I gave a talk on this topic at a recent meeting of the Transport Economists Group, one theme of which was the subject of the previous article. My overall conclusions are set out below.

A number of new trends emerged in the 1990s, or in some cases are still emerging:

• There has been no growth in average distance travelled in Britain for more than 20 years, whether by all surface modes or by car alone. Available data suggests that this holds for the developed economies generally. This contrasts with the previous century and more during which average distance travelled increased steadily.
• One reason for this cessation of growth of distance travelled lies in technological constraints on faster travel. We cannot drive faster on the roads, safely and with acceptable emissions. We have high speed rail to come, but rail is responsible for a minority of trips, and high speed rail for a minority of a minority. These technological constraints mark the end of an era that began in 1830 with the first passenger railway, which harnessed the energy of fossil fuel to permit travel at faster than walking pace.
• Car-based mobility or access to good public transport allow high levels of choice of many regularly used types of destination, thus lessening the need to travel further.
• Travel demand per capita has ceased to be driven by growing incomes. Total travel demand is now determined largely by population growth. However, the pattern of such demand will depend on where the additional inhabitants are housed – if on greenfield sites, more car use; if within existing urban areas at higher density, then more public transport.
• Car use in big cities has passed its peak, in term of mode share, and is now declining. Successful cities attract people to work, study and live, so population density increases. The city authorities recognise that the road network cannot be enlarged to accommodate increasing car use, so investment in urban rail is needed to meet the mobility needs of the population.

So travel in the twenty-first century will be different from travel in the twentieth century, quite apart from the impact of technological developments such as driverless cars.

In contrast to these new trends, one unchanging feature is average travel time of about an hour a day, found for all settled human populations. This constitutes a sound basis for forecasts or scenarios of future travel. Travel/transport models should be constrained to hold average travel time constant in the long run. They also need to recognise the new trends outlined above as regards both model structure and calibration. Most existing models are obsolete.

2015 marks the 50^th anniversary of the first data collection of the British National Travel Survey (NTS). The Transport Statistics Users Group and the Department for Transport held a meeting on 23 September at which I was one of the speakers. My presentation is available Metz tsug 23-9-15 . This article outlines what I said about daily travel. A future article will deal with air travel.

Slide 2 of my presentation shows the main NTS time series, continuous over 40 years (the first data collection 50 years ago was a one off). This covers all modes except international air travel and is derived from 7-day travel diaries, hence covers daily travel. Journey frequency has remained broadly at about 1000 trips a year on average over the period, and average travel time has held steady at around 370 hours a year, or an hour a day. What has changed is the average distance travelled, which rose from 4500 miles in the early 1970s to reach 7000 miles in the mid 1990s, with no further growth in the past 20 years.

The growth in distance travelled, in the unchanged amount of travel time, is the result of investment in the transport system – private investment in more and better cars, and public investment in infrastructure. People have chosen to take advantage of faster travel to reach more distant destinations, not to save time going to previous destinations, to take advantage of increased opportunities and choices. For instance, by travelling faster on the journey to work, you have more choice of jobs accessible from where you live in the time you allow yourself for the commute, and more choice of homes accessible from your workplace; and similarly fast travel allows more choice of shops, schools and so forth.

Benefits of transport investment

Transport economists suppose that the main economic benefit of transport investments takes the form of journey time savings, which are valued because this permits more productive work to be carried out or more valued leisure to be enjoyed. However, the constant average travel time found in the NTS is evidence that there are no such time savings in the long run. So time savings must be short run, succeeded by changed travel behaviour as people choose more distant destinations. This means that we are appraising the value of long lived infrastructure investment based on short run time savings, which is not sensible.

