Variability in costs of electrifying passenger cars in Canada

The high cost of purchasing electric vehicles (EVs) compared to internal combustion engine vehicles (ICEVs) is a major barrier to their widespread adoption. Additionally, the price disparity is not the same for all households. We conducted a total cost of ownership (TCO) analysis to compare the net present value of EV versus ICEV ownership for various household categories across Canada. We observed spatial and behavioral factors, including variations in costs of electricity, temperature, household archetypes and their purchase decisions, and access to charging infrastructure. We found that EVs are more cost-effective than ICEVs for certain daily driving distances, but typical households in Canada generally do not drive enough for lifecycle costs of EVs to be less than ICEVs. The province of Quebec has the most favorable conditions for EV ownership due to high purchase subsidies and low electricity prices. Variability in costs across other provinces and territories is mainly due to differences in rebates, electricity and gasoline prices, and tax rates. Our findings have implications for policymakers and consumers. For consumers comparing ICEVs to EVs based on a fixed budget, which may be consistent with how many households frame their purchase decision, willingness to accept smaller, non-luxury EVs can result in large cost savings. We also find that although temperature variation has a minimal effect on TCO, it does impact the ‘number of charge-ups’—a metric that we introduce to compare how many charging cycles a user may expect over the lifetime of a vehicle. The policy implication of this would be a need to consider regional differences in cold weather patterns when planning charging infrastructure deployment and the extent to which households in shared dwellings may face additional costs. Lastly, our findings strengthen the argument that equitably decarbonizing transportation will also require investment in strategies other than electrifying personal vehicles.


Introduction
Road transportation accounted for approximately 10% of global greenhouse gas (GHG) emissions (in CO 2eq ) in 2019 [1].Several studies have shown that electric vehicles (EVs) have the potential to reduce GHG emissions compared to similar internal combustion engine vehicles (ICEVs), although the actual magnitude of the reduction depends on the energy generation mix of the electricity grid that is used to charge the batteries of EVs [2,3].On average, EVs are estimated to be less carbon intensive than ICEVs in even in the most coal-dependent jurisdictions [4].For example, driving an EV produces less GHG emissions for 97% of the population of the United States [5], and even at current grid carbon intensities, EVs are estimated to have lower life-cycle GHG intensity for 95% of global transport demand [6].However, despite these potential benefits for decarbonizing transportation, EVs remain a small but rapidly growing fraction of the total passenger vehicle stock, with the highest estimates at 1% [7].Nations across the world have signaled their intention to electrify their passenger and commercial fleets, the global stock of EVs only reached 26 million registered units by the end of 2022 [8].
These patterns are also apparent in Canada: Transportation accounted for 25% of GHG emissions in Canada in 2018, making it the second largest contributor by sector [9].The abundance of hydroelectricity and other low-carbon sources in several Canadian provinces offers GHG emissions benefit associated to EV driving [10].As with many other countries globally, electrifying the transportation sector is thus a key element of Canada's decarbonization pathway.Canada's federal climate plan framed this opportunity as 'clean and affordable transportation' [11], emphasizing not only the environmental impact of zero-emissions vehicles (ZEVs) but also the financial capability of a broad range of Canadian households and economic sectors to be included in the transportation transition.The federal government's Emissions Reduction Plan (2020) mandates sales of ZEV to account for 20% of purchase by 2026, 60% by 2030, and 100% by 2035 [12].
A combination of federal and provincial subsidies and tax credits have thus far incentivized the purchase of EVs for Canadian consumers.Along with investment aimed at deployment of charging infrastructure, the federal climate plan boosts the uptake of EVs by disincentivizing ICEVs with a rising carbon tax to reach $170/tonne by 2030, and a clean fuel standard.EV sales prior to the release of the climate plan emphasize the role of provincial and local policies in reducing barriers in uptake.In 2020, 3.7% of all new vehicle sales were electric, but over 92.9% of those sales were in Quebec, Ontario and British Columbia, i.e. the only provinces that offered provincial-level purchase subsidies at that time.Historical experience has shown that several technologies typically follow an 'S-curve' of adoption in the consumer market, where some threshold is eventually reached beyond which adoption increases at a faster rate [13].Domain experts generally hold that the 'tipping point' for EVs is 5% adoption in the consumer market [14].
For a broad range of Canadian households, affordability presents a key challenge to meeting the federal climate plan's goals.Affordability also varies by household purchasing power, patterns of vehicle usage, and other constraints, such as geographic region or type of residence.In other jurisdictions, such variation influences consumer benefits of purchasing and driving an EV [15][16][17], continuing its use [18], the cost-effectiveness of EV subsidy policies for GHG mitigation, and uptake of incentives [19]; yet it remains an understudied phenomenon in the Canadian context.
In this study, we ask: How do the costs and benefits of EVs vary across geographic regions and different households and users in Canada?What additional, complementary policies are needed to enable rapid and equitable electrified transport for a range of households across Canada?The impact of federal, provincial/territorial, and local measures supporting EV deployment will be distributed unevenly, hence the need to understand how these policies differentially impact Canadians, not just the statistically average consumer.An understanding of these variations across Canada can help inform the design of more targeted policy interventions.
Canada presents a good case study to explore the variability in costs that we expect may be transferrable to other parts of the world, beyond the US.Firstly, Canada's northern climate allows an exploration of the impacts of variation in temperature.Secondly, the variability of electricity grid intensities in various regions of the country mimic a diverse range of electricity generation mixes.Thirdly, the existence of a federal carbon tax also makes Canada an interesting case as other countries assess the viability of carbon pricing schemes across a national scale.

