Scholarly article on topic 'An exploratory policy analysis of electric vehicle sales competition and sensitivity to infrastructure in Europe'

An exploratory policy analysis of electric vehicle sales competition and sensitivity to infrastructure in Europe Academic research paper on "Social and economic geography"

Share paper
{E-mobility / "Automotive technology transition" / "System dynamics modelling" / "Charging infrastructure" / "EU policy"}

Abstract of research paper on Social and economic geography, author of scientific article — Gillian Harrison, Christian Thiel

Abstract This research contributes to discussions about policy interventions to stimulate the transition of vehicle technology. Concentrating on passenger cars, an extensive system dynamics based market agent model of powertrain technology transitions within the EU up to 2050 is employed. With a focus on subsidy scenarios for both infrastructure deployment and vehicle purchase, and set within the context of the EU fleet emission regulations, we find that there are important interactions between different powertrain types and with infrastructure provision. For example, strong plug-in electric vehicle (PiEV) policy could inhibit the maturity of hydrogen fuel cell vehicles. Infrastructure provision is important for improving the utility of a PiEV, but we find that in the early market it may have a weaker correlation with uptake than other policy options, until the PiEV stock share is over around 5%. Furthermore, an attempt to install a ratio of much more than one charge point per 10 PiEV may lead to little gains and high costs. PiEV sales are relatively insensitive at target levels over 25 PiEV per charge point. The results of our study can help policymakers to find the right balance and timing of measures targeting the transition towards low carbon alternative vehicles.

Academic research paper on topic "An exploratory policy analysis of electric vehicle sales competition and sensitivity to infrastructure in Europe"


TFS-18630; No of Pages 14

Technological Forecasting & Social Change xxx (2016) xxx-xxx

Contents lists available at ScienceDirect

Technological Forecasting & Social Change

An exploratory policy analysis of electric vehicle sales competition and sensitivity to infrastructure in Europe

Gillian Harrison a'*, Christian Thielb

a European Commission, Joint Research Centre, Directorate for Energy, Transport & Climate, Netherlands b European Commission, Joint Research Centre, Directorate for Energy, Transport & Climate, Italy



Article history:

Received 15 March 2016

Received in revised form 3 August 2016

Accepted 8 August 2016

Available online xxxx

Keywords: E-mobility

Automotive technology transition System dynamics modelling Charging infrastructure EU policy

This research contributes to discussions about policy interventions to stimulate the transition of vehicle technology. Concentrating on passenger cars, an extensive system dynamics based market agent model of powertrain technology transitions within the EU up to 2050 is employed. With a focus on subsidy scenarios for both infrastructure deployment and vehicle purchase, and set within the context of the EU fleet emission regulations, we find that there are important interactions between different powertrain types and with infrastructure provision. For example, strong plug-in electric vehicle (PiEV) policy could inhibit the maturity of hydrogen fuel cell vehicles. Infrastructure provision is important for improving the utility of a PiEV, but we find that in the early market it may have a weaker correlation with uptake than other policy options, until the PiEV stock share is over around 5%. Furthermore, an attempt to install a ratio of much more than one charge point per 10 PiEV may lead to little gains and high costs. PiEV sales are relatively insensitive at target levels over 25 PiEV per charge point. The results of our study can help policymakers to find the right balance and timing of measures targeting the transition towards low carbon alternative vehicles.

© 2016 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license


1. Introduction

Electro-mobility is seen by many to be at the core of future mobility patterns. Electric Vehicles (EV - comprising Plug-in Hybrid Electric Vehicle (PHEV), Battery Electric Vehicles (BEV) and Fuel Cell Vehicles (FCV)), based on an electric motor powertrain, offer a potentially substantial contribution to overcoming environmental problems created by the widespread dependence on conventional automobiles. Conventional Internal Combustion Engine vehicles (ICEV), mainly fuelled with petrol or diesel, have dominated mobility for the past century. As the transport sector currently makes the third greatest contribution to global carbon emissions (IEA, 2016) and accounts for half of daily oil consumption (IEA, 2015), a paradigm shift in mobility is required. Within the context of climate change, this transition needs to occur within the next 30 years to avoid serious irreversible shifts in our climate (IPCC, 2013). In Europe, where transport is the second largest source of greenhouse gas (GHG) emissions, accounting for around a

Abbreviations: BEV, Battery Electric Vehicle; PHEV, Plug-in Hybrid Electric Vehicle; PiEV, Plug-in Electric Vehicle - BEV and PHEV; FCV, Fuel Cell Vehicle; EV, Electric Vehicle - PiEV and FCV; HEV, Hybrid Electric Vehicle; ICEV, Internal Combustion Engine Vehicle; pCP, public (slow) Charging Point; rCP, rapid Charging Point; PTTMAM, Powertrain Technology Transition Market Agent Model; DAFI, Directive on the Deployment of Alternative Fuel Infrastructure.

* Corresponding author at: P.O. Box 2, NL-1755 ZG Petten, Netherlands.

E-mail address: (G. Harrison).

quarter (EC, 2016), the European Union (EU) is committed to reduce GHG emissions economy wide by 40% (versus 1990 levels) by 2030 with road transport playing an important role towards achieving these targets (EC, 2014). Furthermore, the 2011 EC White Paper on Transport set a target of reducing road transport emissions by 60% of 1990 levels by 2050, and within this to "halve the use of 'conventionally fuelled' cars in urban transport by 2030 and phase them out in cities by 2050" (EU, 2011a). Here, "conventionally fuelled" is defined as those vehicles powered by Internal Combustion Engine (ICE) only. Consequently, EVs as a zero tail-pipe emitting transportation option have arisen as a critical enabler for a low carbon economy (EU, 2014a) as well as for improved air quality.

Various studies have considered future EV market penetration, with both short and long term estimates varying greatly (Pasaoglu et al., 2012), and the IEA suggesting only a 9% global light duty vehicle stock share by 2030 and 40% by 2050 under their 2DS scenario (IEA, 2016). This seems far off EU targets, and as such regulation aimed at manufacturers to reduce fleet emissions has been introduced (EU, 2014b; EU, 2014c). Currently, only Plug-in Electric Vehicles (PiEVs - PHEV and BEV), are widely available. However, as yet, and despite rapidly growing sales (ACEA, 2016) they have failed to capture a significant passenger car market share and continue to be dependent on support measures, such as financial incentives (Mock & Yang, 2014; Thiel et al., 2015). One significant reason for this is limited consumer acceptance, due to high upfront costs and the phenomenon of range anxiety (Thiel et al.,

0040-1625/© 2016 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license (


2 G. Harrison, C. Thiel / Technological Forecasting & Social Change xxx (2016) xxx-xxx

2012). In recognition of the cost barrier, many countries are introducing fiscal incentives (ACEA, 2014). In addition, both costs and range anxiety are related to the current capabilities of battery technology. Battery costs are reducing rapidly (Nykvist & Nilsson, 2015), but to alleviate the latter concern policy makers in countries across Europe are encouraging the development of an appropriate charging infrastructure. There are various factors which may impact on the efficacy of infrastructure on uptake, including increasingly cheaper and more rapid chargers, battery capacity and private charging capabilities. This is often termed a "chicken and egg" problem, as infrastructure providers are reluctant to invest without a substantial EV market, yet drivers are wary of entering into e-mobility without the confidence of a reliant, widespread, and interoperable charging infrastructure.

Previous research related to EV policy has focused on the most widespread policies currently applied, which are fiscal incentives for users (ICCT, 2011a; Brand et al., 2013; Diamond, 2009; Gass et al., 2014; Hidrue et al., 2011; ICCT, 2011b; Lane & Potter, 2007; Tran et al.,

2013) and regulation of manufacturer emissions or vehicle efficiency (IEA, 2008; ITF, 2010; Walther et al., 2010; Thiel et al., 2014). This research has generally agreed that due to the high cost differential between EVs and their conventional counterparts, fiscal incentives are required to encourage early adopters to the technology leading to successful pre-mass market penetration. From a supply-side point of view, manufacturers must be also encouraged to invest further in R&D of low carbon technologies in order to bring increasingly affordable and efficient EV into the commercial market. For example, besides fleet emission regulatory targets (EU, 2014b; EU, 2014c; EU, 2009a; EU, 2011b) and member state co-funded R&D projects, the 'European Green Vehicles Initiative',1 has been an important public-private-partnership at EU level since 2008, funding numerous activities under the EU framework programmes for research and innovation (e.g. Framework Programme 7, Horizon 2020). Altogether there are >300 ongoing R, D & D projects with a total budget of nearly 3 billion euros across the EU co-funded by both the EU and member states. On the one hand these projects support technological improvements, most notably for energy storage and control devices, on the other hand, through field tests, they address customer acceptance and vehicle to grid integration (Zubaryeva & Thiel, 2013).

In general, it is not controversial to suggest that policies to introduce sufficient public charging infrastructure are necessary to encourage the introduction of EVs (Bakker &Jacob Trip, 2013; OLEV, 2011). This is to overcome issues of range anxiety, identified to be one of the most significant barriers to EV adoptions in many choice modelling studies (Batley et al., 2004; Beggs et al., 1981; Brownstone et al., 1996; Dagsvik et al., 2002; Eggers & Eggers, 2011; Ewing & Sarigollu, 2000; Potoglou & Kanaroglou, 2007). However, there has been little literature empirically exploring the relationship between minimum charge point provision and EV uptake, instead tending to focus (for example) on socioeconomic or spatial distribution (Namdeo et al., 2014; Zubaryeva et al., 2012; Maia et al., 2015) and charging profiles (Robinson et al., 2013; Donati et al., 2015). Although these may be determinants of EV uptake, current EU policy is focused on guaranteeing a minimum ratio of charge points to EVs in order to avoid market fragmentation and ensure coverage across national borders (EU, 2014a). Recharging infrastructure has been analysed through evidence-based studies, expert elicitation and multi-criteria assessment for determining the policy promoters of EVs. For instance, Zubaryeva et al. (2012) identified that an adequate recharging infrastructure was one of the most important parameters for the large scale deployment of PiEVs in Europe, and (Sierzchula et al., 2014) have found that countries could achieve high adoption rates by increasing their recharging infrastructure levels. Other studies suggest that collaborative schemes between private and public authorities combining incentives and infrastructure are required

for success (Mock & Yang, 2014; Thiel et al., 2012; Rowney & Straw, 2013; Norbech, 2013; Lane & RAC, 2011). Hence, recharging infrastructure can be considered one of the critical parameters in market penetration of EVs.

