The Office of Rail and Road (ORR) is responsible for overseeing the performance of Highways England (HE), a publicly owned company responsible for England’s strategic road network. ORR is consulting on how it should perform its role. I have responded as below:

HE is responsible for a substantial programme of investment in new and improved road infrastructure, each element of which is supported by cost-benefit analysis consistent with the Department for Transport’s Transport Analysis Guidance. The main economic benefit is assumed to be the value of the time saved as a result of investments which increase capacity and are intended to reduce road traffic congestion.

However, there are questions about the estimation of prospective travel time savings derived from the standard models used for traffic forecasts. For example, monitoring of the outcome of widening of the M25 between junctions 23 and 27 concluded that ‘increases in capacity have been achieved, moving more goods, people and services, while maintaining journey times at pre-scheme levels and slightly improving reliability.’[1] No travel time savings were observed beyond the first year after opening, in part at least due to increased traffic, notably an increase of 23% at weekends. These outturns were inconsistent with the forecasts of traffic volumes that were significantly less than observed, and with speeds that were projected to be higher with the road widening than without.[2] The higher speeds were the basis for estimates of travel time savings, leading to the DfT’s estimate of the Benefit-to-Cost ratio of 2.3, which justified the investment.

This example shows that there may be a substantial discrepancy between forecast and outturn traffic flows and speeds. That this is a general problem is indicated by the observed invariance of average travel time over the past 45 years, as found in the National Travel Survey.[3] This implies that the benefits of road investment have been taken, not as time savings, but as increased access to desired destinations, which results in more traffic. This additional traffic is known as ‘induced traffic’, the consequence of increasing capacity, which results in increased externalities related to vehicle-miles travelled, including congestion, carbon emissions, air pollutants, and death and injuries. While HE routinely monitors outcomes of schemes 5 years after opening, this may not be sufficiently long to observe the full extent of induced traffic.[4]

There is therefore reason to suppose that in general the outcome of road investment as experienced by users does not correspond to the rationale for the investment, which is principally to increase welfare and economic growth by reducing congestion and improving connectivity. This discrepancy should be of concern to the ORR.

[1] Smart Motorway All Lane Running M25 J23-27 Monitoring Third Year Report. Highways England. 2108.

[2] https://www.gov.uk/government/publications/vdm-used-to-estimate-traffic-volumes-and-travel-time-saved

[3] Table nts-0101-2018

[4] Sloman L, Hopkinson L and Taylor I (2017) The Impact of Road Projects in England, Report for Campaign to Protect Rural England

 

 

Transport for London has recently published its latest report on Travel in London. At 279 pages, this latest in an annual series is almost certainly the most detailed account of travel behaviour in any city in the world. All credit to TfL.

Table 2.3 shows trip-based mode share. Private transport (very largely car) was responsible for 48% of trips in 2000, declining to 37% in 2015, but thereafter stabilising. Public transport has been stable at 35-36% of trips since 2012, and walking at 24-25% since 2000. Cycling grew from 1.2% in 2000 to reach 2.5% 2018. So the declining trend of car use has ceased in recent years, but it may resume as new rail capacity is opened, particularly Crossrail (the Elizabeth Line). Nevertheless, the target reduction of private transport to 20% by 2041, a feature of the Mayor’s Transport Strategy, looks difficult to achieve.

Section 9.7 discusses the role of licenced taxis and private hire vehicles (PHVs), a topic of much current interest. Taxis (black cabs) have been in slight decline while PHVs have grown substantially in recent years, largely reflecting the entry of Uber into the market. A survey of PHV users in London found that the two main trip purposes were for a night out and to/from airports, but only 28% of PHV trips were for both outward and return legs. App-based PHV users were attracted by specific features: estimate of fare, time for driver to arrive, knowing details of car booked, and estimate of journey time. 30% of PHV users said they had not needed to buy, replace or own a car, which facilitates a shift from individual car ownership.