Total cost of ownership (TCO)
We use a total cost of ownership (TCO) model to evaluate the variable costs and benefits of EV adoption in Canada.The TCO, expressed in net present value (NPV), of a vehicle is calculated as the summation of capital and operating expenses.Capital expenses (Net Capital Cost) include the maximum suggested retail price (MSRP) of a vehicle plus federal and provincial taxes, less the value of salvage or resale of the vehicle.For an EV, it also includes the cost of an EV charging station, less the value of purchase rebates.Operating expenses include fuel and maintenance costs, at net present value, Given the importance of the operational phase in vehicle lifecycle costs, the TCO is strongly a function of vehicle-kilometers travelled (VKT).We focus on an additional metric, the 'breakeven VKT' (VKT BE ), defined as the value of daily VKT at which the difference in NPV between EV and ICEV reaches 0, i.e. the amount of driving an owner must do daily for the total cost of owning an EV to be on par with that of owning the comparator ICEV [15].VKT BE thus allows a comparison of the differential costs of EV ownership under varying conditions in capital and operating costs, and salvage value.A lower value of VKT BE means that an EV is more beneficial at a lower mileage, (2) Finally, we define another metric that is specific to EVs. 'Charge-ups' are the number of times that an EV would be recharged over its lifetime, assuming that its battery goes from a fully discharged state to a fully charged state.While this does not necessarily mimic real-life behavior of how users charge EVs (and may underestimate the number of charge-ups), we use the metric as a simplified comparison of the implications to charging in different scenarios of EV ownership,

Number of charge ups =
Total annual electricity use [kWh] ×Years of ownership Battery pack size [kWh] ×Battery degradation Years of ownership . ( We focus only on direct costs in the TCO which impact households at present day [20].We exclude indirect, or intangible, costs on account of the uncertainty in how these vary geographically and also because we expect such costs to be secondary to direct costs.However, we recognize that other researchers have attempted to include such considerations into TCO analysis of EVs.For example, given that the impact of range anxiety is typically hidden in the conventional approach to TCO, there have been attempts to incorporate it as an intangible cost into the analysis (see review by [21]).Although range anxiety is a major barrier in the adoption of EVs, evidence suggests that this behavioral consideration declines after the purchase and use of an EV [22,23].We also expect that pre-purchase range anxiety will diminish as a factor with increases in the battery capacities of available EV models, the speed of charging technology, and the deployment of and access to charging infrastructure.Thus, we do not include range anxiety as an intangible cost in our analysis and instead use the number of 'charge-ups' metric.
Broadly, conventional TCO studies have focused on demonstrating the long-term superiority of EVs over ICEVs, i.e. the lower operational costs of EVs make them favorable after a certain amount of time [24].The line of inquiry we probe is precisely the opposite, that is, what makes EVs unfavorable-with the goal of identifying what additional policy interventions are needed to support this transition.To do so, we use TCO as a method to explore variability in costs faced by households that purchase an EV.Other researchers have quantified costs based on different travel patterns emerging from variability observed in geographic data [16,17,25] or using abstract categories of consumer behavior [26].Here, we propose a synthesis of the two, as described in the next section.

Variation and archetypes
We conceptualize variation in TCO to exist along two dimensions, spatial and behavioral (figure 1).In different sub-national jurisdictions in Canada, i.e. provinces and territories, we study the effect of variation in fuel prices, electricity prices, electricity grid emissions intensities, tax rates, and climate.In the behavioral dimension, we consider variation in travel behavior, type of vehicle, and access to charging in different types of dwellings.We also consider different approaches to comparing EVs to ICEVs that represent constraints on how consumers choose which vehicle to purchase, that is, based on function or platform (i.e. consumer decides on the vehicle type), or based on comparable capital expenditure (i.e. consumer decides on price cap).
To operationalize how the TCO relationships might vary for a range of Canadian households, we use household and user archetypes that map onto the model parameters.Although these archetype comparisons are simplifications, their utility lies in generating discussion on the real-world impacts of spatial and behavioral variation.
We map spatial variation parameters wholly onto the archetypes of a Typical Household, one for each province and territory across Canada.We assume that the Typical Household in each jurisdiction drives a daily distance that is equal to the average daily VKT of that same location.
Behavioral variation parameters are mapped onto several archetypes: We consider variation across space in the following variables: gasoline price, electricity price, grid carbon intensity, availability of rebates, and tax rates on the purchase of new vehicles (table 1).We selected the provincial/territorial level as the geographic unit for spatial variation analysis as EV deployment policies and tax rates on new vehicle purchases vary at this level of governance.Furthermore, the inter-provincial variation in gasoline prices [27], electricity prices [28], and electricity grid carbon intensity [29] is typically greater than the intra-provincial variation.This is also the most granular level of data availability in the Canadian context.For each province and territory, we took the average electricity rates, sales tax rates, and electricity grid carbon intensities in 2020.For gasoline prices, we used the average of monthly fuel prices of September 2021 to August 2022, to avoid the fluctuations in price during the initial months of the COVID-19 pandemic.

Temperature
One key concern in the Canadian context is the effect of cold weather on the performance of EV batteries [30], so we also consider the spatial variation in temperature.We derived the relationship between ambient temperature and fuel consumption rate for ICEV and EV drivetrains from Du et al [31], who used empirical data collected from 33 600 on-road vehicles to model the additional fuel consumption required for climate control within a vehicle compared to a baseline of 20 degrees Celsius (table 2).We then approximate the effect of ambient temperature by using heating degree-days (HDDs) and cooling degree-days (CDDs)-parameters which are typically used to calculate the energy demand of buildings [32].Here, we used the annual HDD to calculate an average daily ambient temperature for a 'constant heating load' .We then estimated the fuel consumption rate based on this temperature.We applied a similar calculation with CDD for a 'constant cooling load' .We used temperature projections from the Climate Atlas of Canada to approximate HDD and CDD values that are area-weighted for each province and territory [33].We took the average of mean HDD and CDD values in two time periods applied to the climate models used by the Climate Atlas, 1976-2005 and 2021-2050, as an approximation of the temperature effects in 2022.