This paper takes the EC Clean Power for Transport package (EC, 2013a) as a starting point and seeks to explore what impact government policy on infrastructure can have on EV uptake. In particular, we focus on the recently adopted Directive on the Deployment of Alternative Fuels Infrastructure (DAFI) 2014/94/EU (EU, 2014a), and the proposals therein regarding minimum coverage of PiEV charging infrastructure by the end of 2020. We take the approach to identify what impact policy options may have on long term EV penetration. We analyse numerous policy scenarios, recognising that single e-mobility policies should not be considered in isolation as the interaction between multiple policies is highly relevant. For example, a suite of incentives and other demand stimulating policies were employed by Norway, the most successful European country in terms of EV uptake (Mock & Yang, 2014; Norbech, 2013). Our approach seeks to understand how specifically supporting the infrastructural system may characterise uptake within the wider policy environment. For the wider policy environment we consider supply (e.g. fleet emission regulation) and demand stimulating policies (e.g. purchase incentives). To do this, our research employs an extensive system dynamics model of the EU automobile market, which reflects the relevant market agents of users, manufacturers, infrastructure providers and authorities. This research is the application of the model, which is described in detail in a Technical Report (Harrison et al., 2016), and was presented in a previous paper by the authors (Pasaoglu et al., 2016). (Pasaoglu et al., 2016) was designed as an introduction to the model that could then be built upon in future publications such as this, as it focused on only five generic scenarios reflecting three market variables (learning rate, oil price and GDP) and two policy options (vehicle purchase subsidies and fleet emission targets). The purpose of this study is to focus on the provision of infrastructure, including within the context of the previous policy options, in a timely investigation regarding the implementation of (EU, 2014a).

2. Model overview

The Powertrain Technology Transition Market Agent Model (PTTMAM) was developed at the EC Joint Research Centre (JRC) in collaboration with Ventana Systems UK using Vensim™,2 a leading and highly flexible software for system dynamics model building and simulation. The purpose and focus of the model is to study the interaction between, and influence of, the market agents on possible technology transitions within Europe, for each of the 28 member states and across the period 1995 to 2050. The use of system dynamics to analyse possible future scenarios of technology transition in the automotive sector has been explored by many authors (Walther et al., 2010; Bosshardt et al., 2007; Gomez et al., 2013; Harrison & Shepherd, 2014; Janssen et al., 2006; Kohler et al., 2010; Meyer & Winebrake, 2009; Richardson et al., 1999; Rodrigues et al., 2012; Shepherd et al., 2012; Struben & Sterman, 2008; Shepherd, 2014; Leiby & Rubin, 1997; Stepp et al., 2009; Boksberger et al., 2012; Stasinopoulos et al., 2012; Diwaker et al., 2013). Recent overviews of such studies can be found in Harrison and Shepherd (2014) and Shepherd (2014). Many of these have a limited focus, for example on one particular powertrain or country. At the other extreme, Gomez et al. (2013) focus on a simplified high-level global view. Their purpose ranges between forecasting deployment, detailed policy analysis, manufacturer strategies and environmental or economic assessments. To our knowledge, the model presented here is the first attempt to address not only the most relevant interacting agents within the light duty vehicle market (i.e. automotive manufacturers and suppliers, infrastructure providers,

1 2


G. Harrison, C. Thiel / Technological Forecasting & Social Change xxx (2QÍ6) xxx-xxx

policymakers, and consumers) in a European context, but also the competition between all current and future alternative fuel and powertrain technologies, across all member states. A similarly broad model was developed by Walther et al. (2010) and adapted by Harrison and Shepherd (2014), but this focused on the Californian case and was not as detailed or complex as the model we have developed. Some other studies may have been more detailed in their approach towards specific areas, e.g. Kohler et al.'s (2010) consideration of the hydrogen fuelling network, Diwaker et al.'s (2013) focus on the R&D program or Stasinopoulos et al. (2012) who looked at materials used, but this was a level of detail not required in our model. Our approach allows us to capture the complex and sometimes conflicting interests that actually exist in the automotive sector. It also enables us to study the impact of interacting policies on technology transitions in this sector. Nevertheless, the reader should be mindful that a model is a simplified representation of reality and should therefore be viewed as a means of comparing "what if" scenarios, rather than providing predictive or precise results.

The model is detailed and complex, containing over 700 simultaneous equations and 300 constant values, within 13 causal feedback loops. When the base elements are combined with the subscripts employed within the model (such as powertrain type, member state or vehicle size), this leads to over 700,000 elements overall. Within the model, four conceptual market agent groups (manufacturers, users, authorities and infrastructure providers) are represented by key decision rules and feedbacks. In its current form the model does not capture the behaviour of individuals within the system as agent based modelling could, such as proposed or applied by numerous authors (Wolf et al., 2015; El Banhawy et al., 2014; Shafiei et al., 2012a; van der Vooren & Brouillat, 2013; Eppstein et al., 2011; Noori & Tatari, 2016). However, a certain disaggregation of the representative agent groups is partially implemented (for example the distinction of urban and non-urban drivers) and could be furthered, if substantiated by behavioural data. The model inputs have been sourced from numerous publications and expert opinions, but importantly including Eurostat3 and TRACCS4, and for future transport/energy demands the EU 2050 Energy Trends Reference Scenario (EC, 2013b). Some key parameters (such as vehicle demand) have been calibrated to historical data, and sensitivity tests and reality checks were carried out to ensure model robustness. Calibration (or optimisation) of values is a standard approach across many fields applied to the determination of input data not readily available (Valipour, 2015; Valipour et al., 2015; Li et al., 2014; Lim et al., 2014). A more comprehensive overview and introduction to this model is presented in Pasaoglu et al. (2016) and Harrison et al. (2016). Although we urge the reader to refer to these sources, some key elements and updates relevant to this research are presented below.

WtC builds up over time through social exposure to the alternative powertrain (from marketing and interaction with both users and nonusers), but can also decay as that social exposure is 'forgotten', until a tipping point is reached when the powertrain remains within the decision set.

The WtC is then combined with relative financial attractiveness (based on total cost of ownership), and perceived values and importance of seven attributes that characterise each powertrain (Eq. (1)). Together these determine the overall attractiveness as a combined utility for that powertrain, specific to a member state, powertrain, user group and vehicle size class. The powertrain attributes were chosen as to reflect available information on consumer choice on importance (Capgemini, 2008; Deloitte, 2011) but also simplified as development of future powertrains is uncertain. They are however in line with more detailed choice modelling studies such as those cited in the Introduction. The attribute values evolve over time in relation to the maturity of the individual powertrain components that are themselves improved by technology uptake and the R&D effort of the manufacturer. It would be beyond the scope of this paper to go into the development of all the criteria in detail, though it is relevant to note that convenience incorporates the role of infrastructure provision in purchase behaviour (see later). The financial attractiveness is an assessment of the total cost of ownership by the user, which includes the purchase price (as set by the manufacturer and adjusted by authorities with subsidies and taxes), and a proportion of the running costs.

Combined utilityPi = (Attribute valuePi x Attribute importance) x Willingness to considerpj x Financial attractivenessPi (1)

Eq. (1): Combined utility of powertrain Pi.

Attributes = choice, convenience, environment, performance, reliability, safety, popularity.

All values except attribute importance are calculated endogenously.

A standard multinomial logit choice framework, Eq.(2), similar to the approach employed by Struben and Sterman, is used for the assessment of the user's preferred powertrains (and thus the indicated market share) using the calculated combined utility.

Indicated market share =

eCombined utilityPi _1 e^ombined utilityPx_1

Eq. (2): Indicated market share of powertrain Pi. Combined utility calculated using Eq. (1).

2.2. Infrastructure provision

2.1. User purchase decisions

Fundamental to the model is the ability to capture behavioural dynamics of the users within the system. At the centre of this mechanism within the model is a modified version of the highly cited Willingness to Consider (WtC) concept as developed by Struben and Sterman (2008) and adopted in numerous models since (Walther et al., 2010; Shepherd et al., 2012; Kwon, 2012; Shafiei et al., 2012b). WtC "captures the cognitive, emotional and social processes through which drivers gain enough information about, understanding of and emotional attachment to a platform (powertrain) for it to enter their consideration set".



The powertrain attribute convenience, which contributes to the combined utility, represents the access to effective infrastructure for refuelling or recharging a vehicle, as well as the provision of maintenance facilities (which isn't discussed in this paper). Within the PTTMAM, the conceptual infrastructure provider group decides on the type and amount of fuelling and charging facilities to invest in, reacting to signals from the other market agents and based on a desired return on investment (ROI).

22.1. Charging infrastructure

The convenience attribute for PiEV is represented by the average population in each country with access to public (slow) or private (home/work) charging and the proportional achievement of a desired rapid charging network. Although the presence of rapid-charging stations improves the convenience for those with the requirement to recharge on the move, it does not improve access for vehicle owners with no private means of recharging. PHEV convenience is further influenced by a weighting of the relative availability of charging and


G. Harrison, C. Thiel / Technological Forecasting & Social Change xxx (2016) xxx-xxx

refuelling infrastructure. The electricity charging infrastructure feedbacks within the model focuses on the provision of public charge points (pCPs) and rapid charge points (rCPs) as private charging is considered to be deployed in combination with the vehicle sales. The installation of CPs relates to the expected revenue as shown in Eq. (3), though pCP and rCP are considered separately in the model. Both installation and running costs decrease as the number of installed CPs increases, representing a combination of learning effects and economies of scale. The required additional revenue (for desired ROI) is compared to an estimate ofadditional revenue from potential use, determined from an assessment of forecast visits (of forecast urban and non-urban PiEV users with and without private access) and a presumed revenue per visit and frequency of visits. Though early adopters of PiEV are likely to be those with private charging access, as penetration increases it is more likely that some PiEV users would be making sole use of public charging network. Both pCP and rCP are stocks which grow over time.