Assessment

While a long-term target for reduction in car use has merit in that it shapes shorter term decisions, no Mayor is likely to hold office for anything like the time to reach the 2041 target date. A shorter-term target would allow performance to be held to account. And while the recent experience of London is that a steady reduction in the share of trips by car is compatible with the economic, cultural and social success of the city, sustaining this in the longer term would depend on substantial investment in the rail system that provides a fast and reliable alternative to buses, cars and taxis on congested roads. The biggest challenge for TfL and the Mayor is to find means of financing this investment.

BMW and Daimler recently announced that they were withdrawing their joint car-sharing service from the North America and the UK, although it will continue in some European cities. This business, known as ShareNow, which was the successor to BMW’s DriveNow and Daimler’s Car2Go brands, offered app-based short term car rentals, with pick-up and return anywhere withing large urban areas. The reasons given for withdrawal were rising costs and insufficient customer interest. The rationale for entering this shared use market was in case this were to develop into a significant alternative to the private ownership and fleet markets.

The implication of the BMW/Daimler decisions is that shared use seems less promising than many had supposed, not least CoMoUK, the association for the promotion of shared vehicle use in Britain. They see car sharing as a way of providing socially inclusive, low emission mobility which helps break dependency on private car ownership. Pay-as-you-go cars offer affordable, occasional access to cars to benefit individuals. At the same time, they help policy makers meet targets for emissions reduction, improvements to air quality and encouraging use of sustainable modes. However, CoMoUK’s concept of car sharing does not extend to the chauffeur-driven version, Uber and similar, the existence of which is likely to be a reason for the lack of commercial success of ShareNow.

Many observers believe that shared vehicle use is the solution to traffic congestion: if  occupancy could be increased, fewer vehicles would be needed. However, in urban areas there is substantial suppressed demand for car travel, the consequence of the deterrent effect of prospective delays due to congestion. Measures to reduce congestion initially reduce delays, which make car trips more attractive to those previously deterred, generating more traffic. So the limited levels of vehicle sharing that seem likely are probably not going to make much difference to road traffic congestion.

Last month, James Dyson announced the abandonment of his electric car project, worth £2bn of planned investment involving 500 staff. One factor in this commercial decision was the difficulty of developing a solid state lithium ion battery, seen as the next step in the evolution of lithium ion batteries. Getting the battery technology right is crucial for achieving commercial advantage in the electric vehicle market. A new entrant needs to offer a significant improvement in performance if it to grow market share, exemplified by Tesla.

Another likely factor prompting the Dyson decision, though not mentioned in press coverage, is the expectation that the car of the future will have autonomous driving options as well as electric propulsion. Autonomy involves either prolonged costly development or buying in someone else’s technology – both involving considerable risk.

As I have argued previously, the benefits to users of the new auto technologies will be incremental, not transformative. Yet the new technologies will be transformative for the manufacturing industry. There is a risk that returns from incremental improvement in performance will be insufficient to reward the large investment in technology development. Dyson may have made a shrewd judgement in cancelling his EV project.

A number of new transport technologies have entered the market recently or are under development. Innovators have high hopes of benefits to users and returns to investors. In my new book, Driving Change: Travel in the Twenty-First Century, I assess the likely consequences for travel behaviour of four technologies that are making an impact on road transport: electric vehicles, digital platforms, digital navigation, and autonomous vehicles.

The impetus to introduce electric propulsion for road vehicles is to eliminate carbon emissions, which requires the electricity supply system to be decarbonised in parallel. Electric propulsion also eliminates noxious tailpipe emissions and so improves urban air quality. The sustainability of road transport has long been a concern, initially because of reliance on non-renewable oil, and more recently on account of oil’s contribution to global warming. Electrification goes a long way to mitigating this concern. Yet electric propulsion in place of the internal combustion engine does not significantly alter the behaviour of the car and driver.

Digital platforms in the transport sector make possible ride-hailing taxis such as Uber, dockless bicycles and a variety of innovative modes of demand-responsive travel. Platforms are more efficient in matching demand to supply than traditional means, hence their popularity. As well as matching user to vehicle, they enable sharing by users travelling in the same direction, the immediate benefit being lower fares.