Behavioral variation 2.3.2.1. Vehicle type and consumer purchase behavior
TCO studies and consumer purchase tools generally compare an EV to an ICEV that is in the same vehicle category and similar in function, if not the same chassis.Since there is typically a disparity in the price of an ICEV versus an EV, the drawback of this method is that it does not accurately represent behavior of all consumers, some of whom may purchase vehicles based on a fixed budget.For a consumer with a fixed budget, an EV in a smaller vehicle category may be equivalent to an ICEV in a larger vehicle category.We thus use two archetypes, Wealthy Consumer and Budget-Constrained Consumer, to compare the reference EV to an ICEV that is similar in function, or to an ICEV that is similar in price, respectively.
To explore a range of vehicle types, we analyzed three vehicle categories, all of which have EV availability in Canada: subcompact SUV, midsize sedan, and luxury compact SUV (table 3).EVs and their ICEV   Sources [35,57].
comparators for the first two categories were selected from the highest selling models across Canada in the first half of 2022 [34,35].We also included the luxury compact SUV to explore TCO relationships for more expensive vehicles that do not meet the threshold of the federal and provincial EV rebates.

Subcompact SUV (base case):
We take the base case scenario throughout this analysis to be the subcompact SUV, Hyundai Kona, which has both electric and internal combustion drivetrains based on the same chassis.Hence, Hyundai Kona represents the closest approximation of a direct comparison between an EV and ICEV by function.For the Budget-Constrained Consumer, Toyota RAV4, was chosen as the ICEV comparator, given that it was the highest-selling vehicle in Canada in the first half of 2022 [34].
Midsize sedan: Tesla Model 3 was selected as the EV reference for midsize sedan, as the top-selling EV model in Canada [34].In popular media, this vehicle has frequently been compared to Toyota Camry [36].For comparison by function, we selected the most fuel-efficient Toyota Camry, which is also the trim at the lowest price.While there is a large price disparity between these two vehicles, this selection allows Tesla Model 3 to be compared with a popular midsize sedan ICEV that has low operating costs.
Luxury compact SUV: Compact SUV broadly was the top-selling vehicle category in Canada in the first half of 2022, hence we selected the luxury segment of this vehicle category.In addition, Tesla Model Y was the only luxury compact SUV EV available in Canada in 2022.
For this analysis, we choose to compare only battery electric vehicles (BEVs), and not plug-in hybrid electric vehicles (PHEVs).In the case of PHEV, some combination of gasoline and electricity is utilized for operation of the vehicle, where the proportion of the latter has been factored into the operating costs of a TCO analysis as the 'utility factor' [37].While there is some empirical evidence to suggest the value of utility factor that users tend towards [38,39], we consider BEV in our analysis as the bounding case for full electric use in terms of operating costs.Moreover, BEVs are typically more expensive than PHEVs, thus BEV also represents a bounding case for the capital costs.We then expect that in our analysis, the TCO of PHEV would be situated between ICEV and BEV.

Charging access
Type of dwelling can dictate the charging behavior of EV users.Level 2 EV supply equipment (EVSE), i.e. charging stations, can typically be installed in single-detached homes at the homeowner's discretion.On the other hand, shared dwellings, such as apartments or semi-detached houses, may have shared electrical infrastructure or lack dedicated parking, and hence, EV owners in these types of dwellings may face challenges in installing EVSE [40].Such households are expected to rely more heavily on public charging stations than ones with EVSE installations at their place of residence, referred to here as 'private charging' .
Presently, there are approximately 10 000 public charging stations across Canada, including 1650 Level 3 fast charge stations, all of which are largely located close to densely populated urban areas [41].Canada's federal government projects a need for a ratio of 24 EVs per public charging port in 2030 [42].
To explore the variation that a Shared Dwelling Household may face in public charging costs, we compared electricity rates at Level 2 and 3 public charging stations (table 4).While many public charging stations remain free for use, we expect that operators will eventually implement a fee structure.For Level 2 charging, we assume a constant rate of $2 CAD per hour, which is a typical rate charged by fee-for-use Level 2 stations.Level 3 fee structures are only beginning to emerge, making it difficult to determine with certainty what the spatial variation in costs related to the use of publicly available fast chargers.We use the fee structures of Petro-Canada's 'Electric Highway' Level 3 charging station network [43].Although Tesla Superchargers are the most spatially extensive network of Level 3 chargers across Canada, the lack of access to vehicle makes other than Tesla does not allow an adequate analysis of spatial variation in public charging costs.
In addition, we also note that households with access to private charging do not have a need to travel for refueling, unlike households with EVs that must rely on public charging, or households with ICEVs.However, this benefit is considered to be modest, and we do not include in the analysis here [44].

Inter-household variability
Several other factors can contribute to variability of households within a province.We consider two aspects here, daily VKT and choice of electricity pricing structure.For VKT, we use data from Statistics Canada's Canadian Vehicle Survey 2009 [45] to calculate an average daily VKT for each province and territory, which we suppose is the distance driven daily by the Typical Household in each respective jurisdiction.While this survey data may be outdated, we expect that the relative variation across jurisdictions within Canada to remain similar.
This data set represents the highest resolution on VKT across Canada that is publicly available and consistently estimated.Some specific jurisdictions have more granular data: for example, we use travel behavior data in the Metro Vancouver region (Canada's third largest urban center) [46], to comment on our expectations for the intra-provincial household variability across Canada.
The other factor which we explore is the choice that a household makes in its electricity pricing structure, i.e. conventional flat rate, or rates based on time-of-use (TOU) pricing [47].The latter allows households to preferentially charge EVs at a time of day when electricity is priced at a lower rate.However, the TOU rate is only available to households with access to private charging, which thus reinforces the barrier to households that rely on public charging.
To observe the extremes of inter-household variability, we thus use a case study of Metro Vancouver: the Single Detached Household where VKT is higher and with private charging at a cheaper electricity rate [48], versus a Shared Dwelling Household where VKT is lower and reliant on public charging.Coincidentally, the City of Vancouver had the lowest fraction of single-detached houses of all major urban regions in Canada in 2016 [49], and could therefore be among the cities that have the highest proportion of households that face barriers to private charging access.