Desired CP installation

Forecast revenue—Current revenue _ Running costs + Installation cost x (1 + Desired ROI) (

Eq. (3): Desired installation of charging infrastructure in each country.

All values are calculated endogenously, except ROI which is taken to be 0.2.

Maximum and minimum costs are derived from (Gardien & Refa, 2015; Plotz etal., 2014).

22.2. Hydrogen fuelling infrastructure

Similar to CPs, infrastructure providers will add hydrogen fuel to an existing fuelling station if their forecast fuel revenue could deliver a desired ROI (here, a country specific calibrated value in the PTTMAM). The number of desired fuelling stations to invest in is therefore determined using a similar calculation as Eq. (3). Forecast fuel demand is based on a 3 year forecast stock of the relevant powertrain and annual fuel consumption (based on a reference value (Edwards et al., 2014) and modified by improvements to the powertrain as a result of R&D activity within the model). Revenue is based on a fuel cost (Tremove, 2010) and calibrated fuel margin. Investment costs reduce over time depending on the current installed base of hydrogen fuelling stations. In addition, fuelling stations may be dropped if they are no longer deemed sustainable.

2.3. Policy implementation

The PTTMAM was designed to enable to the modeller to create scenarios related not only to market conditions but also policy strategies. There are numerous inputs related to subsidisation and taxation of vehicles, fuel and infrastructure that can be defined by the modeller at either country or EU-wide levels. Moreover, the PTTMAM includes the most significant EU regulation related to the control of manufacturers regarding new fleet emissions. Here, the model structures relevant to the policy levers considered in this paper are briefly described.

2.3.1. Manufacturer emission regulations

Within the model, the current EU light duty fleet emission regulations (EU, 2014b; EU, 2014c; EU, 2011b; EU, 2009b) are captured. Average tailpipe CO2 emissions of new vehicles are assessed against a specific emission target per vehicle as defined in the regulations (Eq. (4)) and authorities charge manufacturers penalties for those fleets with averages above a specified threshold. This excess emission premium is designed to incentivise investment in low carbon vehicle technology, as Manufacturers base their R&D investment decisions on forecast future emission penalties from speculative demand.

Emissions target = Base target + a x Average mass-Mo (4)

Eq. (4): Emissions target (g/km) (as defined in the regulations).

Base target = PC 130 g/km; 95 g/km from 2021. LCV 175 g/km; 147 g/km

a = 0.457; 0.333 from 2020.

Average mass = based on (Thiel et al., 2014)

Mo = 1372 kg up to 2016,3 year average thereafter.

2.3.2. Vehicle purchase subsidies

At the EU level considered in this study, vehicle price subsidies are entered by the modeller as a proportion of the cost differential between the subsidised powertrain and the conventional ICEV powertrain. The effect here is an improvement in the financial attractiveness of those powertrains and thus a greater combined utility that leads to a higher market share. In addition, having the subsidy in place can lead to increased marketing of the powertrain by manufacturers and a demand kick over and above the standard utility as users are aware of the offer being in place, and possibly short-lived.

2.3.3. Infrastructure subsidies

The PTTMAM input for subsidies for infrastructure providers for the installation of infrastructure is entered as a proportion of the costs for each fuel and separate year of the simulation. This subsidy assists the infrastructure provider as it reduces the installation costs making the investment more attractive for achieving the desired ROI. In addition, for charging infrastructure, for which the subsidy only covers pCPs (not rCPs) there is a default assumption of an authority desire of 10 PiEV per pCP, as per the DAFI indicative suggestion to member states. For hydrogen, there is no such limit.

3. Scenarios

The policy scenarios of this research have been modified and widened, compared to earlier applications of the model (Pasaoglu et al., 2016). In this study a number of credible yet realistic levels for each policy have been designed based on this, and an extra subsidy for infrastructure is included to reflect the focus of the study. A table of sources for key variables for the baseline conditions, as used in the PTTMAM Technical Report (Harrison et al., 2016) is shown in the annex. In this research our interest lies in analysing the interaction of various policy options: CO2 regulations on the manufacturer, EV purchase subsidies for the user, and infrastructure subsidies for the

Table 1

Initial policy scenario levels (NB: infrastructure subsidies are all hypothetical).

Scenario level

R regulatory CO2 target (g/km)

P EV purchase subsidy (% of ICEV cost differential)

1100% infrastructure subsidy

0 - No Policy

1 - Current

2 - Projected

3 - Extended 4- High

No penalties in place

2021: 95 2021: 95; 2030: 70; 2050: 41 2021: 95; 2030: 56;2050:27

No subsidies in place

2011-13: 50%; 2014-15: 25%

2011-13: 50%; 2014-20: 25%

2011-13: 50%; 2014-25: 25%

2010-15: 75%;2016-20: 50%; 2021-25: 25%

No subsidies in place 2010-2020 2010-2030 2010-2050


G. Harrison, C. Thiel / Technological Forecasting & Social Change xxx (2016) xxx-xxx 5

infrastructure provider. For the CO2 regulation and infrastructure subsidy duration we consider four levels and for the EV purchase subsidies we consider the same level and an additional "High" level, as detailed in Table 1. All regulation scenario levels, and the purchase subsidy current and high scenario levels are based on current and proposed targets (Pasaoglu et al., 2016). The projected subsidy would seem in line with governmental intentions, for example, the UK government have recently announced an extension of their plug-in car grant until

at least 2018 (OLEV, 2015) and the German government contemplates purchase premiums for EV (FAZ, 2015). The extended EV subsidy scenario prolongs the subsidy to 2025. Infrastructure subsidies are designed to be optimistic and are hypothetical examples employed to understand the impact of increased infrastructure rather than representing specific policies. Our baseline scenario described above is scenario R1P1 (Current regulatory targets and purchase subsidies, no infrastructure subsidies).

Fig. 1. Alternative vehicle shares under initial policy scenarios. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)


G. Harrison, C. Thiel / Technological Forecasting & Social Change xxx (2016) xxx-xxx

4. Results of scenario runs

Fig. 1 presents the results from our model for the initial policy scenarios with no infrastructure subsidies in place. These results will be discussed in more detail in the following sections. However it is worth noting at this point some high level observations. Unsurprisingly, stronger policies and combining policies generally leads to more successful results. This aligns well with observations from previous studies (Pasaoglu et al., 2016; Harrison & Shepherd, 2014; Stepp et al., 2009; Brand et al., 2012; Foxon et al., 2005; ITF, 2008). The "no-regulation" scenarios (R0) experience limited EV success by 2050, where the EV share is in fact lower than in 2020. The scenarios with a CO2 regulation in place (R1) witness stronger EV deployment with stricter targets. The tightest regulatory targets (R2-3) are most beneficial for all EV. Longer term emission targets (R3) have a greater impact on FCV success than on PiEV deployment by 2050. Having purchase subsidies in place does make some impact, especially in 2020 and 2030 results, compared to their no-subsidy counterparts, though the durations tested results in little variation. However, had there been higher purchase subsidies in place from 2010 (P4), EV share would have been much more successful by 2020, though this impact has diminished by 2050 in all but the no regulation scenario.

4.1. Comparison to real world data

It was felt at this time that comparison of our model baseline performance to both historic and future changes would be insightful. At the time of writing, 5 years of data have been gathered on the sales of EV by the EEA for the monitoring of the fleet emission regulation directive. This data has been further analysed in previous work (Thiel et al., 2015), and sales shares based on this are presented and compared with our model results in the table below (Table 2). Bearing in mind that we are not claiming the model to be accurate in a forecasting capacity, the results for these first five years of data would appear to be more or less in line with the sales being witnessed in reality. Although in 2012 and 2013 the model would appear to slightly over-predict sales share, 2014 is similar. However, it should be also borne in mind that at sales of such low shares as at present it is not appropriate to suggest accuracy of the model, but as a reassurance that such promising similarities would lend credence to our later results.

Table 2:

Comparison of model baseline (R1P1) to actual data, sales market shares (in %).

2010 2011 2012 2013 2014

PHEV Actual 0 0.003 0.07 0.21 0.53

Model R1P1 0 0 0.12 0.45 0.62

BEV Actual 0.006 0.07 0.12 0.20 0.31

Model R1P1 0.006 0.10 0.22 0.35 0.30

FCV Actual 0 0 0.0002 0.0004 0.0003

Model R1P1 0 0 0 0 0

42. Comparison to existing studies

It is not the intention for this model to be used in a predictive capacity for future market shares, but to analyse the relative impact of policy options between different scenarios. However, for context, it is useful to see how our model behaves in relation to studies designed to describe EV outlook. For this, in Table 3 we compare our initial policy scenario results to the range reported in a previous study that included a metaanalysis of EV penetration rates from 10 studies (Pasaoglu et al., 2012). This was the most recent study we could identify with all of this data compiled, though a 2016 BNEF study with only one scenario suggested an approximate 5% of sales would be PiEV by 2020 and 40% by 2040 (however under their optimistic conditions BEV is costs competitive with ICEV by 2022 and therefore is the dominant PiEV) (BNEF, 2016). The studies included in this analysis comprised of some with different scenarios (low-high decarbonisation) and some with baseline/reference only. The table thus shows the ranges for low, med/base and high, against the ranges from scenario ranges R0Px, R1 Px and R2-3Px respectively. These groupings would feel to be appropriate representations of such low-high scenarios in line with the comparisons. It would appear that although our model may appear to be on the conservative side for all but HEV, the relative shares between alternative choices seem to be in line with these other studies. Interestingly this also demonstrates the wide range in predictions that does exist due to differing scenarios and assumptions employed between studies, and therefore the lack of certainty in EV sales projections, justifying our approach to concentrate on relative impacts, though some confidence in the range of absolute numbers is provided by the previous observations comparing to real world data.