Some observers see vehicle sharing as an important means to reduce road traffic congestion, allowing travel demand to be met with fewer vehicles of higher occupancy. This would indeed be the case if trip numbers, origins and destination were fixed. However, congestion arises in places where both population density and car ownership are high, such that the capacity of the road network is insufficient to accommodate all the trips that might be made. Some potential road users are deterred by the prospect of unacceptable time delays and so make other choices – of time of travel, mode or destination when there are options, or not to travel at all. Introduction of vehicle sharing would initially lessen traffic, but the resulting reduced delays would attract onto the network previously deterred drivers, so restoring congestion to what it had been. So sharing is not the solution to the problem of congestion.

Digital navigation devices are now generally familiar, based on GPS location, digital maps and route guidance algorithms that take account of congestion. They enhance the quality of the journey for individuals and have the potential to improve the efficiency of operation of the road network through collaboration between private sector providers of routing information and public road authorities. But the main benefit lies in the estimation of journey duration at the outset of a trip, given that the principal problem of congestion perceived by road users is the uncertainty of arrival time.

The development of driverless vehicles is attracting enormous effort, both by established car manufacturers and by the tech sector. Timing and extent of deployment of fully autonomous vehicles remain unclear, as is the likely impact on the road network. At present evidence is limited to small scale trials and modelling simulations, for which the extensive range of possible performance parameters and policies means that any conclusions are at best quite tentative.

Looking back at the main transport innovations of the modern era, we see that they involved a step increase in speed of travel, which allowed individuals to gain access to a wider range of destinations, services, opportunities and choices: the railway in the nineteenth century; the car in the twentieth; the bicycle in its heyday before the car became generally affordable; and motorised two-wheelers today in many low income countries. Step increases in virtual access to people and services have been made possible by electronic and digital innovations: the telegraph, telephone, radio, television, internet, broadband, smartphones, social media. All these innovative technologies have led to huge investments, justified by the returns from access benefits to users.

In contrast, the new transport innovations seem unlikely to increase significantly the speed of travel. They will therefore not be transformative as regards our travel behaviour, although they will transform the manufacturing process and the nature of the auto industry. The car of the future will be electrically propelled, will have extensive digital functionality, and will have autonomous options, yet is unlikely to progress faster through traffic than the car of today.

The new transport technologies are improving the quality of the car journey incrementally, expediting routing, providing assurance about time of arrival, relieving the driver of some tasks, and lessening urban air pollution. Users will take up these technology options if they perceive them to offer value for money. The auto industry will need to bear down on costs to sell such innovations, as it has always done.

Beyond the personal car, digital platforms facilitate the availability of a wider range of shared use vehicles and individual conveyances. These may complement regular public transport services and may contribute to reduced individual car ownership in cities by offering convenient alternatives. Driverless taxis may allow travel at lower cost than conventional taxis, which would increase demand. However, autonomous vehicles in themselves cannot be expected to reduce urban traffic congestion.

Altogether, the new transport technologies seem likely to have less impact on travel behaviour than many suppose.

 

This blog appeared in the Transport Times issue of 11 September 2019.

 

 

 

 

 

 

 

London has had a road user charge since 2003 – a charge for entering a central zone during working hours, known as the ‘congestion charge’. This has proved workable, publicly acceptable and generates useful revenue, although the impact on congestion was fairly short term, as I explained previously. However, the fixed daily charge for entry means that the cost to drivers is unrelated to distance travelled or to traffic levels.The congestion charge is now supplemented by a charge on older polluting vehicles, which will be extended beyond the present central zone to cover much of London in 2021.

A study of more sophisticated alternatives has recently been published by the Centre for London. The proposal is to charge per mile in areas of high demand and poor air quality. Rates would vary by vehicle class and emissions, local levels of congestion and air pollution, and the availability of public transport. The charging system would be linked to individuals, not to the vehicle as at present, and would be based on a smartphone app that offered journey options alternative to the private car. An attractive feature is a refund if a trip takes significantly longer than estimated.