Future scenario
To explore the impacts of future policy directions and technological advancements, we compare the base case scenario for the Typical Household in Quebec at present day to a future scenario in 2030 in which both policy and technology changes would have occurred.For policy changes, we assume that the 2030 carbon tax value of $170 per tonne of CO 2 e is in effect, affecting both gasoline and electricity prices, and that provincial EV purchase rebates are longer available.For technology changes, we assume a reduction in the price of Hyundai Kona EV by an amount equivalent the cost of batteries dropping from $151 per kWh in 2022 to $100 per kWh [50].We expect that decline in manufacturing costs may be offset by increases in the price of lithium, thus we assume that BloombergNEF's forecast for 2026 will hold to 2030.

Effect of battery degradation
Automotive batteries degrade over time and because of charging behavior [51].We include impacts of battery degradation in the TCO model, using annual battery degradation rate of 2% [52] as a baseline.Battery degradation rate has no impact on the TCO relationships, or in simpler terms, a fixed amount of battery storage capacity costs the same to charge, regardless of the number of times a battery is charged.Thus, the qualitative impact of battery degradation on a consumer-i.e. the inconvenience of charging more times to drive the same distance-is entirely hidden by the TCO relationships.Thus, we explore this impact with the total number of charge-ups metric.

Assumptions
We assume a time horizon of 7 years, to approximate the average ownership life of newly purchased vehicles by Canadian drivers, at 6.4 years [53].
For salvage value of vehicles, we used Schloter (2022) to set a depreciation rate [54].Schloter (2022) compiled publicly available data from 8576 vehicles in the US, which we assume is as a reasonable approximation of the Canadian used vehicle market.After 7 years, we derived that ICEVs and EVs retain 40% and 35%, respectively, of their original sale prices.
EVs are generally considered to have lower maintenance costs than ICEVs, although there is some evidence to suggest that new EVs can be more expensive to maintain in the first year of operation and then decline relative to ICEVs thereafter [55].We use constant maintenance costs of USD $0.06 and $0.10 per mile for EV and ICEV, respectively [58].
For NPV, we use 8% discount rate, which is the recommendation by Treasury Board of Canada for cost-benefit analysis calculations [56].We conduct a sensitivity analysis with 5% discount rate to assess the impact that lower cost capital would have on the TCO.

