Abstract
Despite increasing public investment in charging infrastructure for plug-in electric vehicles (PEVs), policymakers know little about drivers’ preferences for publicly-accessible charging stations. Using data from an innovative choice experiment, we estimate demand for PEV charging stations, characterizing willingness to pay for access to types of locations as well as driver tradeoffs between refueling duration and costs. Prospective PEV drivers are willing to pay the actual variable cost of recharging at public charging stations and are willing to pay to cover significant fixed costs at select locations. Not surprisingly, many prospective drivers reveal a positive willingness to accept to wait while refueling, but this varies greatly across latent classes.
Similar content being viewed by others
Notes
PEVs include both battery electric vehicles (BEVs) and plug-in hybrid vehicles (PHEVs). BEVs are powered exclusively by electricity from the on-board battery, while PHEVs are powered by both electricity and gasoline, having both a battery and gasoline engine on board.
There are three types of charging infrastructure that support plug-in electric vehicles, each of which offers progressively faster charging times. Level 1 charging involves 110V charging from standard building electrical outlets. Level 2 offers 220V and 240V charging from dedicated chargers. Fast charging offers much higher voltage and is often comparable to gasoline refueling in terms of wait times.
“U.S. Department of Transportation Names Six Interstate Routes as ‘Corridors of the Future’ to Help Fight Traffic Congestion.” Federal Highway Administration, U.S. Department of Transportation. September 10, 2007. http://www.fhwa.dot.gov/pressroom/dot0795.cfm.
“Corridor: Interstate 5 (I-5)- Washington to California.” Corridors of the Future Fact Sheet. Federal Highway Administration, U.S. Department of Transportation. http://www.fhwa.dot.gov/pressroom/fsi5.cfm.
The CPUC approved SCE to install 1500 at the cost of up to $22 million, SDG&E to install 3,500 at the costs of up to $45 million and PG&E to install 7,500 at a cost of up to $160 million. See https://www.csis.org/analysis/utility-involvement-electric-vehicle-charging-infrastructure-california-vanguard.
See Fig. 5 in the “California statewide plug-in electric vehicle infrastructure” by NREL, 2014. This program was authorized by Assembly Bill 118 (Nùñez, Chapter 750, Statutes of 2007) instructing California Energy Commission to develop and deploy alternative and renewable fuels and advanced transportation technologies to help attain the state’s climate change policies.
For more details on this module, see Sheldon et al. (2017).
The charging choice experiments were only administered to respondents who chose some sort of PEV (approximately 85% of the respondents) at least once in the previous choice experiment. Respondents who never chose a PEV were not administered the charging choice experiments, as they were unlikely to view these choices as relevant or realistic.
For more details see Scarpa and Rose (2008).
The weighted California Household Travel Survey, relative to our weighted sample, exhibits modestly fewer upper middle households ($75–100k; 15% compared to 23%) and greater upper income households (>$150K; 21% compared to 12%). With respect to age, it exhibits a lower number of 18–24 year olds (2% compared to 16%), modestly greater 55–64 years olds (28% compared to 14%) and greater 65+ year olds (19% compared to 10%). With respect to education, it contains fewer households with less than a high school diploma (3% compared to 7%), fewer with a high school degree (11% compared to 25%) and greater with graduated degrees (26% compared to 13%). Finally, with respect to home ownership, it has modestly greater households that own their homes (77% compared to 62%).
For example, the multinomial logit estimated in Table 2 assumes that if an alternative such as grocery store charging were not available, the probability that had been assigned to choosing the grocery store would be equally split amongst the remaining alternatives. In contrast, the mixed logit estimated in Table 2 would allocate the probability according to the parameter distributions and the covariance between preferences for remaining alternatives and for grocery store charging.
While it is possible to make the mean or variance of a mixed logit parameter a function of observed covariates, we ran into frequently incurred problem that such models with more than one or two covariates tend to be numerically unstable and would not converge to a well-defined maximum value. Individual taste parameters can be identified as detailed in Revelt and Train (1999), although the authors caution that estimating a mixed logit as a function of observed covariates such as demographics is ‘more direct and more accessible to hypothesis testing than estimating a mixed logit without these characteristics, calculating expected tastes, and then doing cluster and other analyses on the expected tastes.’
