FEATURE: The “peak car” phenomenon has captured the attention of many but what does it really mean? This is a deep dive into mobility trends of today and where this might steer us into the future.
In a 2017 post, one of us analysed the “peak car” phenomenon, critiquing the data it is based on and querying whether there really were long-run and fundamental changes occurring to how people can and do move about their urban environments.
Here, we discuss peak car further, considering whether more recent vehicle and travel data (particularly from the US and Australia), provide evidence for a more enduring and profound change to urban passenger transport.
We are interested in both the nature of change (how are people travelling instead, if at all?) and evidence regarding causes of change.
The enduring peak car discussion
The peak car concept has achieved a lot of air-time since the global financial crisis.
Eminent international transport researchers including Peter Newman and Jeff Kenworthy described peak car as the attainment and retreat from maximum Vehicle Kilometres Travelled (VKT) per capita, in many cities globally, including a range of traditionally car-dependent cities in North America and Australia.
Significantly, they argued, car ownership and use seemed to be decoupling from income, and city governments were decoupling urban growth from the primacy of the car as a means to meet peoples’ mobility needs.
It is reasonable to expect a plateauing of VKT per capita based on saturation of vehicle ownership and access to vehicles, population demographics, and limits to peoples’ time available to travel, purposeful trips to be conducted and average travel speeds.
Still, analyses by researchers such as Newman and Kenworthy pointed to plateau and decline across many international cities (up to 2011 or so), citing a range of positive megatrends as contributing factors.
In addition to income decoupling and less car-centric urban growth, these megatrends included potential reductions in spatial reach by car (owing to increasing congestion), declining trip lengths, and reduced demand for mobility generally.
Timothy Garceau and others, focusing on the US, concluded similarly regarding the plateau and decline in VKT emphasising fundamental changes in behaviour across cities with quite different spatial and socio-demographic characteristics.
Researchers such as Michael Sivak at the University of Michigan argued that societal changes (as well as economic factors) appeared to contribute to VKT trends.
A recent article in The Economist speaks to a historic ideal for the average Joe/Josephine to strive for vehicle ownership (and use) because cars were the embodiment of personal freedom and representative that one was “keeping up with the Joneses”.
In the US, some researchers used driver licensing rates as evidence of peak car, pointing to particular declines among under 30s. Julie Beck, writing for The Atlantic, speculated that licensing and subsequent car ownership may be less of practical necessities than they once were and, therefore, less necessary for personal freedom.
During the first decade or so of this century, data was showing uptrends in use of active and public transport.
Simultaneously, many (led by millennials and empty-nesters) were flocking – or at least, wanting to flock – to newly-regenerating city centres in pursuit of the amenity that density and diversity can offer as part of urban consolidation trends (indicative of the move away from car-centric growth).
The upward trajectory of fuel prices also disincentivised private vehicle use.
Apparently, travel behaviour change has been reinforced and accelerated by embracement (especially among millennials) of the sharing economy, including growing access to car share, ride-hailing and other mobility services.
These services offer quasi-private vehicle travel on-demand without the hassles of licensing, ownership, insurance and securing parking spaces.
Mobility services have been facilitated by technology and technology-led entrepreneurialism and innovation have influenced patterns of consumption including travel.
Various commentators have argued that millennials are a generation that would rather have a smart phone than a car.
In reality, today’s technophiles do not really have to choose between the two, but the assertion is that the more people rely on devices, the more they would rather play with them than worrying about driving a car.
What a virtuous cycle it could be; growth of inner cities and demand for inner-city living accelerating demand for increased amenity and investment in quality transport choices and services like the mode-of-the-moment, e-scooters.
Ripe conditions, perhaps, for peak car. More recent data and broader analyses may paint a less optimistic picture.
Looking at longer-run and emerging trends
Fuel prices as well as various other macro-economic factors have undoubtedly affected all aspects of private vehicular mobility, including licensing, vehicle purchase, and mode choice and VKT.
There was a widespread view in the 2000s that oil scarcity would trigger a steep and protracted rise in fuel prices, and that this would force travel behaviour changes.
On the contrary, shale and tar-sand extraction, improved fuel efficiency driven by environmental regulation and greater global exploration/discoveries have contributed to much less severe (if still volatile) prices observable at Australian bowsers.
The Bureau of Infrastructure, Transport and Regional Economics (BITRE) observes that the average household’s expenditure on ownership and operation of cars as a share of total household expenditure on goods and services dropped through the first half of the twenty-teens and seemingly not because the average person was driving less.
Plus, typical vehicle capital cost relative to earnings has dropped substantially.