Modelling of the impact of the proposed charging scheme was undertaken although not published. There is some ambiguity in the reported findings (p61 of the published report). On the one hand it is stated that ‘for the average driver making a 10 km journey, we expect this [the charge] to amount to in the region of £1.50 – the cost of a cup of coffee or a bus ticket – although journeys in the most congested and central parts of the city, using the most polluting vehicles, would be charged much more.’ But is it also stated that ‘Charging drivers on the most congested roads the equivalent of a cup of coffee or a bus ticket could reduce emissions and air pollution by up the a fifth.’ So not clear whether the charge on the most congested roads is about £1.50 or much more.

The problem with road user charging in a city like London, with many relatively high income and business drivers, is that a charge of £1.50 for a 10 km trip is likely to be far too little to have a significant impact on congestion. Charges that would make a useful impact would be tricky politically. A proposal to introduce a whole new technology for charging would prompt anxieties about higher charges.

In my view, the best approach would be to increase the present charge of £11.50 per day to something like £15, and then to offer a discount to say £10 if paid via a smartphone app. Once such a new payment method is established, there would be scope for varying the charge to reflect congestion levels (which should be acceptable if the charge were always less that the standard daily charge). This approach would be similar to the introduction of Oyster and contactless card payments in London, with daily caps on charges.

 

Smart Motorways, a flagship programme of Highways England, aims to relieve congestion by converting the hard shoulder into a running lane and by varying the speed limit to smooth traffic flow. To assess performance in practice, Highways England has been monitoring closely the section of the M25 between Junctions 23 and 27 since the widened road opened in 2014. Three annual reports have been published, detailing traffic flows, journey times and safety, and comparing outcomes ‘before’ construction and ‘after’ scheme opening. Traffic growth of 16% was observed at Year 3 compared with before opening, far higher than regional motorway growth over the same period, with increases in weekend traffic of up to 23%.

Such traffic growth seems a noteworthy example of induced traffic, the traffic that arises as a result of increased capacity and which tends to restore congestion to what it had been previously. I therefore made a Freedom of Information request to see the traffic forecasts and economic appraisal that were the basis of the investment decision.

The traffic forecasting report was based on a variable-demand multi-modal model for the M25 area, employing the SATURN suite of programmes, updated to take account of the most recent national datasets for trip ends and similar. Traffic forecasts were made for the assumed 2015 scheme opening year, the 2030 design year and the 2040 horizon year, for the morning and evening peak flows and the interpeak period, comparing the ‘do-minimum’ case, without the investment, and the ‘do-something’ case with it.

The time slices used for the forecast and the outturn monitoring are regrettably different, which limits comparisons at particular times of day. Comparisons may, however, be made of average daily traffic flows (ADT). For the section J23-24 clockwise, for instance, the forecast ADT increase, comparing the scheme with do minimum, was 13% in 2015 and 16% in 2030. The outturn monitoring found an increase of 13% at Year 3 after opening compared with before, in good agreement with forecast.

The economic appraisal report employs the DfT’s TUBA software to derive estimates of monetarised travel time savings and vehicle operating costs (VOC) from the traffic forecasts, comparing do-something and do-minimum cases. The main economic benefit is travel time savings to business users, worth £475m, because the scheme was expected to allow travel at higher average speeds than the do-minimum case. Time savings to non-business travellers (commuters and others) were very largely offset by increased VOC, given the assumed diversion from local roads onto the motorway generating longer trips. As an example of the origin of the time savings, the speed on J23-24 clockwise for the AM peak in 2015 was forecast to be 86 km/hr with the scheme, versus 76 km/hr without. The overall benefit to cost ratio (BCR) was 2.3, later adjusted upwards to 2.9.

However, this forecast increase in speed failed to materialise. There has been effectively no change on average for all days and time slices between before opening and Year 3 for the clockwise travel. For anticlockwise, an average saving of 15 seconds (1.4 per cent) was found for a journey of 16.6 min before the improvement. Time savings of 6% and 9% respectively were seen at Year 1 after opening, but were lost by Year 2.

Generally, traffic flowed at the free flow rate except during the PM peak when it was slower. Surprisingly, the extra lane did not permit a faster flow at this PM peak, even though the increase in capacity of 33% was greater than the increase in traffic volume. Possibly the use of variable speed limits to smooth the flow was at the cost of journey time savings.

