Uber’s buccaneering entry into regulated taxi markets in many cities prompts questions about the purpose of regulation and who benefits. While there is little academic literature on the topic, a 2016 paper* by Harding, Kandlikar and Gulati, focused on North American taxi markets, is illuminating. It is argued that the case for regulation is based on the view that the taxi market suffers from three problems: ‘credence good’, open access and thin market:

  • A credence good is a good or service whose quality cannot be determined by the consumer until after it has been consumed. Questions about reliability of a taxi service may deter users who may be concerned about excess charges or a poor quality vehicle. Regulation that sets standards for quality and price overcomes such market failure.
  • Open access to the market may attract large numbers of new entrants on account of low costs of entry. Given limited demand in the locality, earnings of drivers would fall, increasing the incentive to illegitimate charging and poor vehicle maintenance.
  • A thin market has a small number of buyers and sellers, which reduces the chances of matching supply and demand. The taxi market is thin in that it is geographically dispersed. Regulation of fares prevents exploitation of users when demand exceeds supply.

The entry of Uber and similar ride hailing platforms impacts the taxi market in a number of ways:

  • Barriers to entry for drivers are lowered, and users are attracted, shifting a thin market to a thick one.
  • Fares flex according to demand but are specified before the trip is undertaken. Surge pricing attracts drivers to meet peaks of demand.
  • Quality rating of both drivers and passengers, plus predictable fares, helps ensure consistent standards of service.

Thus the platforms address the shortcomings of traditional taxi markets that have justified regulation, effectively removing two of the rationales for taxi regulation, and largely mitigating the third (open access), Nevertheless, the implications of competition between platforms are as yet unclear. Competition could lead to instability on both supply and demand sides, which could result in collusion by platforms, to the disadvantage of drivers and passengers; while lack of competition may result in monopolistic pricing.

The paper concludes that regulators should allow the ride hailing market to grow and focus on the possibilities of future monopoly and of collusion between platforms.

*Taxi apps, regulation, and the market for taxi journeys. Transportation Research Part A: Policy and Practice, 88, 15-25, 2016.

 

 

I blogged recently about the economics of Uber and other ride hailing services. Further light has been shed by the excellent new book from Andrew McAfee and Erik Brynjolfsson of the Massachusetts Institute of Technology: Machine, Platform, Crowd. From this I gleaned the thoughts below.

Network effects have long been recognised: some services become more valuable to each user as more people use them. The telephone is the historic example, WhatsApp a recent instance. Network effects reflect demand-side economies of scale, where benefits to users, the source of demand, grow as the scale increases (contrasted with supply-side economies of scale where costs fall as scale increases).

Two-sided digital platforms serve to match supply with demand. The most successful take advantage of network effects to become powerful aggregators of both supply and demand. They are early to the business space and pay a great deal of attention to user interfaces and experience, since users may be unwilling to employ more than one or two competing platforms.

In the transport sector, we are concerned with platforms that function ‘online to offline’ – digital access to physical mobility. Examples: Uber, Lyft, Gett and other taxi services, BlaBlaCar and Liftshare for ride sharing, car clubs, dockless bike hire by app, online rail ticketing. As well as matching users with services, the platforms optimise operations, for instance selecting the fastest routes and predicting the location of future demand. The negligible cost of digital scaling means that these platforms can handle huge volumes of information – about user preferences, availability and price of services, payments etc.

In the past such data handling would have been limited to large organisations. Now, the availability of cloud computing with unlimited amounts of capacity helps innovators enter the market, scale rapidly and compete aggressively.

Demand side economies of scale can grow much faster than costs. However, the main challenge for digital platforms arises because the supply side involves physical plant and infrastructure whose capacity is finite, hence capacity, a perishable commodity, must be carefully managed. An important tool is revenue management, pioneered by the budget airlines, where varying price is used to match supply with demand – which needs lots of data and lots of supply and demand to run well.

As well as benefiting from network effects, digital platforms can reduce information asymmetries that inhibit transactions, such as whether you can trust your taxi driver, particularly in an unfamiliar city. Uber asks both customers and drivers to rate each other after each transaction, which allows poor performers to be dropped and increases confidence in quality of service.

Operators of two-sided platforms typically prefer lower prices than their providers of service. The maximum revenue of a taxi service arises at low fares, given the price elasticity of demand. However, two-sided platforms have to satisfy both providers and users. Lower fares increase demand, which will attract more providers onto the system, a benefit to providers. But lower fares also mean less income to drivers.

There is a belief that two-sided platforms for taxis offer network demand side economies of scale such that the biggest platform will dominate each local market. Patient capital to support growth of the market will reduce the marginal costs of arranging a ride, to yield attractive returns to investors.

Assessment

Analysis of the economics of transport digital platforms is at an early stage.  A key question is whether scale economies would tend to result in monopoly, or whether competition would arise on account of low barriers to entry and a gig workforce open to recruitment by the offer of better terms.