Meanwhile, demand for SUVs in Australia, for example has increased dramatically over recent years, influencing total numbers of new cars sold.
Generally, all car sales over recent (post-global financial crisis) years have risen but there is year-to-year variance with respect for purchase trends, with some of the most pronounced variance occurring post-2017 as the economy softens.
Demand for cars in the US
US data show more consistent and sharper growth in all sales through to 2016 with cooling sales thereafter.
According to Statista, post-2014 dips in demand for cars were offset by significant growth in demand for light trucks (for example, pick-ups) with 2018 sales accounting for about 70 per cent of all light vehicle sales.
The post-2016 cooling of US sales reported by Statista correlates with oil price rises and higher gas prices.
Still, according to a February 2019 commentary by Bloomberg, the recent cooling is a portent of an approaching, new peak car moment (ostensibly, some sort of Peak 2.0 following the peak identified in the 2000s).
The cooling itself, the article argues, results from the ubiquity of “new” mobility options such as ride-hailing, as well as reinvigorated civic concerns about congestion and reaffirmed belief in the sharing economy.
The eventual peak will correlate with widespread deployment of automated vehicles on to the mobility market. This explanation begs a more considered definition of “peak”.
In our view, there are limited differences between a kilometre ridden in a light passenger vehicle and one driven, if overall VKT remains the same.
More about the potential effects of vehicle automation, later.
For now, it’s worth noting that May sales of passenger vehicles in the US were strong, correlating with similar uptrends in key economic indicators.
Chris Loader’s detailed and comprehensive manipulation of BITRE and Australian Bureau of Statistics data relating to Australian VKT and licensing trends through to 2018 allows easy review of recent Australian metadata.
The data show a plateauing of total VKT around 2004-2006 and a rise since then.
Most of the growth is accounted for by cars rather than light commercial and heavy vehicles including buses.
Per capita data does reflect the declines in VKT observed by others, but these declines flatten from 2012 and rebound for some cities – Adelaide and Perth, particularly. US data, in the aggregate, show a similar rebound effect from 2014.
The licensing data show that the national licensing rate per 100 persons is trending upwards and there has been some acceleration in licensing rates in recent years in states and territories including Western Australia, Tasmania and Northern Territory.
Slow declines in licensing among people aged 20 to 29 are being offset by older people; especially those 70 and older.
The decline among younger people is particular to New South Wales and Victoria, and less-so, Tasmania.
Mode share data for Australian cities tend to be limited to the Census journey-to-work set.
There are exceptions, such as the Victorian Integrated Survey of Travel and Activity (VISTA) dataset.
Referring again to Chris Loader’s analysis, journey-to-work data at the city level tells a story similar to VKT for states and territories, with dips since 2011 in public transport use for cities including Brisbane, Perth and Hobart, but rises in Sydney and Melbourne.
Average car occupancy appears to be on the decline
A similar, but converse pattern is evident in car only trips. Overall, average car occupancy appears to be dropping.
Still, while public transport use in New South Wales has risen ten times faster than the long-term historical average, anyone driving on a Saturday in Sydney will attest to the fact that the private vehicle remains the mode of common choice.
A recent analysis by the authors shows that the increase in public transport usage is not a mode shift, it correlates with historically high population growth.
In fact, despite the severe crowding on public transport, the same effect is evident on the road network.
With some $40 billion in rail and road investment, mode shift to public transport appears to be a fraction of a per cent over the past three years.
In the US, pundits are lamenting widespread declines in public transport ridership short years after the apparent ridership renaissance reported early this decade by the American Public Transportation Association.
Even then, the apparent renaissance was critiqued for hiding significant variability in trends between cities.
The small handful of cities, such as Seattle, where ridership increases continue to occur, tend to be where investment in public transport infrastructure (and priority), and services, including feeder (bus services) is enduring.
That, and they feature crippling road congestion.
These factors support a basic premise of travel decision-making among the travelling public: cost, convenience and comfort considerations have to be weighed up.
Access to public transport as well as other modes, plus the affordability of those modes are key.
A May 2019 article by University of Sydney researchers Somwrita Sarkar, David Levinson and Hao Wu analysed the polycentricity of Sydney and found enduring spatial inequities.
They concluded that much more change is required in the distribution of homes and jobs to reduce the primacy of Sydney’s Central Business District, and improve housing and employment balances in sub-centres.
These outcomes would likely reduce journey times and lengths as well as improving the price and convenience equation of public transport for more trips including non-commutes and non-peak-hour trips.
