The UK Department for Transport (DfT) recently published its latest Road Traffic Forecasts. The aim is to inform the Department’s strategy, now centred on a five-year Road Investment Strategy that commits £15bn of capital expenditure the inter-urban road network.
The road traffic forecasts are generated by the National Transport Model (NTM), a multi-modal model constructed and operated by the DfT. The latest forecasts are in part a response to criticism that earlier projections of significant traffic growth had failed to take account of the largely flat trend over the last decade – known for short as ‘Peak Car’.
Drivers of road travel demand
To inform the latest round of modelling, the DfT carried out an analysis, now published as a report ‘Understanding the drivers of road travel’, that reviews the current trends and factors behind road use:
Trends: It is recognised that average distance travelled by car has fallen over the past decade, largely due to people taking fewer trips by car, despite growth in car ownership. This decline in per capita distance has been largely offset by population growth. While average car use has fallen, this is largely due to less car use by young men and in urban areas, while women and older age groups have travelled more.
Factors: Traditional factors such as costs continue to be important, including – for young people – the cost of learning to drive and insurance, to be afforded out of lower incomes. But there are new factors, in particular decline in company car use due to changes in taxation rules, more urban living, and congestion that limits car use in urban areas. The DfT conclusion is that after a decade of flat traffic levels, the outlook is for aggregate traffic to start growing again, but at a slower rate than historically as the link between incomes and traffic weakens, with population growth (16% to 2037) now the main driver.
The DfT commissioned a supporting study from the consultants RAND Europe – a ‘rapid evidence assessment’ to understand the reasons for the levelling off of car use in Britain. This selective review covered 21 source papers and addressed the amount of evidence available, which tends to be greater for traditional factors and less for newly emerging factors. So for example, my own analysis that suggested a saturation of demand for travel is viewed sceptically on account of as yet limited evidence.
A general conclusion of the RAND study is that models that rely on aggregate past trends to predict future car travel will not be good enough, given the increasing diversity of the car travel market; it is therefore important that travel demand models incorporate adequate segmentation to ensure that travel behaviour of the different market segments is well represented. The trouble with this very reasonable conclusion, however, is that more segmentation would make an already complex National Transport Model even more complex, raising questions about how to assess its validity.
Road traffic forecasts
The new road traffic forecasts use scenarios for the first time to explore a range of possible influences, a welcome development in the light of the considerable uncertainty about future demand. There are five scenarios, which reflects both traditional variables like oil prices and GDP growth, and also new possibilities: a zero relationship between income and car travel; and the continuation of a recent downward trend in trip rate. (My view is that the zero relationship is a very reasonable basis for a credible scenario, but that the trip rate trend extrapolated to 2040 is rather extreme.)
Let me focus on Scenario 2 in which income no longer drives car ownership or use. Average number of trips and travel distance per person are both forecast to remain broadly constant. Total trips and distance travelled by all modes are projected to increase on account of population growth. However, car traffic in England is projected to increase by 27% by 2040, which is significantly faster than population growth for reasons that are not made clear but include assumptions about declining real fuel costs due to improved fuel efficiency, supposed to effect a shift to car from public transport.
Traffic congestion is forecast to increase for Scenario 2, from about 7% of all traffic experiencing congested conditions at present to 13% in 2040, with average traffic speeds falling from 32.1mph at present to 30.1mph.
Demographic and spatial aspects
Given the weakening of the relationship between income and car use (even to zero as in Scenario 2), demographic determinants have become of central importance, population growth in particular. The breakdown between age groups, gender, employment and types of household are all relevant and are taken into account in the DfT’s National Trip End Model (NTEM), an important component of the NTM.
However, it is not possible to infer the broad distribution between greenfield and urban sites for housing the growing population (another 10m by 2040) from the very detailed NTEM. This distinction is important for forecasting the growth of road traffic since those living in new houses on greenfield sites will use cars, while those housed at higher density in urban areas will make greater use of public transport, as is evident from London.
Previous DfT road traffic forecasts have projected substantial increases for London, even though car traffic has fallen somewhat over the past twenty years – most likely due to road capacity constraints, arising from both decisions in the 1970s not to enlarge the road network to meet the needs of growing car traffic and more recent decisions to allocated more space to bus and cycle lanes and pedestrians. While this past shortcoming of the NTM as regards London traffic has been recognised, the latest forecasts nevertheless project car traffic increase in London by 2040 of 21% for Scenario 2., which is not consistent with the plans of the Mayor and Transport for London (TfL) to improve the road and street network but not to increase its capacity.
TfL recently published a study of trends in travel demand in London, where there has been a substantial mode shift from private to public transport since 2000, with car travel falling by around 15% from its 1999 peak even while the population has been growing. We need data from other cities to see if there are similar developments elsewhere, reflecting the growth of business services located in city centres driven by agglomeration economics.
The road capacity constraints in London and other urban areas are insufficiently recognised in the NTM. Moreover, such constraints are likely to impact on car ownership and car use on interurban roads. Compared with the South East of England, twice as many London households do not own a car, and Londoners travel half the annual distance by car.
The inclusion of a number of scenarios is a welcome innovation for the DfT’s regular road traffic forecasting exercise. Scenario 2, in which income is assumed uncoupled from car use, is consistent with the findings of the National Travel Survey that average distance travelled has changed little over the past twenty years, during which time incomes continued to grow. Nevertheless, on this scenario, car traffic is projected to grow faster than population growth, implying some shift to car from other modes.
In contrast, I would expect car traffic to grow at a slower rate than population growth, since some (perhaps most) of this growth will be accommodated in urban areas where the scope for car use is limited, as is well documented for London. The NTM seems insufficiently to recognise road capacity constraints in cities.
Moreover, the NTM does not recognise travel time constraints, as reflected in the finding of the National Travel Survey that average travel time has changed very little over 40 years. Indeed, the NTM projects average travel time increasing from 68 minutes a day in 2010 to 72 minutes in 2040, compared with the 60 minutes observed in the NTS in 2013. Time constraints are important since they limit the build up of congestion – which results in lower speeds and consequent behavioural changes by road users that serve to fit necessary travel into the unchanging (on average) amount of time available within the fixed 24 hours of the day.
More generally, although those responsible for the NTM have recognised the validity of past critiques and have responded to some degree, more remains to be done. We need more evidence about changes in attitudes and behaviours that underlie the Peak Car phenomenon (a term studiously avoided in recent DfT documents), and further scenarios that reflect emerging evidence. The NTM needs to be updated – it is still calibrated against 2003 traffic data – and to incorporate travel time constraints.
The NTM should be rebuilt to allow others to use what is at present a private DfT model. This would allow the structure of the model to be understood more widely and refined in the light of external analysis, thereby increasing confidence in the forecasts. It is the norm for Government macro models to be made available to others. The Treasury’s model of the UK economy has been used externally for 25 years.
The NTM is projecting car traffic growth at a substantially faster rate than I would expect, based on the National Travel Survey time series data. The Road Traffic Forecasts 2015 arguably contribute to both an overstated case for planned road investment, and overstated difficulty in tackling transport sector greenhouse gas emissions.
10 April 2015