The stated conclusion of the Year 3 monitoring report is that ‘increases in capacity have been achieved, moving more goods, people and services, while maintaining journey time at pre-scheme levels and slightly improving reliability.’ Yet this conclusion undermines the economic case for the scheme, based on forecast time savings. This in turn raises questions about both the validity of the modelling and of the orthodox approach to appraisal.

We know from the National Travel Survey that average travel time has remained essentially unchanged for at least the past 45 years. This implies that any travel time savings must be short run. In the long run, people take advantage of transport investments that permit faster travel to gain access to more distant destinations, services, opportunities and choices, within the limited time they allow themselves for travel. This change in travel patterns would first be seen in optional trips, consistent with the big increase in weekend traffic in the M25 example, and subsequently over the years as people move home and change jobs. The consequential additional traffic – induced traffic – adds to congestion and negates the time savings that are conventionally supposed to be the main economic benefit.

If we are to make transport investments that are good value for money, we need to pay attention to the real-world consequences, and not be misled by the outputs of models that generate the notional time savings to which transport economists are so attached. We need to constrain models to hold average travel time constant in the long run, consistent with the findings of the National Travel Survey. To calibrate models, we need data on origins, destinations and purposes of trips, and how these change when a road is widened. And we need to work out how to value access benefits to users.

 

This blog is based on an article published in Local Transport Today 24 May 2019

 

 

 

 

I recently participated in a conference on the Future of the Car, organised by the Financial Times, which prompted the following thoughts.

The car industry is in a state of flux as it engages with four new transport technologies. Electric vehicles will eliminate both carbon and noxious tailpipe emissions, and go a long way to achieving sustainable surface transport. Autonomous vehicles are expected to work wonders, but timing and impact in big cities is quite unclear. Digital navigation, now in general use, can optimise routing through congested traffic and provide estimated journey times, so reducing the uncertainty that is the most bothering aspect of congestion. Digital platforms, exemplified by Uber, more efficiently match demand to supply, and also facilitate sharing of journeys.

Vehicle sharing is seen by many as the answer to road traffic congestion. Certainly, if the numbers of trips were fixed, increased vehicle occupancy would mean fewer vehicles and less traffic. But trip numbers are not fixed. Congestion arises in areas of high population density and high car ownership. The capacity of the road network is insufficient to accommodate all the car trips that might be made. Some are supressed by the prospect of unacceptable time delays as car owners made other choices – of mode or time of travel, of destination where there are options (as for shopping), or not to travel at all (ordering goods online, for instance, or working at home). Interventions intended to reduce congestion, such as vehicle sharing, initially free up road space and reduce delays, but this attracts previously suppressed trips, restoring congestion to what it had been. This is also the basis for the maxim that you can’t build your way out of congestion, which we know from experience to be generally true.

The most important transport innovations of the past had a crucial characteristic. They achieved a step-change in the speed of travel, allowing us increased access to desired destinations, opportunities and choices, as we saw with the railways, the car, commercial aircraft, the bicycle in its heyday, and motorised two-wheelers today in low income countries. Step-changes in virtual access to services and people have been achieved by a succession of electronic innovations – the telegraph, telephone, radio, television, internet, broadband, mobile phone, social media. All these innovations have been transformative and have rewarded the huge investments that made them possible.

In contrast, the new transport innovations do not permit a step-change in the speed of travel. They will not increase our access and will not be transformative. The car of the future will be electrically propelled, will have extensive digital functionality as well as self-driving options. But the car of the future will not progress faster through urban traffic than the car of today.

What the new technologies offer are incremental improvements to the quality of the journey. Users will take up such options if they are perceived to offer value for money. The auto industry will need to drive down the costs of the innovative features, which of course is what the industry has always strived to do. The future of the car, and of the industry, will be more like the past than many seem to suppose.

A version of this blog appeared in the FT’s Alphaville blog.