The rise of ride-hailing
We anticipate that until relatively recently, peak car researchers did not predict – much like the majority of us – the sharp increase in availability and services offered by new mobility service vendors including Uber and Ola.
These effects have ramped up sharply from around 2015 and research remains exceedingly patchy regarding their nature, magnitude and direction.
Thus, there is limited data regarding how they relate to car- and public transport-based passenger kilometres as well as mode choices.
US research – using richer trip-based datasets than available typically in Australia today – demonstrates the scale of growth in ride-hailing kilometres over the last five years, with Bruce Schaller predicting that by the end of 2018, associated VKT would exceed local bus passenger kilometres.
Also, the per trip VKT by a ride-hailing service exceeds VKT if it is conducted by a private vehicle, owing to dead-head miles (the number of miles a taxi needs to drive from the point of unloading to the point where its new load is ready for pickup).
Others argue that emerging meta-data sets show a close correlation between actual public transport ridership declines and increases to trip numbers using ride-hailing services.
Despite significant growth in ride-hailing trips and associated VKT, the overall share of travel fulfilled by these and other mobility services remains minor, for now.
Still, mode substitution and dead-heading issues are problems for advocates of the sharing economy and tech-led disruption in the transport sector.
Variable trends in private car ownership and usage may mask other trends of equal or greater concern relating to travel choices made instead, including undermining the Peak Car 2.0 argument.
These concerns have found resonance in New York, where a congestion charging scheme is pending to both curtail congestion, which worsened following huge growth in ride-hailing trips and to generate revenue for a chronically under-funded public transport system.
London’s policy-makers have found also that the city’s much-lauded cordon pricing scheme was being undermined by an explosion in car services operating within the cordon travelling often without passengers.
In Australia, infrastructure supply-side responses remain popular and not just to improve the convenience of public transport in the decision-making equation.
New, high-profile plans and projects like Melbourne Metro and Metro 2, Sydney light rail and Perth’s METRONET compete for funding with “congestion-busting” road projects such as North-East Link, Westconnex, and trenches and double-stackers.
Commonly, these are justified on traditional forecasting, safety and travel time savings grounds.
They can serve their intended purpose, in terms of providing some (albeit, often temporary) congestion relief; however, supply-side responses tend to perpetuate particular patterns of travel behaviour.
As such, induced traffic considerations, more comprehensive BCRs and fuller sets of alternatives could factor into project assessment and selection.
In addition, significant rethink is required in financial markets given the many billions nationally being invested currently in new toll roads, which generate profitability through increased VKT.
Returning to price issues and transport affordability, Swedish research by Anne Bastian and Maria Börjesson concluded that economic factors could well explain changes in vehicle use between 2002 and 2012.
They noted also differences within populations based on income, observing limits to vehicle ownership and use (for example, saturation).
We have ourselves noted previously the likelihood of limits to population-level ownership and VKT, and recognise the potential for considerable volatility below these thresholds for all manner of economic and non-economic reasons.
But what of those millennials and their aloof attitudes to car ownership? What about trends, generally towards urban consolidation?
Yet, we think care is needed assuming that decisions made today regarding housing location and its association with mobility (and access) choices will be sustained through life stages, and we are not the only ones.
We also think more scrutiny is required regarding socio-economic and socio-demographic characteristics of those “forsaking cars” and locating in urban centres.
Lang Walker, one of the most successful land use developers in Australia turned focus recently from apartments back to house and land packages, noting that the grand experiment in apartment living was reaching saturation.
His view is that these sorts of housing offers are undersupplied relative to demand.
The cost component of transport decision-making cannot be considered in isolation from other affordability issues – especially those relating to housing.
Plainly, the price and convenience dynamic applicable to peoples’ travel (and car ownership) decisions is influenced strongly by where they live, which in turn, depends on the housing they can afford.
In the 21st Century, housing affordability has decreased globally. Australian house prices surged in the 2000s with particular growth between 2011 and 2017.
Rent and sales prices in major cities across the developed world like London, San Francisco, Toronto, Sydney and Auckland remain at extreme levels even with some evidence of retreat from record highs.
Some reduction does not mean a return to broad affordability; particularly, as wage growth has stagnated in economies such as Australia’s over some years and disposable income evaporates.
This is despite the Reserve trying to reinflate confidence by cutting interest rates to historic lows.
Wealth gaps and transport choices
Undoubtedly, technology and fuel innovation – including in transport – are creating more choice and convenience than ever. They also tend to raise urban wealth.
Still, these factors help to fuel income and buying-power disparities.