Multinomial choice questions involving private goods are not necessarily incentive compatible even when the survey itself is consequential. This is because a respondent must balance an increase in the likelihood that the good is made available for purchase against an increase in the price charged if offered. As such, estimates can potentially be biased in either direction. Because our estimates are lower than fuel cost per mile for almost all gasoline vehicles and in the same general range as home charging costs, if a bias exist it is likely to be in a downward direction.
Our focus is on demand by the early and mid-market PEV adopters who are in our sample. Respondents who never chose electric vehicles were not administered the charging modules. Marginal willingness to pay for public charging is likely higher for respondents with a higher probability of owning a PEV. Since respondents who never chose electric vehicles were not administered the charging modules, the aggregate demand for charging at non-home stations in the longer-term future is likely underestimated here, to the extent households represented by these respondents later buy electric vehicles as their attributes including price and range improve and gasoline prices increase as California’s actions to reduce carbon dioxide and other air pollutants become increasingly stringent.
In California, PEV drivers are currently able to access HOV lanes free of charge. Research indicates this policy has had a significant positive impact on PEV sales in the state (Sheldon and DeShazo 2017). Our results indicate that development of public charging infrastructure could be equally if not more successful in promoting PEV adoption in California.
Because having to recharge one’s vehicle battery is an unfamiliar act for the respondents, the survey includes over 8 screens describing how recharging works as well as what the private costs and benefits of doing so are. We pre-check respondent knowledge of, and ability to distinguish, PHEVs from BEVs. We do not present this choice scenario to the 38 respondents who fail this pre-check. In this choice experiment, we place the respondent in PHEVs not only because it enables us to better a critical policy issue (increasing share of PHEV miles driven using electricity) but also because PHEVs more closely resemble ICEs than would a BEV. Importantly, PHEVs are not range limited while offering the driver flexibility to decide whether to recharge or not. Our pretesting revealed that by this point in the survey, respondents could conceptualize the time-cost trade-off that choosing the recharge represented for them.
At the time the survey was conducted, public charging stations were not common, which is why refueling at the gas station was likely to be perceived by respondents as a status quo.
The CAIC, introduced by Bozdogan (1987), is an extension to the traditional AIC that make it “asymptotically consistent and penalize overparameterization more stringently to pick only the simplest of the ‘true’ models.”
BIC is slightly smaller with fewer covariates, but this requires removing from the control set variables with highly significant coefficients.
There has been little work on the strategic incentives faced by a respondent asked a multinomial forced choice question without a no purchase alternative. Because making an alternative available with a price higher than the respondent is willing to pay can provide no additional utility, any bias in the willingness to pay estimates would appear to be downward. Similar to all multinomial choice questions, more complex patterns of strategic behavior across different types of alternatives cannot be ruled out as these depend on perceptions of how many alternatives will be supplied and the beliefs of other people, with convergence to truthful preference revelation occurring when only one of the alternatives will not be supplied and/or uninformative priors about the beliefs of other people are held by the respondent.