4.3. Baseline for comparison (R1P1)

Under our baseline scenario (R1P1 - current regulation and purchase subsidies), there is a 5% EV market share of sales by 2020, and a 9% HEV share. By 2030 we see that the share of conventionally fuelled vehicles (ICEV) has been reduced by over a third from 100% in 1995 and the EV market share has grown to 20%, and FCV has not yet penetrated the market. ICEV is further reduced to around 40% by 2050, when PiEV have captured 42% of the market, though FCV has still failed.

4.4. Scenarios with no CO2 target (R0Px)

In comparison to our baseline (R1P1), without any policy in place (R0P0), conventional vehicles reduce to only 95% in 2020, 92% in 2030 and 81% in 2050. This technology transition arises solely from HEV capturing approximately 19% of the market, and the remaining EV powertrains are not successful. Although this first observation may seem an obvious one, it confirms that by introducing policy conditions within this model we do achieve technology transition as would be expected. With purchase subsidies but no fleet emission regulations (R0P1-3), EVs still never achieve a significant market share, with only PHEV gaining a share above 2% even by 2050. When the high level purchase subsidies are in place (R0S4), we see more clearly the diminishing

Comparison to other studies (other studies data taken from Pasaoglu et al., 2012).


Other studies Own Other studies Own Other studies Own Other studies Own

2020 Low 4-10 5 1-5 0-1 0-2 0

Med/base 9-17 9 3-5 0-4 1-3 0-2

High 13-18 9 2-15 0-4 3-12 0-2

2050 Low 8 19 18-25 0-2 10-14 0-1 4-5 0

Med/base 7-25 27-37 24-35 13-22 19-25 7-12 12-25 0-3

High 11 18-30 16-27 21-32 24-35 12-20 19-50 16-23


G. Harrison, C. Thiel / Technological Forecasting & Social Change xxx (2016) xxx-xxx

impact of the subsidies from 2020 to 2050, but also that the more generous subsidy amount has led to EV shares 3-4 times greater than the comparative scenarios.

4.5. Current regulation scenarios (RIPx)

Compared to the scenarios R0, which include no regulatory targets or penalties, under R1 we already see an improvement in alternative market shares by 2020, which is more than double of the scenarios without fleet regulation (from 5 to 6 to 9-15%). This impact is more obvious later in the simulation. By 2030, ICEV sales are a third lower than the no regulation scenarios, whereas alternative powertrains now account for almost 40% of all sales, compared to 10% in the previous scenarios. By 2050 the effect of having the regulation in place is a halving of ICEV sales share from the previous scenarios under all subsidy options. However, EV sales reach around a third of the market by 2050, though only the no purchase subsidy scenario (R1P0) has any FCV sales.

By introducing EV purchase subsidies (R1P1-4), in 2020 there is noticeable uptake of PHEV and BEV compared to R1P0 (no EV purchase subsidies), resulting in almost twice the PHEV and BEV shares by 2030, though there is marginal impact from the actual duration of the subsidy availability. However, this additional share would appear to come not from ICEV share but from lowering HEV and FCV shares. At first, this would seem to be counter-intuitive. However, on investigation this phenomenon hints upon the importance of finding the right timing for supply and demand stimulating policies and analysing their interaction. The following two mechanisms can explain this in more detail:

• ICEV purchase price is lowered: As purchase subsidies are introduced, and successful (as seen in the 2020 EV shares), the resultant fleet emissions are reduced thereby leading to lower predicted penalties for the manufacturer or less need for CO2 improvements for the conventional vehicles than under the no purchase subsidy scenario. The price of the higher polluting conventional vehicles are altered by the manufacturer dependent on the predicted penalties (vehicle price is proportionally adjusted to forecast emissions and penalties to deter sales). As such, lower predicted penalties result in a lower conventional vehicle price. This therefore further increases the financial attractiveness (and therefore utility and resultant sales market share) of ICEV relative to other powertrains, in comparison to the no-subsidy scenario.

• R&D investment is lowered: Similarly, with lower penalties being predicted, the manufacturer is less stimulated to invest in R&D as meeting emission targets is supported by subsidised EV sales. Due to this, the cost and technical performance of EV components do not improve as rapidly as in the no purchase subsidy scenario, R1P0. This results in a lower score for the alternative powertrain attributes and a higher purchase price for them. Thus, the utility of the FCV is lower in the R0P1 -4 scenarios and therefore they achieve lower sales market share growth. The same is true for HEV that is not supported by purchase subsidies. BEV and PHEV sales are boosted by the subsidies but this favourable effect is attenuated after 2030 due to lower R&D investment yielding less favourable utilities between 2030 and 2050 when compared with R1P0. FCV gains no success when subsidies are in place and this results in a severe inhibition of maturity. As FCV is less mature than the PiEVs at the start of the simulation, and becomes available later than these powertrains, it is more greatly affected by the lack of R&D investment. As FCV sales do not develop, neither does the hydrogen infrastructure, thus further lowering the utility of the vehicle compared to R1P0. This then perpetuates throughout the simulation, as predicted sales (based on current sales) are also lower than R1P0 and accordingly the forecast profits from FCV are lower. As potential profits determine R&D spend, investment is

never as much as R1P0 and therefore components never mature to the same level, so sales of FCV remain low.

4.6. Projected CO2 emission regulations (R2Px)

Comparing the projected regulations scenarios (R2) to the current regulation scenarios (R1), there are no differences in 2020, as the conditions are the same. By 2030, we see a higher deployment of alternative powertrains when the projected regulations are in place. Most noticeable is that FCV has a small market share that was not achieved in the previous scenarios. The stricter CO2 target and related predicted penalties for the manufacturers drive decisions for R&D investments. The manufacturers make the same investment decisions as in the current regulation scenarios, up until 2025 (the model is set to consider known changes 5 years in advance). From this point, although investment in BEV and PHEV is marginally increased, most R&D funds go into FCV as it has the greatest potential technology/cost improvement remaining, and therefore provides the greatest leverage for reducing the fleet CO2 emissions and avoid penalties. The biggest impact by 2050 of the projected emissions occurs when no subsidies are in place (R2P0), as PiEV sales are increased compared to the R1 scenarios by around 50%. In all the purchase subsidy scenarios, 2050 PiEV sales are only slightly improved compared to the current regulations scenarios. ICEV sales are more than halved to around 20%, with the majority of these sales attributed to FCV, now selling better than BEV. This would suggest that more ambitious, long term fleet emission targets are most beneficial to FCV. In relation to the previous discussion on FCV market failure due to lack of investment, the higher targets encourage the R&D investment in FCV required to ensure success. In future work, a closer look at the tipping points to where this occurs may be useful in order to better set target and penalties in future regulation. Similar to the previous discussion on current emission regulations, although having purchase subsidies in place in addition to regulations would seem to make an impact on EV uptake, particularly in the earlier years and for PHEV, they would also seem to be marginally less favourable for HEV and FCV by 2050, and could result in making ICEV more attractive to the user in later periods. Furthermore, there is little impact from increasing the duration of the purchase subsidies.

4.7. Extended CO2 emission regulations (R3Px)

Our final set of initial policy scenarios considered even more ambitious CO2 emission regulations, compared to those previously discussed in 4.6 (R2Px). Similar trends observed also persist within this category of scenarios. There is no difference by 2020 to the current regulations as there is no difference in the scenario design for this time-frame. By 2030, we see a slight increase in alternative market shares on the projected regulations (around 1% point for each powertrain), despite a 20% higher emissions target in 2030. Again, this lower than expected impact arises from a combination of investment decision dynamics and intra-powertrain competition. However, by 2050 there is a much more noticeable impact compared to the projected emission regulation scenarios. ICEV shares are halved, HEV reduced by around a third, and BEV, PHEV and FCV have increased by around 40, 30 and 20% (resp.), when purchase subsidies are in place, and 50, 40 and 20% without. Again in these scenarios the early addition of EV purchase subsidies can create a certain technology lock-in in favour of BEV and PHEV, penalising the currently less mature FCV, though the effect is somewhat diminished by 2050.

4.8. High purchase subsidy scenarios (RxP4)

As seen in Fig. 1 and in comparison to the P0-3 scenario counterparts, high purchase subsidies can make a difference on sales, and


G. Harrison, C. Thiel / Technological Forecasting & Social Change xxx (2016) xxx-xxx

have led to the greatest EV shares by 2020 of all scenarios by over 100% or 300% when no purchase subsidies are in place. By 2030, the impact is not so great, other than in the no regulation case (R0). Here, the ICEV share is up to 8% points lower, BEV has a 2% share where there was none before, and PHEV share has more than tripled to 7%. When regulations are in place, however, ICEV shares are similar to the previous scenarios when high purchase subsidies are in place, though there is a 20% improvement in total EV sales shares from being more competitive with HEV, rather than ICEV. Similar observations carry through to 2050, where the subsidies have also reduced FCV sales, and ICEV share is slightly higher for similar reasons as already remarked upon. Therefore with just these simple tests we can deduce a similar conclusion to previous work that even high purchase subsidies do not appear to have long term impact, but can stimulate the EV market when it is in its infancy, which may be particularly beneficial in cases of an otherwise failing market (Shepherd et al., 2012). Although this finding may not be novel, it does support confidence in the integrity of the PTTMAM.