Spatial variation: typical households across jurisdictions
We evaluate the TCO relationships for Typical Households in all ten provinces and three territories in Canada.The Typical Household in Quebec has the most favorable conditions for EVs, requiring the lowest VKT BE -or daily distance for the difference in NPV of EV and ICEV to be equivalent-at 46 km per day (VKT BE = 40 km d −1 at 5% discount rate, or 13% lower).Although none of the Typical Households across Canada drives as much as what would be required for NPV to breakeven with a comparable ICEV, the difference varies markedly across jurisdictions.In Quebec, vehicles are driven 18% less than the distance that is required for VKT BE , whereas in Nunavut, average VKT is nearly a factor of 12 lower than what is required for VKT BE (figure 2).
From a financial perspective, the Typical Household in each jurisdiction faces varying costs to own and operate an EV versus an ICEV.Most favorably, the Typical Household in Quebec would over the lifetime of a vehicle pay $2.4 K CAD more for an EV than an ICEV, whereas the Typical Household in Nunavut pays $17.7 K CAD more.The high cost faced by the latter is not only on account of the weakest ∆NPV to VKT relationship, but also the low average VKT at 17 km per day.In addition to costs, each jurisdiction also faces varying impacts in the number of 'charge-ups' that an EV would require over a 7 year lifetime.The Typical Household in Quebec would charge an EV nearly twice as much as the Typical Household in Nunavut, given the low average VKT of the latter (table 5).
To explore what drives the variation of costs across the country, we compare the Typical Household in Quebec with all other jurisdictions to quantify the effects of spatial parameters-i.e.gas price, electricity price, rebate, tax rate, and temperature-on VKT BE , individually while holding all other parameters equal (figure 3).For example, the contribution to VKT BE variation due to the difference in rebates between Ontario and Quebec is considered as the hypothetical scenario of Quebec with same rebate amount as that of Ontario.Negative values in this figure indicate that a spatially varying parameter contributes to a higher VKT BE than in Quebec (i.e. is less advantageous for an EV), and vice versa for positive values.If we take the net of the differences of each of these individually varied spatial parameters, we note that it does not equal the actual difference in VKT BE that we observe between Quebec and another jurisdiction (represented by the dots in the figure).This is due to the interaction effects that occur between the spatial parameters.Nevertheless, the all-else-equal analysis allows a comparison of the relative impact of individual parameters as well as an estimate of the magnitude that each contributes to variation.Taken individually, the impact of the spatial variation parameters are as follows.
Rebates: Reductions in the capital cost of the EV and charging equipment reduce the net total cost, thus rebates shift the entire TCO relationship to lower VKT BE .All provinces and territories have a lower total rebate than Quebec, contributing 10.5-28 km more of travel per day to VKT BE .
Electricity cost: Higher electricity prices increase the operating costs of an EV, resulting in a higher VKT BE .Electricity prices contribute 5-12 km of travel more per day relative to Quebec.
Gasoline cost: Lower gasoline prices make the cost of operating an EV less favorable, resulting in a higher VKT BE .Quebec has among the highest gasoline prices in Canada, so in comparison, gas prices across most of the rest of Canada contribute to a higher VKT BE .
Tax rate: Higher tax rates increase the capital cost of both vehicles.However, in the base case examined here (i.e.Hyundai Kona EV versus ICEV), the EV costs more than its comparator ICEV.As a result, a higher tax rate penalizes the EV more than the ICEV.Quebec has among the highest tax rates in Canada, so in comparison, the lower tax rates across most of the rest of the country contribute to a lower VKT BE .However, the differences in tax rates between jurisdictions is low (6%-10% of vehicle price), with the exception of Alberta and the three territories, where new vehicles are not subject to a sales tax.This reduces VKT BE by over 7 km per day relative to Quebec, where the tax rate is 9.975%.
Temperature: Here, we observe that temperature has as negative effect on VKT BE in jurisdictions with a warmer climate than Quebec.This counter-intuitive result occurs because there are two temperature-dependent effects, one related to EV efficiency and the other to ICEV efficiency.Since electricity is cheaper per energy unit in Quebec than the price of gas, the net differential in the sensitivity calculation for the same reduction in temperature favors EVs, even though the efficiency effect favors ICEVs.(Electricity is cheaper per energy unit than gasoline in all jurisdictions except Saskatchewan, Northwest Territories and Nunavut.We conducted a similar all-else-equal analysis for the latter and found that, indeed, a comparison to colder temperature would result in higher VKT BE .)Hence, with all else being equal, jurisdictions that are   on average colder than Quebec have a small net benefit, and the provinces that are warmer on average have a small net penalty relative to Quebec.We note that what is most salient here, however, is the relatively minor contribution of temperature to variation in the TCO compared to all other spatial parameters.Quebec is the most favorable jurisdiction for EVs because rebates are the primary driver of TCO variation observed across the country.The exceptions are Northwest Territories and Nunavut, where high electricity cost-due to heavy reliance on diesel and fuel oil for power generation [59]-is a larger contributor to the difference with Quebec.Differences in gasoline prices and tax rates generally contribute less to variation across Canada, except in cases where there is a distinguishing feature in these two parameters, such as the absence of sales tax on new vehicles in four jurisdictions.Alberta also has no fuel tax at the provincial level, resulting in the lowest gasoline price in Canada (by 4%-21%).The stark difference in Alberta's gasoline price with the rest of the country (12% cheaper than the average of the rest) amplifies the contribution of this parameter to VKT BE relative to other jurisdictions.
Finally, the contribution of temperature is the least of all spatial parameters analyzed here.The range of difference between the coldest and warmest jurisdictions in Canada is less than 2.5 km per day in VKT BE , thus we can conclude that the variation in temperature across Canada is likely not great enough for many households to perceive as an impact to total costs of EV ownership.

Behavioral variation 3.2.1. Vehicle type and consumer purchase behavior 3.2.1.1. Comparison by function
From the perspective of a Wealthy Consumer, unconstrained by price and able to purchase a vehicle as desired, EVs cost more than ICEVs.The subcompact SUV (Hyundai Kona EV Preferred), luxury compact SUV (Tesla Model Y Long Range AWD) and midsize sedan (Tesla Model 3 Standard Range RWD) would require the Wealthy Consumer to drive 68, 82 and 240 km per day, respectively, break even in costs with a similar ICEV (figure 4).As expected, the capital cost of the luxury SUV includes a premium above the non-luxury subcompact SUV.On the other hand, the midsize sedan has a far higher VKT BE , primarily due to the low capital cost of the vehicle selected as a comparator (Toyota Camry LE).However, this selection was deliberate to emphasize the market availability of low-cost, fuel-efficient midsize sedan ICEVs.Conversely, the Tesla Model 3 is priced higher relative to ICEV vehicles in its own category (Toyota Camry LE) than Hyundai Kona and Tesla Model Y relative to their comparators ICEVs.

Comparison by price
The Budget-Constrained Consumer is expected to operate under a fixed budget for car payments or vehicle purchase.Thus, comparison by price is a more realistic scenario to understand the total costs of ownership for many consumers.Here, EVs have the advantage of lower operating and maintenance costs compared to ICEVs that are similar in price.For all three of the vehicle categories analyzed, the Budget-Constrained Consumer purchasing an EV yields a positive NPV even at low daily driving distances (at 40 km per day, NPV is $22k, $24k, $19k for subcompact SUV, midsize sedan, and luxury compact SUV, respectively).
Although EVs have a higher cost for the Wealthy Consumer archetype (i.e.comparing like for like), switching from a larger ICEV to a smaller EV would result in cost savings.We took the NPV of each vehicle category in Quebec at the province's average VKT, 40 km per day, and compared the cost differential between a switch from an ICEV to an EV in the same vehicle category (Wealthy Consumer) or an EV at the same price (Budget-Constrained Consumer) (table 6).For a Wealthy Consumer willing to switch from a luxury compact SUV ICEV to one of the two smaller EVs, there is a cost savings of $32k to $55k in NPV.The Budget-Constrained Consumer with sufficient budget for a luxury full-size SUV ICEV would save $19k to 80k in NPV by switching to smaller EVs.We note that in our analysis, the Budget-Constrained Consumer has a high floor on budget constraints.A consumer constrained to a budget equivalent to the price of the ICEV sub-compact SUV (Hyundai Kona) at $23k would, in fact, not have any options for an EV.Such a consumer is effectively priced out of the market, presently.

