including range and limited fuel availability for alternatives such as hydrogen fuel cell vehicles.2 A sensitivity analysis illustrating uncertainties about the market’s response to alternative vehicles and fuels is described in Section 5.7. There is also controversy about the magnitude of the social cost of GHG emissions and the social cost of the United States’ reliance on oil and petroleum-based gasoline. The estimates used in this report are drawn from the most recent literature but do not reflect the full range of uncertainty. Finally, it is extremely difficult to model all of the feedback effects that will inevitably result over time as technology development and markets interact.
Despite the inherent uncertainties in attempting to forecast four decades into the future, the committee’s modeling effort here uses the best available evidence and information and makes plausible assumptions where sound data are missing. Analysis of the results from the two models then provides useful insights about what various vehicles and fuel combinations can achieve, the nature of the processes by which changes will occur, and the general magnitude of potential costs and benefits of different policy options.
VISION is designed to extend the transportation sector-specific component of the National Energy Modeling System (NEMS) used by the Energy Information Administration (EIA). It provides longer-term forecasts of energy use and GHG emissions than does NEMS. While not as detailed or comprehensive as the NEMS model, VISION provides greater flexibility to analyze a series of projected usage scenarios over a much longer timeframe. It has been used extensively in the literature.
For the purposes of this study, VISION has been modified in a number of ways. The most up-to-date assumptions from the committee about vehicle efficiencies, fuel availability, and the GHG emissions impacts of using those fuels have been included. It is assumed that new-technology vehicle sales ramp up slowly and that new sales for a particular vehicle type never increase by more than about 5 percent of total new LDV sales in a given year. In addition, only one plug-in hybrid electric vehicle (PHEV), a PHEV-30 with a real-world all-electric driving range of 25 miles, is included. It is assumed that because of their limited range, battery electric vehicles are to be driven 1/3 fewer miles per year than other vehicles (Vyas et al., 2009) and that any decrease in miles driven by electric vehicles will be offset by increased mileage from other vehicles. Total new car sales and annual vehicle miles traveled (VMT) are assumed to be the same as in the projections from the Annual Energy Outlook 2011 (AEO; EIA, 2011a), and there is no assumption of a “rebound effect”3 if the cost of driving a mile declines. Adjustments to VMT can be included separately in any VISION run assessment.4 Finally, GHG estimates from biofuels include both emissions from production and from indirect land-use changes (see Chapter 3).
The committee uses the VISION model to explore how a focus on specific technologies or alternative vehicle and fuel types has the potential to reduce oil use and GHG emissions to achieve the study goals. The committee then turns to the LAVE-Trans model to shed light on how policies might be used to achieve the needed transitions.
The Light-duty Alternative Vehicle Energy Transitions (LAVE-Trans) model uses a nested, multinomial logit model5 of consumer demand to predict changes in the efficiency of vehicles and fuels over time, including a possible transition to alternatively fueled vehicles. Any transition to these advanced vehicles faces a number of barriers, including high costs due to the lack of scale economies and lack of learning, consumer uncertainty about safety or performance, and the lack of an energy supply infrastructure. Each of these barriers has been incorporated into the LAVE-Trans model so that the costs of overcoming them and, alternatively, the benefits of policies needed to do so can be measured (subject to the limits of current knowledge).
The model incorporates an array of factors that affect and are derived from consumer behavior, including the rebound effect; “range anxiety” and perceived loss of utility, particularly as it pertains to the availability of a fueling infrastructure; aversion to new technology and its reciprocal effect, early adoption; and the significant discounting of future fuel benefits over the lifetime of the vehicle. Nine variables influence the market shares of the alternative advanced technologies:
2Thanks to recent research, such issues are better understood than they were a decade ago (e.g., UCD, 2011; Bastani et al., 2012), yet much remains to be learned.
3Improvements in the efficiency of energy consumption will result in an effective reduction in the price of energy services, leading to an increase of consumption that partially offsets the impact of the efficiency gain in fuel use. This is known as the “rebound effect.”
4If a 5 percent reduction in vehicle miles traveled is plausible under certain policies, then the estimates of GHG emissions and oil use can be reduced by 5 percent.
5A multinomial logit model is a standard model often used to represent consumer choice where there is a finite set of discrete options. The probability of choosing among the set of available options is governed by representative parameters for a particular class of consumer. A nested model refers to multiple layers of choice (see Daly and Zachary, 1979; McFadden, 1978; Williams, 1977). For example, the first level of choice in the LAVE-Trans model is between choosing whether or not to buy an LDV. If a consumer chooses to buy an LDV, the next level of choice is between purchasing a passenger car or a light truck. Then, within a particular class of vehicle there are multiple options, such as whether to purchase an ICEV, FCEV, or BEV. Further description of the LAVE-Trans nested multinominal logit model can be found in Section H.2 in Appendix H.