4.9. Infrastructure subsidy scenarios (RxPxI1-3)

The results of the tested infrastructure scenarios are presented in Fig. 2. 2020 is not shown as there is no impact by this point, despite infrastructure subsidies being in place for ten years, compared to the initial policy scenarios. Furthermore, as can be seen, there is little impact even by 2050 for any duration of infrastructure subsidy. There is only a marginal decrease in conventional vehicle share that is slightly more pronounced when the longer duration subsidies are in place. Under current regulations (R1), FCV still does not achieve any market penetration despite the 100% subsidising of H2 infrastructure. This is because the powertrain itself is never strong enough to compete with the PiEVs that were introduced earlier in the simulation. For the projected and

extended emission scenarios, R2-3, the longer duration infrastructure subsidies have led to slightly lower FCV shares, as the subsidies are more beneficial for the already stronger PiEV powertrains. The PiEV on the other hand, have already profited and similar competing dynamics as previously observed have set in to suppress the FCV market. In fact, FCV never has shares greater than the comparative scenarios without infrastructure subsidies are in place, and only when based on the highest purchase subsidy scenario (P4) does the longest duration infrastructure subsidy achieve the highest FCV share. In this case, the increased utility of the FCV from the increased number of hydrogen fuelling stations due to subsidies leads to the greater market shares, taking shares otherwise given to PiEV under the comparative scenarios. The reason for this would appear to be in the detail at an attribute level. Although the infrastructure subsidies have increased the convenience attribute (relative to the base), leading to greater combined utility (and thus shares) up until around 2030, it is the popularity attribute (based on vehicle stock), causing a lower combined utility from this point. This is because the additional infrastructure has been favourable to the PHEV rather than BEV, where shares are lower (or at least the same). Once this effect kicks in it begins to perpetuate as the BEV popularity remains consistently lower than the base scenario, causing even lower shares. This effect has then led to lower forecasts for future BEV, and thus lower forecast infrastructure requirements. The lower investment in infrastructure has then reduced the convenience for PHEV and thus lower shares for this powertrain also. In summary, it would seem to be that the infrastructure subsidies tested here are most beneficial for PHEV, but without strong manufacturer regulation is not strong enough to be self-sustaining for any EV penetration.

These results have to be considered in conjunction with the deployed infrastructure. Fig. 3 shows the cumulative number of pCPs installed through the simulation under all scenarios. Solid lines are


Fig. 2. Infrastructure scenarios results. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)


G. Harrison, C. Thiel / Technological Forecasting & Social Change xxx (2016) xxx-xxx

Fig. 3. Charging point installation. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

those scenarios with no infrastructure subsidies in place. For those based on current vehicle purchase subsidies (P1), there is little difference when the short-lived infrastructure subsidies (dotted line) are in place. This is because the PiEV shares are only marginally higher as discussed above. When the infrastructure subsidies are extended beyond 2020 (dashed lines), the rate of installation is greater, but still no deviation between these scenarios is seen until after 2030. Under the most ambitious regulatory scenario (R3), there is a greater pCP installation without infrastructure subsidies than any duration of subsidies combined with the lower regulatory targets. Having the higher purchase subsidies in place (lighter coloured lines) follows similar trends, though the shortest-lived infrastructure subsidies does result in a higher initial rate of infrastructure deployment than the base, as early PiEV shares are higher. Setting these findings within the context of the impact of the infrastructure subsidies on PiEV shares, it would seem to be clear that pCP installation is disproportionate to market shares. In other words, infrastructure subsidies may be successful in increasing pCP provision under certain conditions, but the magnitude of the effect does not feed back into PiEV sales success.

The share of fuel stations carrying hydrogen that have been deployed under the different infrastructure subsidy levels are shown in Fig. 4. Under the current fleet regulation scenarios (blue lines), as no FCV share builds up, there is never enough stimulus to build H2 infrastructure, even with the longest infrastructure subsidies in place. In fact,

for all remaining scenarios, there is no infrastructure until the late-2020s when, in our model results, the FCV enters the market with larger numbers. As such, even the mid-term infrastructure subsidies only support an initial small number of H2 fuelling stations. H2 coverage grows to only a maximum of 7.5%, with the extended regulations being most successful. Under the projected and extended fleet emission regulations, there is only a significant H2 infrastructure installation when the infrastructure subsidies remain in place through to 2050 (longer dashed lines). Yet still, due to low FCV shares, coverage does not exceed 40%. As can be seen in Fig. 4, those scenarios with high purchase subsidies (lighter lines) have lower H2 infrastructure deployment, as the FCV shares are lower, as observed above.

The results reveal that the PiEV profit more from the infrastructure subsidies than the FCV, and this leads to a certain technology lock-in, resulting in a reduction of deployed FCV. The lower FCV share again has an impact on the effectiveness of the hydrogen infrastructure subsidies. This stresses the importance of finding the right balance, degree, and timing of policy measures in order to avoid undesirable effects.

5. Sensitivity to charge point targets

Focus now turns specifically to PiEV and their charging infrastructure. As mentioned in 2.3.3, the DAFI indicates a desire for a ratio of 10 PiEVs per pCP (PiEV/pCP) for member states. To understand the

Fig. 4. H2 fuel station coverage. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)


G. Harrison, C. Thiel / Technological Forecasting & Social Change xxx (2016) xxx-xxx

Fig. 5.2050 Percentage change in PiEV share and CP installation under R1P1I3 conditions with varying PiEV/CP targets. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

sensitivity to this, and in the light of the disproportional relationship between pCP installation and market share identified in the previous section, we ran our baseline scenario with long term infrastructure subsidies (R1P1I3) several times at different PiEV/pCP ratios, looking at the impact in 2050 on sales market share, installed pCPs and assumed cost. As can be seen in Fig. 5, which demonstrates the divergence as the PiEV/ pCP ratio differs from 10, a general observation is that doubling the PiEV/pCP ratio roughly halves both the number of installed pCPs and cost to authorities in our model results, up until around 25PiEV/pCP, after which the target has no bearing on the total costs or installed pCP. Indeed, whereas with a target of 10PiEV/pCP around 50% of installed pCPs were funded by the authorities, for 25 PiEV/pCP this reduces to 5% and below that <0.2%. It does not, however, have a similar correlation with market share. It can be seen that as the PiEV/pCP ratio decreases, both installed pCPs and their costs have a significantly greater and more rapid increase than market share, which is more pronounced in Fig. 5 due to the secondary axis. On the other hand, increasing the PiEV/pCP ratio follows a shallow reduction as PiEV stock is still low. This suggests that the successful impact on share is relatively insensitive to desired ratio at levels below 5 or above 25 PiEV/pCP.

Leading from this insight, we explored if our model results would indicate some saturation level of the PiEV/pCP ratio target. We carried out tests (of the varying PiEV/pCP ratios used previously) under the infrastructure scenarios with long term 100% infrastructure subsidies when projected or extended emissions targets are in place and current or high purchase subsidies are in place, as well as the baseline (I3, I6, I9, I15, I18). The resulting evolution of PiEV stock and sales shares are shown in Fig. 6. The charts reveal the impact of the purchase subsidies

over time, which was not clearly observed in the previous results where we concentrated on three distinct points in time. Both the sales shares (on the left) and the stock shares (on the right), follow similar trends, as one would expect. Concentrating on sales, the two scenarios with the high purchase subsidies in place (P4), have advanced the take-up of PiEV compared to the current purchase subsidy scenarios P1. With the highest level of subsidy in place the rate of uptake was much more rapid. The 2015 local peak in P4 scenarios shows the impact when the offered discount is reduced from 75 to 50%. It highlights the importance of carefully designing the phase in and out of subsidy schemes as the market will react very sensitively to abrupt changes. This effect could be observed in Estonia in autumn 2014, when the EV subsidy scheme expired (Thiel et al., 2015). In the period during which the 50 and 25% discounts are in place (2015-2025), the calculated rate of uptake does not appear to differ from the scenarios which have no subsidy in place at this point. This suggests that the magnitude of the EV purchase subsidy would need to be high enough so that the EV price does not exceed the price of an equivalent conventional car by > 50%. A second observation is that the extended regulatory targets R3 have not only led to shares around 10% greater than their counterparts, as discussed in 4.7, but also seem to be marginally less sensitive to charge point targets as the range between the results is smaller. The 2050 sales share bandwidth between pCP targets for each scenario varies between 16 and 30% for all tests and 15 and 26% between only 10 and 100 PiEV/pCP. We also notice that in all scenarios, there is a very low sensitivity to pCP provision when the target is >25 PiEV/pCP, whereas the most ambitious targets of < 5 PiEV/pCP, could make a large difference. The final observation is that there is no obvious deviation between PiEV/pCP ratio scenarios until after the PiEV sales share is between 10 and 20%. For the stock this share is at about 5%. It should be noted that our model does lack the sophistication of a spatial element regarding infrastructure deployment, and as such may be vulnerable to underestimating the impact of public charging post provision, eg regarding "charging highways". Nonetheless, our findings support evidence that early PiEV adopters are more likely to rely on home charging (Tran et al., 2013; Namdeo et al., 2014; Hoen & Koetse, 2014; Plötz et al., 2014; Campbell et al., 2012), and furthermore indicate the need for widespread infrastructural deployment as EV adoption becomes more mainstream with users who are more reliant on public charging infrastructure.

The final step of this part of the study was to look at individual countries. We therefore focus on the range of results from the eight different PiEV/CP targets for selected countries under high infrastructure subsidies, current purchase subsides and projected regulatory targets (R2P1I3). We have not replicated the policies that exist in each country in reality, but still apply an EU wide policy. Future research will focus on this. We chose four countries to consider, representing different levels of GDP and vehicle stock growth, as shown in Table 4.