 

 

 

 

 

 

 

 

 

 

Check refs

An interesting article by Len Sherman in Forbes magazine argues that unregulated taxi services are characterised by bounded demand, abundant supply given low barriers to entry, relatively undifferentiated service quality, low customer switching costs, high variable costs and virtually no economies of scale. Historically, this led to regulation, for example the set fares and ‘Knowledge’ requirement for drivers of London’s black cabs, which allowed profitable operation and acceptable remuneration, at the expense of consumers who paid higher fares.

Uber’s buccaneering approach bypassed taxi regulation, and allowed rapid market penetration of a service that is superior in many respects to existing taxi and private hire services and hence very popular. The question is whether Uber can be profitable, which Sherman doubts.

The economics of ride hailing, or demand-responsive transport (amongst the variety of terms in use), is not straightforward. These businesses compete for both customers and drivers. Low fares attract customers but deter drivers who have other choices for getting clients. Reducing the commission taken (20-25% for Uber) could benefit drivers and customers but would reduce profitability and disappoint investors. Raising fares could benefit both drivers and profitability but would encourage competitors to enter the market.

Much depends on whether the market for ride hailing tends to monopoly, so that the dominant incumbent is able to deter new entrants by short-term fare reductions; or whether sufficient competition would naturally arise because barriers to entry are not high (digital platforms being replicable). Lyft, Uber’s main US competitor has been gaining market share, which suggests a competitive market may be possible.

For governments, there are a number of challenging questions about the future regulation of taxis and private hire vehicles:

  • Create barriers to entry that result in effective monopoly supply, with improved driver remuneration but higher fares (the situation with regulated fares)?
  • Or aim for a level playing field for competing suppliers, to encourage innovation and benefit consumers?
  • Segment the market with distinct classes of providers, or allow new entrants with novel offerings, such as demand-responsive minibuses?
  • Protect remuneration of drivers in a competitive market by application of the ‘minimum wage’ concept?
  • Regulate overall numbers of ride hailing vehicles to mitigate congestion and to limit diversion of passengers from buses, or regard such services as helpful in reducing individual car ownership?

One approach to the problems experienced by Uber drivers in London is being developed by the New Economics Foundation – a driver-owned alternative that is just as convenient and competitive on price, but treats its passengers and drivers with respect. However, the question is whether the economics of ride hailing allow better rewards for drivers, over and above ‘respect’, particularly since a driver-owned cooperative would be constrained by limited capital in the face of competition from well-capitalised commercial enterprises able to sustain short-term losses while seeking market dominance.

Perhaps a better approach to improve the position of drivers would be to devise an app that allowed them to achieve improved outcomes while continuing to work for Uber and other ride hailing businesses. The key point is that transport services are very time sensitive and therefore vulnerable to interruption. It is noteworthy that Ryanair, which had long declined to recognise the pilot unions, has recently agreed to do so when faced with a strike in the run up to Christmas.

Organising collective action by drivers by means of an app might be one approach to securing better terms. But it might also be possible to take advantage of Uber’s surge pricing system in which prices rise when demand exceeds supply, thus deterring some customers and attracting additional drivers. An app that allowed drivers collectively to constrain additional supply could prolong the duration of surge pricing to their benefit.

The operation of such driver-organising apps in an already complex industry would need careful analysis, based on game theory concepts. It is possible that the outcomes for drivers would be better than head-on competition by a driver cooperative.

A kind of precedent is the Vegas Kickback app, which rewards drivers, over and above the fares earned, who take customers to commission-paying destinations in Las Vegas – nightclubs, massage parlors, gun ranges etc.

Charles Musselwhite has edited a new book on transport and travel in later life. I have a chapter on Future Transport Technologies for an Ageing Society. I discuss how the new digital technologies are affecting both the transport system based on civil and mechanical engineering technologies, and how we choose to travel. There are a number of ways in which innovations would be of benefit to those in later life, and a number of policy approaches that would help achieve such benefits.

In pre-Budget briefing, the Chancellor expressed enthusiasm for government investment to help get driverless cars on UK roads by 2021. In this he followed the example of his predecessor, George Osborne, who was keen that Britain takes bold decisions to ensure that it leads the world in new technologies and infrastructure. The Budget was followed by the launch of the Government’s Industrial Strategy, which featured new modes of mobility as one of four Grand Challenges, and stated that the Government wants to see fully self-driving cars, without a human operator, on UK roads by 2021.

There are two related reasons why the Government might attempt to ‘pick a winner’ of this kind: industrial policy and transport policy. If, as George Osborne claimed, driverless cars represent ‘the most fundamental change to transport since the invention of the internal combustion engine’, then support for autonomous vehicles should form part of our industrial strategy as well as our transport policy.

The Government has indeed been active in supporting trials and future deployment of driverless vehicles on British roads. Projects are underway in Bristol, Coventry, Greenwich and Milton Keynes; R&D is being funded; Codes of Practice for on-road testing published; and legislation introduced in Parliament to ensure that vehicle insurance covers both the motorist when driving as well as the car in automated mode.