We should ask what is the long run benefit from transport investment. This can be seen in East London, Docklands and beyond, which is being regenerated through rail investment – Docklands Light Railway (Slide 3), Jubilee Line Extension, Overground, with Crossrail under construction. Public investment in the rail system makes land accessible for development by private sector developers who construct residential and commercial property, which accommodates London’s growing population, both homes and employment. This is both the strategic narrative and the causal mechanism whereby transport investment facilitates economic growth. It is not sensible to value these rail investments on the basis of notional short run time savings. The real value is seen, for instance, in the high rents paid for office accommodation at Canary Wharf, reflecting both the access made possible by rail investment and the high value of the economic activity carried out there.

Strategic Road Network

Constant average travel time also provides an illuminating perspective on investment in the Strategic Road Network, shown in Slide 4 where the colour coding indicates the degree of congestion, from black through red and amber to green. In general, congestion occurs near to population centres, where local traffic impedes long distance users. On the M25 around London, for example, about half the traffic is long distance and half is local, importantly people going to and from work, which generates morning and evening peak congestion. Such congestion is seen by the Department for Transport and its roads agency, now known as Highways England, as a reason to increase capacity by adding carriageway. Evaluations of completed major schemes of improvement on the Strategic Road Network show average time savings at peak hours to be about 3 minutes per vehicle. Multiplying a few minutes by a large number of vehicles and by the standard monetary values of time yield economic benefits sufficient to justify the large investment in expensive civil engineering works. This is an example of justifying investment in long lived infrastructure on the basis of short run time savings.

But what happens in the long run? Faster travel resulting from widening the road is to the advantage of local users who can exercise more choice of jobs and homes, when they come to change these, by travelling further in the same amount of time. This generates extra traffic – known as ‘induced traffic’ – which restores congestion to its former level, so that long distance road users are no better off. This is the basis for the maxim ‘You can’t build your way out of congestion’, something that UK Transport ministers used to say when they didn’t have a big budget for roads investment.

What investment in the road network achieves is improved access that facilitates local development. This may be inadvertent as a consequence of misconceived efforts to tackle congestion. Or it may be intentional in support of local plans agreed between developers and planning authorities. However, it is difficult to devise a national road investment strategy on the basis of such local needs.

If we can’t build our way out of congestion, how can we tackle this challenge? Surveys of road users indicate that the main problem with congestion is the uncertainty of journey time. To deal with this we need to provide road users with good predictive journey time information before they set out, so they can plan their trips to arrive on time. This is increasingly possible as the digital technologies get faster, cheaper and more powerful. Digital technologies are very likely to be far more cost-effective than civil engineering in improving the performance of the road network.

Peak or Plateau

Returning to Slide 2, we see no growth over the past 20 years. About three-quarters of the 7000 or so miles travelled each year on average is by car, so it is no surprise that car use per capita has also changed little over this period (Slide 5). This is found not just for Britain but for the developed countries generally, and is known as ‘Peak Car’, although ‘Plateau Car’ might be a better term since, while it is clear that cessation of growth of car use started well before the recession, it is not certain that long term decline has set in.

Since per capita travel by all modes has stabilised, future total travel demand will be driven mainly by population growth, which in Britain is significant. But the pattern of demand growth will depend on where the additional inhabitants will live. If they are to be housed on greenfield sites, they would use cars and the road system would need to be enlarged. If, however, they are housed in existing urban areas where the scope for additional road capacity is limited, then public transport investment would be important.

Peak Car in London

London is an example of a city with a rapidly growing population with no greenfield sites suitable for house building, so that the brownfield sites and infill are being developed, and people are living at higher densities in existing properties. Slide 6 shows how London’s population grew until 1940, followed by a 50 year period of decline as people left a damaged, overcrowded city to seek a better quality of life in new towns, garden cities and the like. But around 1990 the tide turned as people saw again the attraction of city living, with numbers now projected to reach 11.3m by 2050.

While London’s population has been growing for the past 25 years, car traffic and car use has not (Slide 7). This is because of road capacity constraints. Having begun to build elevated motorways in the 1960s to accommodate growth of cars, this approach was abandoned in the face of popular resistance to the damage created in the urban environment. The historic street network has been retained, but with reduced capacity for vehicles on account of bus and cycle lanes, more pedestrian space, as well as parking controls in the inner boroughs and the congestion charging zone in the centre.