Fig. 6. Sales (left) and Stock (right) market shares under various scenarios. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this


G. Harrison, C. Thiel / Technological Forecasting & Social Change xxx (2016) xxx-xxx 11

2010 GDP and vehicle stock and 2010-50 growth for representative countries.

GDP: H > 30 k, L < 10 k; GDP growth: H > 100%, L < 50%; Stock: H > 10 M, L < 3 M; Stock growth: H > 80%, L < 40%. Country 2010 GDP/capita 2010-50 GDP/capita growth 2010 vehicle stock 2010-50 vehicle stock growth

Germany (DE) High Medium High Low

Ireland (IE) High Medium Low Medium

Poland (PL) Low High High High

Bulgaria (BG) Low High Low Medium

Demonstrated in the results (Fig. 7), it was found that the generalisation derived about the EU in the previous section, that deviation occurs around 5% of stock share, i.e. charge point provision takes affect after that point, did not appear to hold entirely true and appears to be slightly dependent on country characteristics. Our model results indicate that those countries with greater levels of GDP/capita have more rapid EV deployment (DE and IE), and appear to require a similar EV stock share than the EU28 level for infrastructure provision to take effect. For the countries with lower GDP/capita EV uptake starts later and would appear to be impacted at lower stock share than 5%, but are less sensitive overall as they have a lower range between PiEV/pCP targets. It is not surprising that a higher GDP/capita leads to more rapid or earlier EV deployment, given that EV's have a higher purchase price than conventional counterparts when they first appear on the market. Users in countries with lower GDP/capita may simple be priced out of the EV market. It further makes sense that DE and IE would have more similar behaviour to EU as their average GDP/capita is closer (though slightly higher) to EU average than PL or BG (which are both much lower). The countries with higher overall achieved EV shares are more sensitive to charge point provision, demonstrated by the wider bandwidth between the PiEV/pCP target scenarios, than their counterparts, confirming what was said above for the EU also on a country level. The countries with lower vehicle stock (IE, BG) also had wider bandwidth than their counterparts. The lower sensitivity to CP provision of countries with larger stock and smaller EV share witnessed in PL and BG would suggest that these countries may have benefited from the more rapid deployment of infrastructure in other countries that has reduced infrastructure costs. Thus the provider will expect higher return on investments and install more infrastructure before the government subsidies are needed (bearing in mind that in this study, due to structural limitations of the model, infrastructure targets are related to government subsidies). The finding that each member state responds differently to the policies gives further credence to tailored policies for member states, as long as they are in line with EU

regulations and do not lead to market fragmentation. Furthermore, it highlights the importance of the interaction between member states, and how poorer states may benefit from the activities of those who can bear the higher early stage costs. This can give insight for the MSs when designing their national policy frameworks for the implementation of the Directive on the deployment of alternative fuels infrastructure.

6. Conclusion and outlook

This research has sought to bring about further understanding for policymakers regarding the interaction between e-mobility and related infrastructure. The model results of the tested scenarios have determined the following key policy insights: Purchase subsidies:

• Even very high purchase subsidies alone did not lead to long-term EV market success.

• Subsidies benefit available EV technologies in the short term, even in the absence of other policies.

• Longer duration of subsidies did not make significant impact. They can provide market impetus in the beginning but market growth beyond initial deployment needs to be sustained by market mechanisms other than subsidies.

• Due to technology competition dynamics, offering EV purchase subsidies before all technologies are available could lead to technology lock-in and inhibit long-term maturity of less developed technologies.

Fleet Emission regulations:

Having long term emission target regulations in place is necessary for technology transition.

Competition for R&D funds between alternative powertrains seems favourable for FCV when long term targets are in place, provided

Fig. 7. Influence of CP provision (desired ratio of EV to CP) on EV stock share in various member states. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)


G. Harrison, C. Thiel / Technological Forecasting & Social Change xxx (2016) xxx-xxx

that hydrogen is produced via low carbon pathways.

• The most ambitious long term targets benefit PHEV and BEV when subsidies are also in place.

• Higher regulatory emission targets appear to reduce sensitivity to charge point provision.

Infrastructure provision:

• Success of infrastructure subsidies is strongly tied in with success of other policies and existing competition dynamics.

• Charge point provision appears to impact PiEV uptake when PiEV stock share is over 5% in the EU but would appear to be dependent on total stock volume and GDP.

• EV take-up is relatively insensitive at target levels below 5 or over 25 PiEV/pCP.

We can conclude that without stringent fleet emission regulation targets in place for automobile manufacturers a significant transition towards e-mobility, as a potential enabler to meet wider transport and emission reduction targets, is unlikely to take place, regardless of supportive policies aimed at users. Furthermore, it would appear that FCV could have negligible deployment in the absence of strong regulation. There are significant interactions between EV options, especially when vehicle subsidies are in place. Whereas vehicle subsidies are favourable towards PiEVs, this can lead to very low shares of FCVs, which can be tipped back into success by H2 fuelling infrastructure subsidies and fleet CO2 emission penalties. These findings would suggest that subsidies are only beneficial in the earlier years of market introduction and should cover all technologies. Nonetheless, our results would also give some support that White Paper targets may be met.

Our findings indicate that some form of minimum infrastructural targets could be beneficial though further research is needed to identify the exact saturation levels, and it is beyond the scope of this model to comment on the impact of spatial deployment. There is a correlation between EV uptake and infrastructure subsidies, but in our modelled scenarios it appears to be weaker than vehicle purchase subsidies or manufacturer fleet CO2 targets. Our study results support the hypothesis that early EV adopters are less reliant on the provision of public charging infrastructure. Greater infrastructure provision is necessary to increase the convenience of PiEVs and thus their overall utility as perceived by the user, but for the user to realistically consider purchasing a PiEV in the first place, more exposure to the new powertrains is required in early years to build awareness and enter the users' decision set.

Tipping points for regulatory targets should be considered more in depth, based on the initial observations presented here, in preparation for future policy discussion and implementation. This could include a consideration of upstream emissions. Finally, this study has not directly

considered vehicle size/segments, which could offer further insight for discussions. The work presented in this paper relies on a complex model. As described in Section 3, the model is extensive and detailed, but remains a simplified representation of the decision processes of market players. The purpose of the model is to capture system interactions and feedbacks as concisely as possible and focus on impacts that system changes can make on overall outcomes, with a view of understanding the key relationships and tipping points within the system being studied. The attraction of this is that recommendations for policy design can be made on the basis of computer simulations avoiding costly policy experimentation. Many assumptions have been made to ensure that the model is as simple as possible, yet representative enough to consider important feedback loops. Naturally, this introduces uncertainties into the confidence in the model results, as does any uncertainty over the decisions within the scenario design. However, the strength in system dynamics modelling lies not in a prediction or forecasting capacity, but in the understanding of behaviours of variables, and the comparison of scenarios. It is important that any party drawing from the results of this study is fully informed of this limitation. Therefore, the impact of uncertainties of model assumptions are somewhat mitigated as they are consistent across all scenarios, and model calibration and optimisation routines have suggested that data deviation should have minimal impact on behaviour. Uncertainties in scenario assumptions are less troublesome, because as time goes on, even if they turn out to be not as realistic as assumed, they are still valid scenarios at this point in time, spreading between minimum and maximum possibilities. Furthermore, the model is adaptable enough to be improved over time with continued learning, and availability of new data and information. Future iterations may also include other under-researched system elements such as differing business models employed by the manufacturer or modal shifts adopted by the consumer. Going forward, much more detailed analysis, focusing on sensitivities and tipping points, will be carried out over a greater range of scenarios to reflect on further policy options, not just regarding charging infrastructure but also wider energy and transport policy goals. Such analysis can be further enriched by linking with other models such as detailed emission, GIS, power dispatch or energy system models.


The model used in this research was initially developed by one of the authors and Guzay Pasaoglu (JRC) with Lee Jones, Andrew Hill, and Alexandre Beaudet (all Ventana UK). Additional feedback on the manuscript and previous versions was provided by Panagiota Dilara (European Commission, DG GROW) and anonymous reviewers. The views expressed are purely those of the authors and may not in any circumstances be regarded as stating an official position of the European Commission.

Appendix A

Table 5

Sources for key variables under baseline conditions for indvidual member states unless marked EU (as used in (Harrison et al., 2016)).


Historic trend

Future projection


Average household size GDP per capita

Demographic breakdown

Average annual KM

Road network length

Growth in oil price

Growth in alternative fuel price

Electric fuel cost

ICEV fuel cost

(Eurostat, 2014a) EC, (2013b) (Eurostat, 2014b)

World Urbanisation Prospects (UN, 2013) TRACCS (EMISIA, 2013) and trend extrapolation (Eurostat, 2014c; CIA, 2014) and expert assumption Extrapolation from EU Reference Scenario (EC, 2013b) Expert assumptions based on growth in oil price (Eurostat, 2014d)

(Tremove, 2010; BCN, 2014) and expert assumptions

EU Reference Scenario (EC, 2013b)

Calculation based on EU Reference Scenario (EC, 2013b)

EU Reference Scenario (EC, 2013b)


G. Harrison, C. Thiel / Technological Forecasting & Social Change xxx (2016) xxx-xxx

Table 5 (continued)


Historic trend

Future projection

Hydrogen fuel cost Electricity CO2 intensity Hydrogen CO2 intensity (EU) Biofuel blends (EU)

Circulation tax Registration tax Fuel tax growth Fuel tax

Expert assumptions based on other fuel costs EU Reference Scenario (EC, 2013b) (Pasaoglu etal., 2012) EU Energy Statistics (EC, 2012)

TRACCS (EMISIA, 2013) and trend extrapolation

(Tremove, 2010)

EU Reference Scenario (EC, 2013b)


ACEA 2014. Overview of Purchase and Tax Incentives for Electric Vehicles in the EU, in. European Automobile Manufacturers Association, Brussels.

ACEA, 2016. Alternative Fuel Vehicle registrations: +20.0% in 2015; +21.1% in Q4. Association of European Automobile Manufacturers ( (accessed 03/08/16)).

Bakker, S., Jacob Trip, J., 2013. Policy options to support the adoption of electric vehicles in the urban environment. Transp. Res. PartD: Transp. Environ. 25,18-23. http://dx.doi. org/10.1016/j.trd.2013.07.005.

Batley, R.P., Toner, J.P., Knight, M.J., 2004. A mixed logit model of UK household demand for alternative fuel vehicles. Int. J. Trans. Econ. 31, 55-77 (

BCN, 2014. Europe's energy portal.

Beggs, S., Cardell, S., Hausman, J., 1981. Assessing the potential demand for electric cars. J. Econ. 16,1-19.

BNEF, 2016. Global EV sales outlook to 2040. Advanced Transport - Research Note. Bloomberg New Energy Finance LP.