References
Achtnicht M, Georg B, Hermeling C (2012) The impact of fuel availability on demand for alternative-fuel vehicles. Transp Res Part D: Transp Environ 17(3):262–269
Bozdogan H (1987) ICOMP: a new model-selection criterion. In: 1. Conference of the international federation of classification societies, pp 599–608
Brownstone D, Bunch DS, Train K (2000) Joint mixed logit models of stated and revealed preferences for alternative-fuel vehicles. Transp Res Part B: Methodol 34(5):315–338
Bunch DS, Bradley M, Golob TF, Kitamura R, Occhiuzzo GP (1993) Demand for clean-fuel vehicles in California: a discrete-choice stated preference pilot project. Transp Res Part A: Policy Pract 27(3):237–253
California Department of Transportation (Caltrans) (2013) California household travel survey final survey report
Chen TD, Kockelman KM, Kahn M (2013) The electric vehicle charging station location problem: a parking-based assessment method for Seattle. Transp Res Rec 1254(28–36):2013
DeShazo JR, Song CC, Sinn M, Gariffo T (2015) State of the states’ plug-in electric vehicle policies, UCLA Luskin Center Report
Eisel M, Schmidt J, Kolbe LM (2014) Finding suitable locations for charging stations: implementation of customers’ preferences in an allocation problem. In: IEEE international electric vehicle conference (IEVC), Florence, pp 1–8
Ewing G, Emine S (2000) Assessing consumer preferences for clean-fuel vehicles: a discrete choice experiment. J Public Policy Mark 19(1):106–118
Franke T, Neumann I, Bühler F, Cocron P, Krems JF (2012) Experiencing range in an electric vehicle: understanding psychological barriers. Appl Psychol: Int Rev 61:368–391
Golob Thomas F, Bunch David S, David B (1997) A vehicle use forecasting model based on revealed and stated vehicle type choice and utilisation data. J Transp Econ Policy 1997:69–92
Hidrue MK, George RP, Willett K, Meryl PG (2011) Willingness to pay for electric vehicles and their attributes. Res Energy Econ 33(3):686–705
Ito N, Kenji T, Shunsuke M (2013) Willingness-to-pay for infrastructure investments for alternative fuel vehicles. Transp Res Part D: Transp Environ 18:1–8
Jia L, Hu Z, Liang W, Tang W, Song Y (2014) A novel approach for urban electric vehicle charging facility planning considering combination of slow and fast charging. In: International conference on power system technology (POWERCON), Chengdu, pp 3354–3360
Kitamura M, Hagiwara Y (2010) Honda ‘lacks confidence’ in electric-car demand. Bloomberg.com. http://www.bloomberg.com/apps/news?pid=newsarchive&sid=a_kxOOLkD.cU. Accessed 5 May 2017
Kurani KS, Thomas T, Daniel S (1996) Testing electric vehicle demand in ‘hybrid households’ using a reflexive survey. Transp ResPart D 1(2):131–150
Li S, Lang T, Jianwei X, Yiyi Z (2017) The market for electric vehicles: indirect network effects and policy design. J Assoc Environ Res Econ 4(1):89–133
McFadden D (1974) Conditional logit analysis of qualitative choice behaviour. In: Zarembka P (ed) Frontiers in econometrics. Academic Press, New York, pp 105–142
Nylund KL, Tihomir A, Muthén BO (2007) Deciding on the number of classes in latent class analysis and growth mixture modeling: a Monte Carlo simulation study. Struct Equ Model 14(4):535–569
Potoglou D, Kanaroglou PS (2007) Household demand and willingness to pay for clean vehicles. Transp Res Part D: Transp Environ 12(4):264–274
Qian L, Didier S (2011) Heterogeneous consumer preferences for alternative fuel cars in China. Transp Res Part D: Transp Environ 16(8):607–613
Revelt D, Train K (1999) Customer-specific taste parameters and mixed logit. In: Working Paper. University of California, Berkeley
Scasny M, Zverinova I, Czajkowski M (2015) Individual preference for the alternative fuel vehicles and their attributes in Poland. No 8575, EcoMod2015, EcoMod. https://EconPapers.repec.org/RePEc:ekd:008007:8575. Accessed 5 May 2017
Scarpa R, Rose JM (2008) Design efficiency for non-market valuation with choice modelling: how to measure it, what to report and why. Aust J Agric Res Econ 52(3):253–282
Sheldon TL, DeShazo JR (2017) How does the presence of HOV lanes affect plug-in electric vehicle adoption in California? A generalized propensity score approach. J Environ Econ Manag 85:146–170
Sheldon TL, DeShazo JR, Carson RT (2017) Electric and plug-in hybrid vehicle demand: lessons for an emerging market. Econ Inq 55(2):695–713
Sweda T, Klabjan D (2011) An agent-based decision support system for electric vehicle charging infrastructure deployment. In: 7th IEEE vehicle power and propulsion conference, Illinois
Train KE (1998) Recreation demand models with taste differences over people. Land Econ 74(2):230–239
Williams B, DeShazo JR (2014) Pricing workplace charging: financial viability and fueling costs. In: Transportation Research Board 93rd annual meeting, no. 14-1137
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
About this article
Cite this article
Sheldon, T.L., DeShazo, J.R. & Carson, R.T. Demand for Green Refueling Infrastructure. Environ Resource Econ 74, 131–157 (2019). https://doi.org/10.1007/s10640-018-00312-9
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10640-018-00312-9