If the population of London is growing but car use is not, it follows that the share of journeys by car must be falling, as seen in Slide 8. Car use (driver and passenger) has fallen from 50 per cent of all trips in 1993 to 37 per cent currently. Walking plus cycling has changed little (cycling is growing, but from a low base), while bus and rail have increased, the result of substantial investment.

In Slide 9 the car share of trips in London is extrapolated to cover the century 1950-2050. Data prior to 1993 are based on the assumption that car use in London grew at the same rate as car ownership nationally, which we know from vehicle registrations. Future data is based on the assumption that London’s road capacity will not be increased but that rail investment will continue, in line with the policies of the Mayor and Transport for London. On this basis, the share of journeys by car will decline to about 27 per cent by 2050, despite which (or perhaps because of which) London is very likely to continue to thrive economically, culturally and socially.

The peak seen in Slide 9 I designate as ‘Peak Car in the Big City”. This shift away from car use in successful cities, as exemplified by London, together with the cessation of car use per capita nationally, are helpful in mitigating transport greenhouse gas emissions, as I previously described. The peak marks a transition: from an era in which the growth of travel demand was driven by increasing income to an era in which it is driven by population growth; and from the twentieth century in which increasing prosperity was associated with increasing car ownership to the twenty-first century in which increasing prosperity is associated with decreasing car use in successful cities (and static car use per capita beyond).

It is hardly possible to forecast a peak of the kind seen in Slide 9. A transport planner in London in the 1980s, thinking about the future, would see a 40 year trend in population decline, a 40 year trend in growth of car use, and would extrapolate both in to the future, as would a transport model. What the planner and the model would fail to foresee is the change in the economic structure of London and changed attitudes to urban living and car use. Models assume continuity between past and future, relying on historic relationships (elasticities and the like) to predict the impact of change in exogenous factors such as income and population growth, oil prices, and technological developments. However, the peak phenomenon implies a break in continuity and the need for forward-looking relationships. This is difficult for modellers to cope with since the bulk of the data they have for model calibration is historic.

The peak of car use in London occurred in 1990, 25 years ago, yet the modellers at the Department for Transport responsible for the National Transport Model still do not recognise this – they are in denial about Peak Car. In consequence this Model projects big increases in car traffic in London – depending on the scenario of up to 37 per cent by 2040 in the most recent Road Traffic Forecasts. This is hard to reconcile with the historic trend (Slide 7) and the plans of the Mayor, which do not involve increasing road capacity. The Model is not consistent with observed behaviour in London, nor with the evidence we have for other big cities.

Assessment

To think about the future, we would best focus on the invariant average travel time (Slides 2 and 10). The hour a day has hardly changed in 40 years, and probably for a lot longer, and is found in all other countries. The scope for increase is limited by the 24 hours of the day and all the other activities we need to undertake; and for decrease by the activities we need to reach beyond the home, as well as by the need to get out and about regardless of destination. So it would be sensible to constrain all transport models to hold average travel time constant in the long run. This would change model outputs when used to analyse the impact of a proposed investment, comparing the ‘do something’ with the ‘do minimum’ cases. Rather than time savings, the output would be changed access, land use and land value, which would be appropriate when the purpose of the investment is to stimulate economic growth – comprising population growth and productivity growth, both of which involve land use change.

My conclusions as regards daily travel are summarised in Slide 20. The NTS provides clear evidence that there are no travel time savings in the long run, and that there has been no increase in per capita travel for 20 years, the basis of the Peak Car phenomenon. The current theoretical frameworks for investment appraisal and forecasting are inconsistent with this evidence and hence need to be rethought. Using an economic framework inconsistent with the evidence results in sub-optimal investment decisions.