Jobs (and living opportunities) in centres are great but the liveability of centres for a broad cross-section of the population depends on both the types of jobs available and affordability of housing.
Even urban evangelists like Richard Florida acknowledge both the triumphs and failures of the return to city centres, noting unprecedented geographies of affluence and poverty, especially in some of North America’s and Australasia’s most liveable cities.
According to recent Canadian research, AirBnB, which is to short-term accommodation what Uber is to the mobility sector, has taken about 31,000 homes (about 1.5 per cent of supply) out of the country’s rental market.
This unintended consequence of technology-led innovation appears to contribute increasingly to reduced housing supply and therefore, all else equal, unaffordability.
In San Francisco, the hyper-successful tech sector has started to take more responsibility for the income polarisation and affordability issues it has contributed to.
About three quarters of this is planned on land owned by the company, which includes holdings in relatively car dependent locations.
Other researchers look at these sorts of issues from alternative perspectives, concluding that trends in urban consolidation are tied to restrictive land use policies and limitations to housing supply more so than population preferences and footloose entrepreneurialism.
These consolidation trends contribute directly to housing affordability with flow-on effects on peoples’ buying power for relatively capital-intensive items like cars.
In our view, there is good evidence of a range of reasons for unaffordability and the lack of consensus helps to show that causes are complex.
So, the return to centres – especially among young people – may signify the trade-offs that a proportion of this generation chooses to make to balance access to employment and education with what they can afford to rent or buy.
In this context, cars become luxury goods that can be forsaken for necessities.
Kevin Naughton’s and David Welch’s article refers to a tech exec from Boston who decided to sell his luxury car and spend up to $20,000 a year commuting using ride-hailing services. We should all be so lucky to be able to value our time so highly and shell out such money on chauffeuring. Presumably, the execs can choose to live where they like.
Still, choice depends on context even for wealthy tech execs. In Dublin, the European headquarters of Google and Facebook, there is no ride-hailing other than taxi and even getting a taxi can be difficult. In this city, the exec would probably retain their luxury car.
There is a lack of readily-available and powerful data demonstrating that the urban influx has been voluntary, profound, preferred, representative (for example, it reflects a broad social strata), and sustained. The same can be said of data showing peak car.
As researchers, we tend to analyse car use based on necessity and choice, a distinction made well by Susan Handy and others at UC Davis.
The ideal is to avoid the former and provide convenient, cost-effective and comfortable alternatives to minimise the latter.
Increasingly, we should also be analysing car ownership on the same basis, recognising that living in some of our urban centres means not owning a car may be less voluntary and more necessary on price grounds.
If there is a sustained drop in accommodation costs relative to income, should we assume that inner cities will continue to boom?
Another relative exodus to the suburbs might fuel more growth in VKT especially if governments have favoured increased land supply to try to take heat from the housing market.
One of the problems with urban datasets is they tend to measure what happened, not why and are thus poor at predicting future behaviour. Furthermore, even surveys do not deal well with”if” and”when’, which will reveal what millennials do when they enter new life stages.
What can we learn from all of this?
Should we be pessimistic given all of these facts, figures and trends, and give up on urban development and transport strategies including smart growth, polycentricity and investment in high quality mass transit?
Should we also spurn innovation in the mobility market and regulate ride-hailing (and variants) out of our cities?
The answer is no.
But we must respect the inherent diversity of our cities and respond to the diversity of needs with regard for mobility and access.
We must not rest, thinking we have conquered car dependence or even car preference.
Focus must become tighter on public transport ubiquity, convenience, affordability and comfort if we want liveable, equitable and affordable cities.
Various peak car practitioners are encouraged by some evidence of changing preferences and choices among some of the population and particular to their life circumstances.
They are positive, also regarding some of the relative increases in funding for public and active transport infrastructure, and adoption of some smarter growth policies.
These things represent progress, not perfection with respect for our cities and the transport systems that support them.
These require more effective governance frameworks, flexible funding mechanisms tied more closely to pay-as-you-go revenue streams and robust investment decision-making based on more holistic appraisal frameworks.
For example, the research community has developed tools such as LUPTAI, SNAMUTS and Arup’s Accessibility Toolkit to gauge spatial disparities and evaluate the impacts of new infrastructure projects, while the Urban Liveability Index under development at RMIT is intended to provide intra-city and longitudinal evidence regarding spatial disparities in accessibility, among other variables.
These should be used to supplement more traditional transport modelling approaches to project evaluation.
Various urban growth and transport policies have failed to address a range of issues satisfactorily, including congestion, geographies of affluence and poverty, and driving by necessity rather than choice.