Boksberger, B., Ulli-Beer, S., Bouza, M., Wokaun, A., 2012. Assessing threats and opportunities of induced technology change: long and short cycles in the car maker industry. 30th International Conferences of the System Dynamics Society, St. Gallen, Switzerland.

Bosshardt, M., Ulli-Beer, S., Gassman, F., Wokaun, A., August, 2007. Developing a Diffusion Model of Competing Alternative Drive Train Technologies (CADT model), in: 2007 International Conference of the Systems Dynamics Society, Boston, 29 July - 2.

Brand, C., Tan, M., Anable,J., 2012. The UK transport carbon model: an integrated life cycle approach to explore low carbon futures. Energ Policy 41,107-124. 10.1016/j.enpol.2010.08.019.

Brand, C., Anable, J., Tran, M., 2013. Accelerating the transformation to a low carbon passenger transport system: the role fof car purchase taxes, feebates, road taxes and scrappage incentives in the UK. Transp. Res. A Policy Pract. 49, 132-148. http://dx.

Brownstone, D., Bunch, D.S., Golob, T.F., Ren, W., 1996. A transactions choice model for forecasting demand for alternative-fuel vehicles. Res. Transp. Econ. 4, 87-129.

Campbell, A.R., Ryley, T., Thring, R., 2012. Identifying the early adopters of alternative fuel vehicles: a case study of Birmingham, United Kingdom. Transp. Res. A Policy Pract 46, 1318-1327.

Capgemini, 2008.10th Annual Global Automotive Study: Tracking Consumer Buying Behavior in Both Mature and Emerging Markets.

CIA, 2014. The World Factbook. (accessed 23/09/14).

Dagsvik, J.K., Wennemo, T., Wetterwald, D.G., Aaberge, R., 2002. Potential demand for alternative fuel vehicles. Transp. Res. Pt. B-Methodol. 36, 361-384. http://dx.doi. org/10.1016/S0965-8564(01)00013-1.

Deloitte, 2011. Unplugged: Electric vehicle realities versus consumer expectations. Deloitte.

Diamond, D., 2009. The impact of government incentives for hybrid-electric vehicles: evidence from US states. Energ Policy 37, 972-983. 2008.09.094.

Diwaker, V.D., Nedumthaly, S.R., Prakash, R., 2013. Objective based modelling and simulation within a context of an electric vehicle development program. EVS27, Barcelona, Spain, November 17-20 2013.

Donati, A., Dilara, P., Thiel, C., Spadaro, A., Gkatzoflias, D., Drossinos, Y., 2015. Individual Mobility: From Conventional to Electric Cars, in: JRC Science and Policy Report. Publications Office of the European Union, Luxembourg.

EC, 2012. EU energy in figures. Statistical pocketbook 2012. publications/doc/2012_energy_figures.pdf (accessed 24/09/2014).

EC, 2013a. Clean Power for Transport: A European Alternative Fuels Strategy. European Commission.

EC, 2013b. Energy, Transport and GHG Emissions. Trends to 2050. Reference Scenario 2013. European Commission.

EC, 2014. In: E. Commission (Ed.), Communication - A Policy Framework for Climate and Energy in the Period From 2020 to 2030 [COM(2014)15].

EC, 2016. Communication: A European Strategy for Low-emission Mobility tCOM(2016) 501 Final Available from: European Commission.

Edwards, R., Hass, H., Larive, J.-F., Lonza, L., Maas, H., Rickeard, D., 2014. Well to wheels technical report 4a. JEC well to wheels analysis. Well to wheels analysis of future automotive fuels and powertrains in the European context. In: Godwin, S., Hamje, H.,

Krasenbrink, A., Nelson, R., Rose, K.D. (Eds.), JRC Technical Reports. European Commission, Joint Research Centre, Institute of Energy and Transport, Luxembourg.

Eggers, F., Eggers, F., 2011. Where have all the flowers gone? Forecasting green trends in the automobile industry with a choice based conjoint adoption model. Technol. Forecast. Soc. Chang. 78,51-62.

El Banhawy, E.Y., Dalton, R., Anumba, C., 2014. The agent based modelling of e-mobility. Transportation Electrification Conference and Expo (ITEC), IEEE, Dearborn, MI, USA

EMISIA, 2013. TRACCS Project, in.

Eppstein, M.J., Grover, D.K., Marshall, J.S., Rizzo, D.M., 2011. An agent-based model to study market penetration of plug-in hybrid electric vehicles. Energ Policy 39, 3789-3802.

EU, 2009a. Directive 2009/33/EC of the European Parliament and of the Council of 23 April 2009 on the Promotion of Clean and Energy Efficient Road Transport Vehicles. European Commission.

EU, 2009b. In: European Commission (Ed.), Regulation (EC) No. 443/2009 of the European Parliament and of the Council of 23 April 2009 Setting Emission Performance Standards for New Passenger for New Passenger Cars as Part of the Community's Integrated Approach to Reduce CO2 Emissions From Light-duty Vehicles.

EU, 2011a. White Paper 2011: Roadmap to a Single European Transport Area - Towards a Competitive and Resource Efficient Transport System. European Commission.

EU, 2011b. Regulation 510/2011 of the European Parliament and of the Council Setting Emission Performance Standards for New Light Commercial Vehicles as Part of the Union's Integrated Approach to Reduce CO2 Emissions From Light-duty Vehicles. European Commission.

tpdeEU, 2014a. Directive 2014/94/EU of the European Parliament and of the Council of 22 October 2014 on the Deployment of Alternative Fuels Infrastructure. European Commission.

EU, 2014b. Regulation (EU) No. 333/2014 of the European Parliament and of the Council: Amending Regulation (EC) No. 443/2009 to Define the Modalities for Reaching the 2020 Target to Reduce CO2 Emissions From NEW passenger Cars. European Commission.

EU, 2014c. Regulation (EU) No 253/2014 of the European Parliament and of the Council Amending Regulation (EU) No 510/2011 to Define the Modalities for Reaching the 2020 Target to Reduce CO2 Emissions From New Light Commercial Vehicles. European Commission.

Eurostat, 2014a. Table demopjan - population on 1 January by age and sex. http:// (accessed 28/04/14).

Eurostat, 2014b. Table nama_GDP_c - GDP and main components - current Prices. http:// (accessed 29/04/14).

Eurostat, 2014c. Table road_if_road/motorwa: length of other roads by category of roads/ motorway. carm&lang=en (accessed 18/03/14).

Eurostat, 2014d. Table nrg_pc_204_h - Electricity - Domestic Consumers - Bi-annual Price.

Ewing, G., Sarigollu, E., 2000. Assessing consumer preferences for clean-fuel vehicles: a discrete choice experiment. J. Public Policy Market. 19,106-118. 10.1509/jppm.

FAZ, 2015. (accessed 05/02/15).

Foxon, T.J., Gross, R., Chase, A., Howes, J., Arnall, A., Anderson, D., 2005. UK innovation systems for new and renewable energy technologies: drivers, barriers and systems failures. Energ Policy 33, 2123-2137.

Gardien, L., Refa, N., 2015. EV Infrastructure and Charging Data in the Netherlands. ELAAD NL

Gass, V., Schmidt, J., Shmid, E., 2014. Analyis of alternative policy instruments to promote electric vehicles in Austria. Renew. Energy 61, 96-101. renene.2012.08.012.

Gomez, J., Jochem, O., Fichter, E.W., 2013. EV market development pathways: an application of system dynamics for policy simulation. EVS27, Barcelona, Spain, 17-20 November.

Harrison, G., Shepherd, S.P., 2014. An interdisciplinary study to explore impacts from policies for the introduction of low carbon vehicles. Transp. Plan. Technol. 37, 98-117.

Harrison, G., Thiel, C., Jones, L., 2016. Powertrain Technology Transition Market Agent Model (PTT-MAM): An Introduction, in: JRC Technical Report. European Commission, Luxembourg.

Hidrue, M.K., Parsons, G.R., Kempton, W., Gardner, M.P., 2011. Willingness to pay for electric vehicles and their attributes. Resour. Energy Econ. 33,686-705. 10.1016/j.reseneeco.2011.02.002.

Hoen, A., Koetse, M.J., 2014. A choice experiment on alternative fuel vehicle preferences of private car owners in the Netherlands. Transp. Res. A Policy Pract. 61,199-215.


14 G. Harrison, C. Thiel / Technological Forecasting & Social Change xxx (2016) xxx-xxx

ICCT, 2011a. Policies That Work: How Vehicle Standards and Fuel Fees Can Cut CO2 Emissions and Boost the Economy. Climate works foundation and International Council on Clean Transportation.

ICCT, 2011b. A review and Comparative Analysis of Fiscal Policies Associated with New Passenger Vehicle CO2 Emissions. International Council on Clean Transportation.

IEA 2008. In: Onoda, T. (Ed.), Review of International Policies for Vehicle Fuel Efficiency. OECD/IEA.

IEA, 2015. World Energy Outlook 2015. OECD/IEA, Paris.

IEA, 2016. Energy Technology Perspectives 2016: Towards Sustainable Energy Systems. OECD/IEA, Paris.

IPCC, 2013. Climate change 2013: the physical science basis. Summary for policy makers. Intergovernmental Panel on Climate Change.

ITF, 2008. The cost and effectiveness of policies to reduce vehicle emissions. International Transport Forum.

ITF, 2010. Stimulating low carbon vehicle technologies: summary and conclusions. International Transport Forum.

Janssen, A., Lienin, S.F., Gassmann, F., Wokaun, A., 2006. Model aided policy development for the market penetration of natural gas vehicles in Switzerland. Transp. Res. A Policy Pract 40, 316-333.

Kohler, J., Wietschel, M., Whitmarsh, L., Keles, D., Schade, W., 2010. Infrastructure investment for a transition to hydrogen automobiles. Technol. Forecast. Soc. Chang. 77, 1237-1248.

Kwon, T.-H., 2012. Strategic niche management of alternative fuel vehicles: a systems dynamics model of policy effect. Technol. Forecast. Soc. Chang. 79, 1672-1680.