I see three problem areas. In my view we are investing too little in urban rail because we do not recognise the economic benefits associated with changed land use. Urban rail allows successful cities to grow to higher densities, which results in agglomeration benefits – economic, cultural and social – and which also mitigate greenhouse gas emissions. We are investing too much in civil engineering works on inter-urban roads in the futile hope of reducing congestion, although there is scope for supporting development where planners and developers wish to develop sites that require improved road access. And we are investing too little in the digital technologies that are likely to be far more cost-effective than the expensive civil engineering technologies in dealing with congestion. What we need here is to put monetary values on journey time reliability and to understand the response of road users to predictive information about journey time, as a basis for investment appraisal of digital approaches.

The UK Government Office for Science is running a Foresight project on the Future of Cities. I was asked to contribute an essay, the key points of which are:

• Successful cities are characterised by growing populations as people are attracted to work, study and live. As population density increases, agglomeration effects contribute to economic productivity and to similar cultural and social benefits.
• Increasing population density precludes enlarging the road network to accommodate growth of car-based mobility. Instead, rail systems must be expanded to provide fast and reliable travel for those who work in the city. So the share of journey by car declines.
• In London, car use peaked at 50 per cent of all trips in around 1990. It has now fallen to 37 per cent and should fall further to 27 per cent by mid-century, on the basis of current projections and policies. This will make a significant contribution to mitigating transport greenhouse gas emissions.

Plans are being developed for additional road crossings of the River Thames in East London. Downstream of Central London there are at present two road tunnels (Rotherhithe and Blackwall), a ferry at Woolwich, and a bridge/tunnel at Dartford on the M25 orbital motorway. These are subject to delays at peak times, and the limited possibilities for crossing the river are seen as an impediment to the growth of East and South East London.

However, building additional road capacity runs counter to the general direction of transport policy in London in recent years, which has been to invest in additional rail capacity but not in road capacity. The result has been a steady shift away from car use while London has thrived economically, culturally and socially. So are additional river crossings a sensible idea?

The case for additional river crossings has been made most forcibly in the report of a Commission on East Thames Crossings, set up by the Centre for London and chaired by Andrew Adonis, a former Secretary of State for Transport who is a visionary thinker. Four new crossings are proposed: a tunnel at Silvertown to relieve congestion on the Blackwall Tunnel; a crossing – preferably a tunnel – at Gallions Reach; a bridge further downstream at Belvedere-Rainham; and a further bridge to supplement the Dartford Crossing.

In the view of the Commission, the ‘case for action is now overwhelming. New crossings will improve access to jobs, customers and suppliers, increasing business productivity and employment. The increased accessibility will also provide a boost for house building, so helping to tackle London’s severe housing shortage. Crucially, new crossings will also connect otherwise somewhat isolated communities to the opportunities which are now beginning to spread across the area.’

Transport for London (TfL) is preparing plans for the Silvertown Tunnel with the intention to seek powers to build in 2016, and is developing the concepts of new bridges at Gallions Reach and Belevedere. The Department for Transport (DfT) is considering two options for a Lower Thames Crossing at or beyond Dartford.

Questions

The proposals for new river crossings raise two key question: whether the additional road traffic that will result is likely (a) to be sufficient to foster a worthwhile amount of development, and (b) so big as to cause significant additional congestion on the road network. TfL has issued a report on the traffic impact of new crossings, based on a demand model. The assumption is that use of the new crossings will be charged, with tolls set at the rate for the Dartford crossing (£2.50 for a car at peak times). It is also assumed that the Blackwall Tunnel will be charged, to prevent diversion from the adjacent Silvertown Tunnel, although that would result in diversion of some traffic to non-charged crossings upstream.

The findings of the traffic modelling are quite complex but generally indicate reduced congestion while overall flows across all crossings are little changed. This is because the model does not allow for any land use changes that could occur as a result of changes in travel accessibility. However, the hoped for development as a result of the new crossings would depend on land use changes – new homes and new places of employment. So the modelling underestimates the likely growth of traffic and provides limited insight into the consequences for congestion.