Charging access
Shared Dwelling Households (i.e.public charging only) tend to face greater costs to charge EVs than Single Detached Households (i.e.private charging only).To explore the effects of private versus public charging, we contrasted these two household archetypes in Quebec.Whereas real-world consumer behavior may involve mixed use of private charging and public charging at Level 2 and 3, the three scenarios-private charging only, public charging at Level 2 only, and public charging at Level 3 only-provide a range of possibilities.These three scenarios are progressively more expensive for the consumer (see table 4 for charging rates).Given that Level 3 public charging prices are the highest in Quebec, and the private (i.e.residential) rates are the lowest, the magnitude of difference between the bounding cases is thus a proxy of the variation across jurisdictions between a Single Detached Household and a Shared Dwelling Household (figure 5).
At the average VKT of 40 km per day in Quebec, the Shared Dwelling Household charging at Level 2 would pay nearly $6000 more than the Single Detached Household in NPV.For the Level 3 charging only, the Shared Dwelling Household would pay nearly $9000 more than the Single Detached Household.With respect to VKT BE , the high price of Level 3 charging in Quebec in a distance that is well over what is driven daily by the Typical Household in Quebec, into a range that only a Superuser may drive.

Inter-household variability
As we have observed, EVs are more favorable as daily VKT increases, and therefore households that travel less are face greater costs from EVs. TOU pricing structures for electricity have the potential to exacerbate this difference.To explore the impact of inter-household variability, we compare costs across the Metro Vancouver region of British Columbia.In all municipalities of this dense urban region, VKT in 2017 [10] is  lower than the provincial average.Thus, municipalities in this region may represent the higher ranges of the costs faced by households in the province.
Here, we compare the most favorable case, the Single Detached Household in the peripherally located suburb of Maple Ridge with TOU pricing, to the least favorable case, the Shared Dwelling Household in Vancouver that is constrained to a flat rate on electricity pricing.In other words, it is a case of long daily commute and cheap electricity at home ($7.7k), versus a case of short daily distance and limited to expensive electricity at a public charger.The difference between the two cases, $6.6k, is the range of additional cost that households in Metro Vancouver would face in purchasing an EV over an ICEV (table 7).
This variation in additional costs is driven largely by VKT variability (all else equal, 41%-87% additional costs for an EV in Vancouver, on account of the lower VKT compared to Maple Ridge).In the case of Vancouver, the daily VKT is, in fact, low enough that the cost of a charging station installation is not offset by the savings resulting from the cheaper electricity rate of private charging, thus the Single Detached Household faces a (slightly) larger cost than the Shared Dwelling Household.

Future scenario
In the 2030 scenario, the combined effects of the carbon price at $170 per tonne and a plausible reduction in vehicle price outweigh the lack of purchase rebates: VKT BE would reduce from 46 to 28 km per day for the Typical Household in Quebec (figure 6), well below the average daily VKT that is driven in the province.
In the long-term time horizon to the 2035 ZEV passenger vehicle sales mandate, the political feasibility of imposing taxation to meet the sales mandate may be limited in under a government that may be less willing to accept EVs as emissions reduction measure.Industrial policy may be one approach that avoids political backlash and helps to ensure the cost competitiveness of EVs with ICEVs.

Using costs of ownership to explore policy gaps
In applying the TCO approach to Canada, a country with wide geographic variability in factors implicated in vehicle ownership costs, we find that households across the country face varying costs of owning an EV versus an ICEV.These costs vary both spatially, because of differences in gasoline prices, electricity prices, grid carbon intensity, purchase rebates, tax rates, and temperature; and behaviorally, because of differences in charging access, travel patterns and consumer preferences.Less VKT is required for an EV to equal an ICEV in total costs of ownership when gasoline prices are high, electricity prices are low, purchase rebates are available, tax rates are low, and temperature is warmer, and when households have access to private charging and prefer a non-luxury vehicle.Our results are aligned with similar findings in the US which show that there is variability in costs faced by households in different locations and that differ in daily VKT, and that variability of these parameters is in no case sufficient for EVs to cost less than ICEVs [17,25].No similar study exists for Canada, but our analysis of archetypes offers insight on why, as other research has suggested, EVs are suitable for only a fraction of Canadian households [60].
Rather than emphasizing the point that EVs are typically more expensive than ICEVs for most consumers, our analysis allows us to identify possibilities for policy interventions.The approach that we take in this study extends the literature that uses TCO to assess variability for different types of consumers to develop policy insights.The archetypes that we use are generalizations of possible households, and thus, provide an approach for policymakers to assess the variability that exists across the country, and thereby identify potential gaps in EV deployment policy.In other words, in exploring exactly who wins or loses in the era of transition to EVs, we can glean insights on where policymakers can look to next for increasing EV adoption.
From a methodological perspective, this study presents an alternative to how TCO analyses are typically framed.Comparing an EV to an ICEV that is equivalent in price, i.e. the 'Budget-Constrained Consumer' , may be a more behaviorally realistic comparison for many consumers.Moreover, this framing more clearly demonstrates the advantage of EVs even at low VKT.

