declared that no competing interests exist.