Gordon Stokes Transport  

Peak car - the future?

Transport Home Photography Home Transport Photos

Last updated - Jan 2013

Pages in this series

1 - The idea of Peak Car

2 - Graphic analyses of trends

3 - More analyses of changes

4 - Attempt to forecast future car use

I wrote this material on peak car in 2012-13. Although the data is somewhat out of date now, I think the basic principles still apply.

What does the future hold?

What this means for the future is difficult to predict. Forecasters tend to use past trends at least as an indicator of the future and to look for relationships that are constant to show how if one things changes, the thing you're trying to forecast may do too. In the past it was noted how closely the amount of car travel followed economic growth so this formed a major element of forecasts.

But since the early 1990s it has been noted that GDP and traffic levels have 'de-coupled' - economic growth as measured by GDP continued to grow but transport activity grew less fast, and since the mid 1990s (as noted above) the amount of car use per person remained roughly constant, until it fell in 2008. While official forecasters have related recent events to their models and imply that the relationships still stand it is probably worth looking at the future from a different angle.

What is put forward here (partially, but hopefully soon complete) is a model that takes the status quo (as of around 2008) as being the 'base', and predict what will happen as the population ages, and assuming that drivers drive about the same amount as they do now. It also assumes that younger people will take up car access at the same rate as they have been recently observed to - equal for men and women and slower than in the 1990s. While this is a 'base' forecast it is possible to change parameters (or tweak it) to allow for different possible changes - for example, higher mileage related to a return to economic growth with lower fuel costs, or further growth in alternative modes of travel, and higher costs of driving.

It is a simple model in its conception and is intended to be illustrative of various scenarios rather than a solid forecast to challenge official forecasts.

Forecasting population

Population forecasts are taken from the Office of National Statistics 2010 forecasts which provide a forecast for each year of age for males and females, for every year up to 2036. Rather than use the 'central' forecast (which is their best bet on what will happen taking immigration and other issues into account) the 'natural' population growth ones are used. This is done so that factors such as immigration levels can be taken into account within scenarios, rather than starting with a base that assumes a fairly high degree of immigration.

Forecasting access to a car

Data on access to cars is taken from the National Travel Survey using the variable "i203 Driver access to a vehicle". People are grouped into 3 groups:

  • "Main drivers" - people who are recorded as the main driver of a household or company owned vehicle. This is the person who chirped up "me" when asked who was the main driver of a vehicle used by the household
  • "Other driver" - People with a car licence who are not the main driver of a household vehicle
  • "Non drivers" - people who are under 17, and others without a licence

For the base model these figures were derived from an average for the years 2002-10.

The 'base' forecast is made on the assumption that inertia means that once people have car access they do not give it up, until age and ill health force them to relinquish it. For younger people, they are assumed to gain car access at the same rate as people in 2008-10, up to the age of 44. The model goes through each year from 2010 up to 2036 using formulae for gaining access during 'younger age', retaining last year's through middle to 'older age', and then a formula means that a certain percentage lose car access from the age of 74 (accelerating towards age 90). This formula attempts to mirror the observed loss in car access for similar age cohorts shown in the analysis section.

The following figures show how this pans out for 2012 and 2033 for men and women. The dark parts of the charts are for 'main drivers' while the lighter parts show 'other drivers'. The rather outdated 'blue for a boy' and 'pink for a girl' are used.





These profiles can be altered, and are, within different scenarios. In particular some scenarios will assume that women 'catch up with men' in terms of reaching the same saturation level. Others will assume that the rate to saturation is faster or slower. And in some cases it will be assumed that some people forego access to a car, maybe due to lack of a need for one.

Forecasting kilometres per driver

The base scenario assumes that annual kms per driver remains the same as that from 2008-10. This is in line with what was observed in Figures 5 and 6 on the graphic analyses pages.

What it predicts

Running the model from 2010 to 2036 shows that the overall amount of car kilometres (black line) will rise slowly until about 2027, and then start falling. In 2027 it would be 3.5% above the current level, and by 2036 it would have fallen to about 3.1% above the current level. The pink line for women shows more growth than that for men. By 2036 the population would be 5.8% higher (using the 'natural' growth only forecast). So car distance per person would have fallen slightly, having risen very slightly up to 2020. The scale on the graph is in million kms per year for the solid lines showing traffic. The dashed line shows kilometres driven per adult.



Shortcomings of the 'base' scenario

Any base scenario for a prediction 20 years hence is not likely to be perfect. In this case, rather unusually, the assumption is for no change other than demographics and aging - the propensity to use a car remains static, as does the distance per driver. Conventional economic forecasters will probably be horrified by this approach, but it does allow greater transparency in determining other scenarios than when change is part of the basic assumption.

It is also just as plausible, given the trends of the last 15 or more years, that the use of the car will become less or more, or stay the same. While forecasts from the 1970s were done at a time when car access and mileage was rising, it is just as sensible to assume that little will change.

But things will change:

  • It seems unlikely that men will continue to drive so much more than women, though it is as likely that they would converge towards somewhere in between the current levels, as that they would to rise to male levels or fall to female levels
  • The economy will change - but will the car retain its earlier place as the major item of expenditure after housing and food, or have other items taken its place, and have public transport and cycling opportunities (in towns and cities) improved to a level where fewer people want a car?
  • Prices will change. Currently it looks as though fuel prices will rise and maybe quite dramatically at times, but electric vehicles (when they are affordable to more) may make fuel prices much lower

Scenarios based around these will be explored in the next page (when I write it).

Source of data used as base for model - National Travel Survey from the UK Data Archive, with additional variables provided by the DfT National Travel Survey Team. The research for this has been done in relation to my Visiting Research Associateship with Transport Studies Unit in the University of Oxford. I am writing this up as a TSU Working Paper which I hope will be published in the spring of 2013.

Back to Transport Index

Gordon Stokes, 2013