 

 

I blogged recently about the review of the law governing autonomous vehicles (AVs) that is being carried out by the British Law Commission. One aspect not considered in the consultation document is the implications for vehicle connectivity. I have submitted a response to the consultation to make the following points.

Adaptive Cruise Control, available in some current production vehicles, adjusts the gap to the vehicle in front using sensor data to vary speed. Connected (or Cooperative) Adaptive Cruise Control (CACC), an emerging technology, takes data from vehicles further ahead using vehicle-to-vehicle (V2V) communications, which allows shorter headways and attenuates traffic disturbance to achieve smoother flow. The shorter the headway, the less scope for driver control. In the limit, CACC allows platooning of vehicles on inter-urban roads – ‘road trains’ involving vehicles in close proximity with a lead driver in control. The postulated benefits of platooning include better fuel economy and decreased emissions resulting from lower aerodynamic drag, improved traffic flow and capacity, and lower labour costs. A number of truck manufacturers have initiated trials and more are planned. The 2016 European Truck Platooning Challenge involved six manufacturers sending platoons from starting points in five different countries to end at Rotterdam, gaining valuable practical experience.

Much of the past impetus for the development of connected vehicles came from research funding provided by the US Department of Transportation. The main motive was to improve safety through vehicles exchanging data about speed and location, and vehicles acquiring data from roadside infrastructure about road conditions. There are two competing telecoms technologies under consideration: short-range WiFi for V2V being developed by the car manufacturers for platooning; and a system based on the next generation 5G mobile phone technology, to be installed in both vehicles and infrastructure, and able to signal at longer range.

Unlike truck drivers, the generality of drivers of AVs would be able to choose the gap to the vehicle in front. It is not clear why, without an incentive, they would choose a gap smaller than that with which they are comfortable, which may not be very different from current headways. Accordingly, to increase road capacity by reducing headway there would need to be some incentive that would impact on individual drivers. This might be a road user charging regime that charged on the basis of the length of carriageway effectively occupied.  Another kind of incentive to reduce headway would be dedicated lanes that are less congested and faster flowing than other lanes, analogous to High Occupancy Vehicle lanes on US highways. Acceptable incentives would be needed if manufacturers were to go beyond equipping vehicles with the existing Adaptive Cruise Control. If manufacturers are to be held be responsible for the safe functioning of AVs, adding V2V or vehicle-to-infrastructure connectivity (V2I) to reduce headway would exacerbate this responsibility by introducing functionality that depends on that of other manufacturers and suppliers, and that increases the risk of security breaches.

More generally, connected vehicles operating at short headways would require reconsideration of the safety regime concerned with the crashworthiness of individual vehicles. A system of connected vehicles would require consideration of fault modes at system level, for instance the consequences of faults in individual vehicles in a platoon or faults in connectivity, including faults from hostile interventions. It would not be surprising if there were trade-offs between headway and safety that limited possible increases in effective road capacity.

The current industry-wide technological thrust is directed towards vehicle autonomy, not connectedness. The logic is that the general application of V2V connectivity would need to follow on from the successful adoption of vehicle autonomy – because the benefits of such connectivity largely depend on response times of connected vehicles that are faster than achievable by human drivers. There would then be a question of whether access to certain classes of roads should be limited to vehicles equipped with V2V technology. However, the efficiency gains from such dedicated infrastructure use would depend on widespread adoption of V2V communications, for which, as noted, there may be a lack of incentive for vehicle manufacturers to develop.

Although the potential of connectedness is unclear at present, it would be sensible for a new legal regime to allow for this possibility. On the railway, statutory regulations place a duty on infrastructure managers and operators to develop safety management systems, as well as a duty on operators to work together to ensure safety. However, the railway is a closed system that can be used only by authorised operators, whereas roads are generally open to all qualified users.

One approach would be to regard V2V and V2I connectivity as simply extensions of visual detection of other vehicles and roadside signs and signals. From this perspective, the legal framework for AVs may not need to be modified to deal with connected vehicles. On the other hand, it is arguable that the absence of some kind of system-level safety requirement would be incompatible with public expectations in the event of a crash involving numbers of connected AVs.