Policy implications
As observed in other parts of the world, Canadians who initially adopted EVs had use cases that resulted in cost savings over ICEVs, or attached some non-financial value to EV ownership, such as pro-environmental behavior [61].Achieving the 2035 mandate of 100% ZEV sales will require moving beyond early adopters and towards Typical Households.This will likely require a cohesive and mutually supportive suite of policies at all three levels of governance, i.e. federal; provincial and territorial; and local.Given that decarbonization of the transportation sector is a key component of Canada's overall climate commitments per the Paris Agreement, the task is even more critical for Canadian policymakers.The following policy implications emerge from our analysis of TCOs for households across Canada.Where possible, we also remark on potential impacts of, and challenges that may arise from, implementation.
EVs can exacerbate transportation inequalities: Passenger vehicle ownership is already inequitable for some social groups in North America [62], and the additional costs of EVs can further the affordability gap [63], create new access barriers related to public charging infrastructure [64], and also perpetuate the 'lock-in' of car-based transportation [65].As our analysis shows how much more expensive EVs are in some places and for some households, this study adds evidence that the transportation inequalities inherent in a personal transportation-oriented system cannot be alleviated by widespread adoption of EVs.Thus, while there is a significant emissions reduction potential that could be realized by electrification of passenger cars, other pathways to decarbonizing transportation should not be deemphasized-active transport and public transit must also play in deploying clean and affordable transportation for Canadians [66].Although calls to increase the feasibility of active transportation have historically faced political challenges [67], we are optimistic that the rare circumstances of a system-level transition can be leveraged to draw more attention to transportation inequalities.
Incentivizing smaller EVs will reduce total costs for consumers: Policies that are designed to encourage consumers to replace larger ICEVs with smaller EVs would take advantage of the lower ICEV-EV price disparity for smaller, non-luxury vehicle categories.As an increasing number of cheaper EV models enter the market, policymakers could reassess the purchase price cap on rebates as a means of limiting the size of vehicles that are incentivized.Alternatively, as the market moves beyond early adopters, a luxury vehicle tax based on size may also encourage consumers to purchase smaller, non-luxury vehicles.Additional considerations for policymakers include the well-established co-benefit of greater traffic safety from smaller vehicles [68] and the emerging understanding that electric SUVs have less emissions benefit than smaller EVs [69].However, the increasing consumer preference over the past two decades for light trucks and SUVs in North America [70,71] could limit the effectiveness of such a policy.
We must also draw attention here to the fact that there are presently no EV alternatives for Budget-Constrained Consumers who are less affluent to make the switch from ICEVs.Given that the current availability in the EV market is barrier for less affluent households, it would therefore be pertinent to track how the price of vehicles is spread across the fleet over time.Otherwise, some households risk having a delayed entry into, or perhaps even being left out of, the transition to EVs.
Cold weather may require more public charging infrastructure deployment: Canada's colder weather requires more energy consumption for climate control by EVs than by ICEVs.While the impact to TCO is minimal, households in colder climates must charge their EV batteries more frequently resulting in greater demand for public charging infrastructure.For example, the Typical Household in Quebec would have to charge an EV a total of 434 times, 5% more than in warmer Alberta, even though the Typical Household in both jurisdictions travels the same average VKT, at 40 km per day.Further analysis is needed on the spatial density requirements that account for greater demand for public charger utilization in colder regions.As Canada's climate is colder than the US, public charging station per vehicle deployment goals for the latter country [72] may not be sufficient in the context of Canada.Naturally, the constraining factor here is the limit on public funding available to invest in public charging infrastructure.However, a thorough identification of charging needs, including consideration of the cold weather penalty, would allow funders to take advantage of emerging practices in reducing charging infrastructure costs, e.g.procurement in large volumes to access discounts [73].
High public charging costs will be faced by households in shared dwellings: Households in shared dwellings (such as apartments) are typically disadvantaged by constraints on charging EVs at home [74].Such households may have to rely more heavily on public charging infrastructure, and therefore disproportionately face higher costs than households that are able to charge at home.Policymakers should turn their attention to helping offset the additional costs associated with EV costs for those living in dwellings without access to private charging.Such policies may be most appropriately implemented at a local level to complement other policies that are designed to improve access to charging for such households.One approach could be that public chargers adopt differential pricing so that certain households could pay a lower rate.However, such a policy may be difficult to implement given that charging infrastructure is typically operated by firms that set the price to a market rate.Alternatively, TOU pricing in public chargers may be easier to implement and would allow anyone to benefit from lower pricing during off-peak periods.For households reliant on public chargers, this could offset some of the additional operating cost versus households with access to private charging.
Incentivizing superusers is an effective emissions reduction strategy: Superusers, or use cases of high VKT, stand to gain the most TCO benefit from using an EV versus an ICEV [16,75].In addition, incentivizing superusers would also minimize the cost of emissions abatement.The first case of such an approach would be a law passed in the US state of Vermont to offer EV purchase incentives to individuals using more than 1000 gallons of fuel annually and to those with low to moderate income [76].Our analysis supports the notion of shifting incentive availability to behavioral metrics based on VKT, rather than a blanket availability of incentives.In the US, the latter has resulted in EV adoption by mostly wealthy households [77], and although such households typically have higher daily VKT than low-income ones [78,79], many of those EVs are purchased as a second vehicle [39] and are driven less than ICEVs [80].We would then expect that a key implementation challenge of such a policy would be to identify which low fuel-efficiency vehicles are driven the most.To go a step further by coupling this with an additional criterion of income threshold could address equity concerns, but nevertheless, the former criterion alone would increase the emissions reduction potential beyond a blanket availability of EV incentives.
In the absence of purchase subsidies, a carbon price may be key to increasing the adoption of EVs: The combined effect of a carbon tax of $170 CAD per tonne (expected by 2030) and plausible purchase price reductions could help adjust the total costs of ownership of an EV versus an ICEV when purchase subsidies end.On the other hand, if the carbon tax were to be repealed by a future change in government, the total costs of EV ownership would be worse off than present day.While 'taxing to adoption' has political consequences, a carbon price does have a utility in adjusting the economics of ICEV versus EV.In regions of Canada where the absence of provincial-level purchase subsidies is a deterring factor, such as in the Atlantic region [81], the carbon price may play a more prominent role in driving adoption of EVs.
Given the 2035 ZEV sales mandate on light-duty vehicles, Canadians seeking to purchase a new vehicle would increasingly face limited choice options for ICEV.In such circumstances, it could be expected that some households would defer the purchase of new vehicles to replace current ones, and consequently, the average ownership life of vehicles would increase.The result would be a delay in the stock turnover of light-duty vehicles to ZEV, dampening the emissions reduction potential.
Decarbonization of the electricity grid is critical to maximizing emissions reduction potential but may require additional policy support to reduce consumer costs in affected jurisdictions: Provinces with lower carbon intensity electricity have greater emissions reduction potential with EV adoption than ones with higher carbon intensity [82].While we focus on consumer costs in this study, the analysis could be extended to assess the variation of emissions abatement costs of EVs across Canada.Given that the ZEV sales mandate affects the entire country, not just the provinces with clean electricity grids, there is a risk that the distribution of EVs across Canada in 2035 may not achieve the maximum emissions reduction potential.Thus, federal policy focusing on decarbonization of the electricity grid will be critical in optimizing emissions reduction potential.Projects such as the Saskatchewan-Manitoba intertie and the Atlantic Loop would distribute the benefits of low-carbon electricity provinces with others that have a less favorable endowment of energy resources [83].
If the capital costs of such projects were borne partly by consumers via higher electricity prices, the result would be a greater disparity in NPV between ICEV and EV.EVs would cost even more for Typical Households in jurisdictions transitioning to lower carbon intensity electricity.Thus, there would be a need to assess whether additional EV adoption policy supports are needed to mitigate this potential trade-off.
While the above policy recommendations could help to advance EV adoption, it should be noted that transport electrification may not full achieve GHG mitigation targets for the transportation sector, as research suggests is the case for the US [84].This highlights the emerging need for policymakers to couple transport electrification with other measures that could reduce the demand for travel.