Lane, B., Potter, S., 2007. The adoption of cleaner vehicles in the UK: exploring the consumer attitude-action gap. J. Clean. Prod. 15,1085-1092. j.jclepro.2006.05.026.

Lane, B., 2011. In: RAC (Ed.), Market Delivery of Ultra-Low Carbon Vehicles in the UK: An Evidence Based Review.

Leiby, P., Rubin, J., 1997. Sustainable transportation: analysing the transition to alternative fuel vehicles. Asilomar Conference on Policies for Fostering Sustainable Transportation Technologies.

Li, G.-D., Masuda, S., Nagai, M., 2014. Predicting the subscribers of fixed-line and cellular phone in Japan by a novel prediction model. Technol. Forecast. Soc. Chang. 81, 321-330.

Lim, D.-J., Anderson, T.R., Inman, O.L., 2014. Choosing effective dates from multiple optima in technology forecasting using Data Envelopment Analysis (TFDEA),. Technol. Forecast. Soc. Chang. 88, 91-97.

Maia, S.C., Teicher, H., Meyboom, A., 2015. Infrastructure as social catalyst: electric vehicle station planning and deployment. Technol. Forecast. Soc. Chang. 100,53-65. http://

Meyer, P.E., Winebrake, J.J., 2009. Modelling technology diffsuion of complementary goods: the case of hydrogen vehicles and refuelling infrastructure. Technovation 29, 77-91.

Mock, P., Yang, Z., 2014. Driving Electrification. A Global Comparison of Fiscal Incentive Policy for Electric Vehicles. International Council on Clean Transportation.

Namdeo, A., Tiwary, A., Dziurla, R., 2014. Spatial planning of public charging points using multi-dimensional analysis of early adopters of electric vehicles for a city region. Technol. Forecast. Soc. Chang. 89,188-200.

Noori, M., Tatari, O., 2016. Development of an agent-based model for regional market penetration projections of electric vehicles in the United States. Energy 96, 215-230.

Norbech, T.E., 2013. Incentives and infrastructure - crucial elements in the build-up of Norway's EV fleet. EVS 27, Barcelona, Spain, November 17-20, 2013.

Nykvist, B., Nilsson, M., 2015. Rapidly falling costs of battery packs for electric vehicles. Nat. Clim. Chang. 5,329-332.

OLEV, 2011. In: D.f.T. Office of Low Emission Vehicles, UK Government (Eds.), Making the Connection: The Plug-In Vehicle Infrastructure Strategy.

OLEV, 2015. (accessed 15/01/15).

Pasaoglu, G., Honselaar, M., Thiel, C., 2012. Potential vehicle fleet CO2 reductions and cost implications for various vehicle technology deployment scenarios in Europe. Energ Policy 40,404-421.

Pasaoglu, G., Harrison, G., Jones, L., Hill, N., Beaudet, A., Thiel, C., 2016. A systems dynamics based market agent model simulating future powertrain technology transition: Scenarios in the EU light duty vehicle road transport sector. Technol. Forecast. Soc. Chang. 104 (March 2016), 133-146 (0.1016/j.techfore.2015.11.028).

Plotz, P., Gnann, t., Kuhn, A., Wietschel, M., 2014. Markthochlaufszenarien für Elektrofahrzeuge. Fraunhofer-ISI.

Plotz, P., Schneider, U., Globisch, J., Dütschke, E., 2014. Who will buy electric vehicles? Identifying early adopters in Germany. Transp. Res. A Policy Pract. 67, 96-109.

Potoglou, D., Kanaroglou, P.S., 2007. Household demand and willingness to pay for clean vehicles. Transp. Res. Part D: Transp. Environ. 12,264-274. j.trd.2007.03.001.

Richardson, B.C., McAllinden, S.P., Belzowski, B.M., Booms, C., Ebarvia, B.N., Hill, K., 1999. Method for estimating market acceptance of technologies and fuels of the partnership for a new generation of vehicles. University of Michigan Transportation Research Institute. Report for US DOE, DOC and DOT.

Robinson, A., Blythe, P.T., Bell, M.C., Hubner, Y., Hill, G.A., 2013. Analysis of electric vehicle drive recharging demand profiles and subsequent impacts on the carbon content of electric vehicle trips. Energ Policy 61, 337-348. 2013.05.074.

Rodrigues, L.L.R., Gupta, K., Motlagh, F.G., 2012. Vehicle fuel consumption modelling: a system dynamics approach. World Congress on Engineering, London, UK. July 4-6.

Rowney, M., Straw, W., 2013. Leading the Charge: Can Britain Develop a Global Advantage

in Ultra-low Emission Vehicles? IPPR Shafiei, E., Stefansson, H., Asgeirsson, E.I., Davidsdottir, B., Raberto, M., 2012a. Integrated agent-based and system dynamics modelling for simulation of sustainable mobility. Transp. Rev. 33,44-70. Shafiei, E., Thorkelsson, H., Asgeirsson, E.I., Davidsdottir, B., Raberto, M., Stefansson, H., 2012b. An agent-based modelling approach to predict the evolution of market share of electric vehicles: a case study from Iceland. Technol. Forecast. Soc. Chang. 79,1638-1653. Shepherd, S.P., 2014. A review of system dynamics models applied in transportation. Transportmetric. B Trans. Dyn. 2,83-105. 916236.

Shepherd, S., Bonsall, P., Harrison, G., 2012. Factors affecting future demand for electric vehicles: a model based study. Transp. Policy 20, 62-74. tranpol.2011.12.006.

Sierzchula, W., Bakker, S., K, M., Wee, B.v., 2014. The influence of financial incentives and other socio-economic factors on electric vehicle adoption. Energ Policy 68,183-194. Stasinopoulos, P., Compston, P., Jones, H.M., 2012. Policy resistance to fuel efficient cars and the adoption of next-generation technologies. 30th International Conference of the System Dynamics Society, St Gallen, Switzerland. Stepp, M.D., Winebrake, J.J., Hawker, J.S., Skerlos, S.J., 2009. Greenhouse gas mitigation policies and the transportation sector: the role of feedback effects on policy effectiveness. Energ Policy 37,2774-2787. Struben, J., Sterman, J.D., 2008. Transition challenges for alternative fuel vehicle and transportation systems. Environ. Plan. B Plan. Des. 35,1070-1097. 1068/b33022t

Thiel, C., Alemanno, A., Scarcella, G., Zubaryeva, A., Pasaoglu, G., 2012. Attitude of European car drivers towards electric vehicles: a survey. European Commission: Joint Research Centre Scientific and Policy Reports. Thiel, C., Schmidt, J., Van Zyl, A., Schmid, E., 2014. Cost and well-to-wheel implications of the vehicle fleer CO2 emission regulation in the European Union. Transp. Res. A Policy Pract. 63,25-42. Thiel, C., Krause, J., Dilara, P., 2015. Electric Vehicles in the EU from 2010-2014 - Is Full

Scale Commercialisation Near? European Commission, Luxembourg Tran, M., Banister, D., Bishop, J.D.K., McCullocn, M.D., 2013. Simulating early adoption of alternative fuel vehicles for sustainability. Technol. Forecast. Soc. Chang. 80, 865-875. http :// Tremove, 2010. TREMOVE economic transport and emissions model. Transport and Mobility Leuven ( UN, 2013. World urbanisation prospects, 2011 revision. Data on urban and rural populations. Percentage of Population Residing in Urban Areas, 1950-2050. United Nations. Valipour, M., 2015. Opimization of neural networks for precipitation analysis in a humid region to detect drought and wet year alarms. Meteorol. Appl. 23,91-100. http://dx. Valipour, M., Sefidkouhi, M.A.G., Eslamian, S., 2015. Surface irrigation simulation models: a review. Int. J. Hydrol. Sci. Technol. 5. (5150).

van der Vooren, A., Brouillat, E., 2013. Evaluating CO2 reduction policy mixes in the automotive sector. Environ. Innov. Soc. Trans. 4,60-83. 001.

Walther, G., Wansart, J., Kieckhafer, K., Schnieder, E., Spengler, T.S., 2010. Impact assessment in the automotive industry - mandatory market introduction of alternative powertrain technologies. Syst. Dyn. Rev. 26,239-261. Wolf, I., Schröder, T., Neumann, J., de Haan, G., 2015. Changing minds about electric cars: an empirically grounded agent-based modelling approach. Technol. Forecast. Soc. Chang. 94, 269-285. Zubaryeva, A., Thiel, C., 2013. Paving the way to electrified road transport: publically funded research, development and demonstration projects on electric and plug-in vehicles in Europe. European Commission, Joint Research Centre Scientific and Policy Reports. Zubaryeva, A., Thiel, C., Zaccarelli, N., Barbone, E., Mercier, A., 2012. Spatial and multi-criteria assessment of potential lead markets for electrified vehicles in Europe. Transp. Res. A Policy Pract. 46,1477-1489.

Dr. Gillian Harrison is a postdoctoral research fellow in electro-mobility modelling at the Directorate for Energy, Transport and Climate, Joint Research Centre of the European Commission, based in the Netherlands. Her interdisciplinary PhD research combined system dynamic modelling, engineering and moral philosophy in the appraisal of policy options for the uptake of low carbon vehicle technologies. Prior to this Gillian worked in environmental management and business reporting in the UK water industry, holding an undergraduate Masters in Chemistry with Forensic Science, a post-graduate certificate in Water and Environmental Management and a Masters in Low Carbon Technologies. Her current research interests are focused around sustainable and equitable transport solutions in a carbon-constrained and automobile-centric society.

Christian Thiel is the project leader of the "Electo-mobility Modelling" project in the Sustainable Transport Unit, at the Directorate for Energy, Transport and Climate, Joint Research Centre of the European Commission. Before joining the European Commission, he worked for 12 years in the European Engineering Centre of Opel. He holds a Masters degree in Environmental Science (Geooekologie) from the Technical University Braunschweig, a bachelor's degree in Biology from Université Paris VI and an undergraduate degree in Economics from the University Frankfurt (Main).