Limited investments in “blue ribbon” infrastructure, showcase revitalisation projects, and the land use uplifts that these may yield, tend to favour a relatively limited slice of the urban population, and are not serious approaches to deal with historic and emerging problems.
The Cheonggyecheon revitalisation project in Seoul, South Korea is cited often as evidence of success with respect for urban renewal (in this case, regeneration of an urban stream and supportive active transport infrastructure following removal of an elevated arterial road).
Still, for all its merits, the economic uplift the project has contributed to is cause for concerns regarding potential socio-economic exclusion.
Urban space needs to be utilised more flexibly
This means cleverer zoning, investment incentivisation and building practices, something Tim Williams discusses in one of his articles in The Fifth Estate in the context of Sydney.
Furthermore, the political clout of NIMBY lobby groups needs tempering because the perceived importance of low density “character areas” cannot trump the need for cities to feature more diverse housing product and support sustained mode switch away from cars, because of the increased viability of alternatives as urban densities increase.
In the US, various state and local governments are revising zoning laws to facilitate development of more diverse housing stock in locations where previously, only single-lot product was permissible.
From a transport policy and strategy perspective, commitment to revised infrastructure appraisal and delivery practices needs to bridge election cycles and extend beyond blue ribbon projects to common practice.
This requires a rethink of what constitutes key performance indicators for projects, including criteria such as consumer savings and affordability (beyond just estimates of travel time savings for car drivers), improved mobility options for non-drivers and energy conservation.
Increasingly, price signals need to be used to moderate travel behaviour in peak hours via peak routes.
Pay-as-you-go, congested-based pricing should apply to light passenger travel whether it is by driving or being driven.
This all means commitment and thinking beyond assuming millennials will drive the fall of the private vehicle (aided and abetted by technology), and everyone wants to live in a city centre.
It does mean more relevant data are needed to help us understand changes in travel behaviour.
Luminaries like Jarrett Walker continue to call on industry to work harder to make sense of what is going on with our transport systems and why.
What might we anticipate tomorrow?
The oft-maligned private vehicle will remain an essential feature of the mobility landscape for the foreseeable future.
It remains convenient and relatively affordable for a lot of travel – whether a person drives or is driven.
Moreover, we should not overlook the opportunities private vehicles provide for people to participate socially and economically in their communities.
A shift worker shuffling kids on a low income is a long way from being convinced that car ownership is an option.
Still, fundamental changes are required and should be facilitated to policy, planning, pricing, business models and consumer behaviour for peak car and sustained reductions in VKT to be realised.
There are two trends that will place increasing pressure on our cities and their supporting systems, and present added challenges to peak car theory.
The first is the ongoing pressure for urban expansion, and associated consumer demands for different housing typologies.
These, in turn, will be influenced by macroeconomic trends, including relative wage rises, inflation, interest rates and overall consumer buying power.
Even increased densities within existing urban boundaries may lead to growth or at least, maintenance of VKT if new housing tends to be in outer and middle suburbs rather than in higher-density centres.
The quality of that higher density living will determine its attractiveness relative to other housing typologies given the amount of land available to metropolitan fringes in Australia.
Given farebox revenue often does not get close to covering costs of providing and operating public transport, added suburban growth patterns are likely to make providing cheap and convenient public transport services impossible, as that other macroeconomic trend of reducing taxation per head of population further constrains public investment in service provision.
The second trend is the increasing application of complex technology in the transport sector.
Already, technology-enabled ride-hailing is cause of concern regarding light vehicle VKT.
In the near future, Connected and Automated Vehicles (CAV), promise low-cost, demand-responsive hyper-mobility.
Automakers, technology firms, venture capitalists and investment banks are going all-in, globally in the deployment race.
The profit incentive will tend to favour increased vehicle use at the expense of walking, cycling and public transport use.
CAV are likely to redefine the cost, convenience and comfort equation, making consumers potentially more willing to spend longer travelling.
This may redefine peoples’ expectations regarding where they live relative to where they work, learn, shop and recreate.
Moreover, this can fit in nicely with the sharing economy – people do not have to own their own CAV if a shared unit is only a smartphone thumbprint scan away.
Peak 2.0 might be a false summit.
Ryan Falconer is an independent researcher on cities and transport. He holds a PhD in Sustainability in Technology Policy and formerly led Arup’s cities business in Western Australia. Terry Lee-Williams leads Arup’s transport planning and strategy teams in Australasia, and has a substantial history in transport-related technology, major projects, policy, governance and delivery.
All views expressed in this article are the authors’ own and do not necessarily represent the views of their employers.