Limitations, other considerations, and future work
Post-pandemic travel patterns: Although we use the most recent data available on VKT variation across Canadian provinces (2009), we note that this data is out of date.In the US, average VKT continued to increase after 2009 [58], so we would expect that Canada may have also experienced a similar pattern of VKT growth.Patterns of travel behavior have changed in the post-COVID-19 pandemic era [85], including less trips as a result of an increased incidence of teleworking [86] and a shift from public transport modes to private ones [87].Although the long-term impacts are unclear, some evidence suggests that the net effect in the years immediately following the pandemic could be a slight decrease in car traffic from pre-pandemic levels [88].Following our results, a post-pandemic teleworker who drives less would then have a larger gap towards meeting the required VKT for an EV to breakeven with an ICEV.
Efficiency of EV subsidies: While we note that EV subsidies are critical in achieving decarbonization goals, we must also acknowledge that the implicit subsidies available to fossil fuels [89] are a fundamental issue in necessitating such large-scale subsidization of EVs in the first place.One study estimates that the implicit subsidy on gasoline in Canada is equivalent to $0.20 per liter [90].Removal of the implicit subsidy would decrease breakeven VKT for the Typical Household in Quebec to 39 km per day (provincial average = 40 km per day).Recognizing the political challenges of fossil fuel subsidy reform as a climate policy measure [91], we nevertheless make note that the economic rationality of EVs is inherently disadvantaged by implicit subsidies on gasoline.
Technology advancements: A limitation of studies on EV costs is that the rapidly changing technologies may render results outdated in the near-term future.For example, vehicle-to-grid (V2G) technology (and favorable fee structures with local electric utilities) could reduce the operating costs of EVs; smart power distribution could increase the overall number of EVs that could be charged at public stations and dwellings with shared electrical capacity; rapid charging could reduce recharging time; and the introduction of wireless and inductive charging to supplement conventional conduction-based charging infrastructure could help to reduce range anxiety [92,93].Although our analysis does not include intangible costs, such as range anxiety and the time to refuel, our results provide insights on which future trends in EV and charging infrastructure technology could be prioritized to reduce tangible costs.The introduction of new technologies may also necessitate other costs.Generally, V2G technology could reduce the operating costs of EVs, but this would have to be weighed against an increase to the capital cost of an EV capable of bidirectional charging.For EV users living in dwellings with shared electrical capacity, the deployment of smart power distribution would have the knock-on benefit of reducing reliance on public charging infrastructure, thereby reducing the costs that such households face.On the other hand, the installation costs of such smart upgrades may be borne by the households themselves, which would increase the initial capital costs.As the value of these costs become clearer, future TCO analyses could incorporate such considerations.

Figure 1 .
Figure 1.Parameters of spatial and behavioral variation included in the study design.

Figure 3 .
Figure 3. Spatial variation of breakeven VKT relative to Quebec.

Figure 4 .
Figure 4. Comparison of net present value differential across different vehicle types.

Figure 5 .
Figure 5. Private versus public charging in Quebec.

Figure 6 .
Figure 6.Net present value differential in Quebec in 2023 and a 2030 scenario.

Table 1 .
Provincial and territorial electricity, gasoline, and temperature variation.

Table 2 .
Relationship between fuel consumption rate and ambient temperature.

Table 3 .
Electric vehicles and internal combustion engine vehicle comparators.
a Denotes comparator vehicles that are in a larger vehicle size category than the reference EV.

Table 4 .
Public charging rates across provinces and territories.
[43]tro-Canada's Electric Highway network is not presently available in these jurisdictions.We assume a middle-range rate of other jurisdictions, $0.33 per minute.Source[43].

Table 5 .
Number of charge-ups for typical households over vehicle lifetime.

Table 6 .
NPV difference of switching between vehicle categories (in thousand dollars) for QC at VKT = 40 km day −1 .

Table 7 .
Intra-regional variation in VKT and choice of electricity billing structure.