We know from the National Travel Survey that average travel time remains unchanged in the long run at close to an hour a day. This means that investments that increase speed of travel result in people making longer trips to access more opportunities and choices, which in turn results in changed land use and enhanced land and property values reflecting the greater access. So we may expect that if new crossings are built, they will fill with traffic to the point where congestion inhibits further growth – the basis of the maxim that you can’t build your way out of congestion. The benefit will be seen as development of land and property, which is what is desired, but congestion would not be relieved.

How much development may we expect from additional river crossings? TfL’s assumption is that these will have one general traffic lane and one HGV/bus lane in each direction. So the scope for growth of car-based commuting seems limited. A further constraining factor is the tolls assumed both to help finance the new crossings and to manage demand. At £2.50 per car per traverse, this adds £25 to a regular weekly commute, on top of the usual running costs (average for all households £34 pw), which could be enough to deter most drivers who are not accustomed to paying tolls for travelling to work.

A study commissioned by TfL from the consultants Atkins considered the scale and distribution of the economic benefits of additional river crossings. The study identified property market areas in East London and how these might benefit from additional connectivity by car. The conclusion is that the northern side of the river has over twice as much floor space capacity that could support employment than on the south side, with the majority of this difference in the office sector. This potential imbalance in employment growth, combined with a relatively even distribution of potential housing growth, would lead to a greater demand for trips from those on the south side of the river commuting to the north, reinforcing the need for new river crossings.

Atkins estimate that in East London there is potential capacity for over 243,000 residential units, 2.5 million sq.m of office, 440,000 sq.m of retail and 1 million sq.m of leisure floor space. The biggest net improvement in connectivity, of the options considered, is from Silvertown + Gallions Bridge, which would result in a gross impact of around 20,000 additional residential units and 400,000 sq.m of commercial floor space. This increase is modest, the consequence of increases in access of less than 10 per cent across the wider area, and prompts the question of whether building new river crossings represents the best value for money in improving connectivity and stimulating development. The alternative would be more radial rail routes to central London, consistent with Atkins’ observation that while demand for housing in East London is strong, only sites with good links to the employment centres of central London are coming forward for development.

Width of the river

There is an inherent problem with river crossings in East London – the width of the river. If this were as narrow as in west London, then many more bridges would be possible (subject to repealing the navigation rights of tall ships, which could likely be justified). If the Thames were twice as wide as it is, then intermediate crossings at Gallions and Belvedere would be ruled out on cost grounds. With the width of the river as is, a couple of additional crossings between Blackwall and Dartford are feasible, but even with these additions, there would still be substantial constraints on cross-river traffic.

Perhaps the best use of new road crossings would be for buses , but to carry useful numbers of commuters over the distances to access employment opportunities, a Bus Rapid Transit system would be needed, with dedicated lanes protected from other traffic and fewer stops than on normal bus routes.

Assessment

The broad approach developed in London in recent years has involved investing in rail but not increasing road capacity – an approach that has contributed to the economic success of the city. Professionals and business people are willing to use rail for work journeys since this is faster and more reliable than car travel on congested roads. This helps free road space for commercial and public service vehicles. So the proposal to build additional road crossing across the Thames seems retrograde, particularly when new rail crossings have been and are under construction.

There is an opportunity cost for new river crossings – what could be done by spending the money in another way. Arguably, new and improved radial rail routes into East and South-East London would be likely to stimulate more development, both housing and commercial, taking advantage of lower land costs beyond central London.

Of the proposed new crossings, the two most likely to go ahead – Silvertown and Dartford – are seen as strategic routes where investment is justified by the present high levels of traffic and expected population growth. A further crossing at Dartford would be funded from DfT’s enlarged road construction programme. It is the two crossings in between, mainly for local traffic, that are most questionable – and therefore not evidently a high priority.