There are two routes to driverless vehicles. The evolutionary approach, pursued by all the main international auto manufacturers, offers to relieve drivers of tedious tasks, for instance by means of adaptive cruise control, which regulates the speed and space to the vehicle ahead. The revolutionary route, pioneered by Google (now branded Waymo), dispenses with the driver entirely. Other US businesses with disruptive approaches are active developing driverless technologies, including Tesla and Uber.

With the enormous worldwide effort underway, it is going to be difficult for the UK projects to make much impact – unless they have some breakthrough technologies under test, of which there is no public evidence. Nevertheless, Government support for driverless vehicles might be justified if the impact on the transport system were clearly beneficial.

Transport benefits

Autonomous vehicles would need to be demonstrably safer than their human-driven counterparts to be publicly acceptable. This should be achievable, given that most crashes are caused by human error. So we may expect a safety benefit from driverless cars, and lower insurance costs. Beyond that, it seems likely that there will be two main impacts.

First, replacing the human driver with a robot would lessen the cost of licensed taxis and private hire vehicles, enhancing their competitiveness, which is why Uber is so keen on this technology. Such robotic vehicles would fill the present gap in service provision that exists between high-capacity, low cost public transport and low-capacity, high cost taxis.

Proponents of public transport are anxious lest demand for efficient high-occupancy buses  is reduced. On the other hand, the ready availability of robotic taxis would reduce the attractions of individually owned cars, and robotic shared-use minibuses would allow door-to-door conveyance at reasonable charges. So we can envisage a future in which the availability of shared-use autonomous vehicles fosters a shift away from private ownership in urban centres, with a beneficial impact on traffic congestion.

Second, private ownership is likely to remain popular in suburbs and beyond. Driverless cars will allow new options, for instance sending the vehicle home unoccupied, after delivering the occupant to their workplace, for use by others in the household. This could reduce car ownership per household, but would increase vehicle-miles travelled. Another option would be ‘parking on the move’ – programming your car to cruise round the block while you do your shopping. However, such unoccupied vehicles would add to traffic and worsen congestion in urban areas, so would need to be regulated, to give priority to occupied vehicles.

The main problem of the transport system is road traffic congestion. The test for any new technology is its likely impact on congestion. For autonomous vehicles, there are many possible impacts, both positive and negative. In the absence of evidence from deployment at scale, the outcome is uncertain. So the transport policy case for support for driverless vehicles remains unclear. Yet that lack of clarity justifies cautious support of development, testing and deployment, to understand better the implications of what could yet turn out to be an important innovation.

Traditional transport technologies based on mechanical and civil engineering develop quite slowly. In contrast, digital technologies are fast and disruptive. Autonomous vehicles are where the digital hare has to ride on the back of the mechanical tortoise, with as yet uncertain consequences for the speed of travel.

A version of this blog appeared in The Conversation on 21 November 2017.

 

 

I was a member of a Commission on the future of London’s roads and streets convened by the Centre for London. Our report was launched yesterday.

This was a worthwhile exercise that stimulated thought and discussion amongst the expert members. The report will contribute to the actions that will be needed to implement the plans of the Mayor’s Transport Strategy. Some of the toughest issues concern the management of demand for both road space and kerb space while maintaining traffic flow and improving the quality of place in a city growing at the rate of about 100,000 people a year.

Travel time and congestion

Britain’s National Travel Survey has been monitoring our travel patterns for the past 40 years. One remarkable finding is that average travel time has held steady over this period at about 375 hours per person per year, close to an hour a day. There are only 24 hours in the day and many activities that have to be fitted in, leaving an hour or so for daily travel. The history of travel is largely about travelling faster as incomes have grown and technologies advanced, allowing us to go further in the same amount of time.

This travel time constraint is an important influence on road traffic congestion. As traffic builds up, speeds fall and trips take longer. This puts pressure on time available for travel and some road users change their plans – travelling at a less busy time, or to a different destination (where options exist, as for shopping), or by a different mode, or not to travel at all. So congestion is self-regulating, and gridlock is rare, particularly when an urban traffic management system is used to adjust the timing of traffic signals to make best use of road space, as in London.

Road traffic congestion occurs in areas of high population density and high car ownership. There is insufficient road capacity to accommodate all the trips that might be made. Many potential road users are deterred by the prospect of time delays. These suppressed trips mean that congestion is difficult to mitigate since measures aimed at reducing car use in effect create space for drivers previously deterred.

Policies that it is hoped would reduce congestion by getting people out of their cars tend to disappoint. Promoting walking and cycling is good for health and for the environment but has little impact of urban car use. Increasing road capacity induces more traffic, hence the maxim that you can’t build your way out of congestion, which we know from experience to be generally true.

Congestion charging

London pioneered congestion charging, aimed at reducing demand for car use in the central area. On introduction in 2003 there was a marked impact – both car use and delays were reduced. But when the charge was increased from £5 to £8 in 2005, there was no further effect, and over the next few years delays reverted to the previous level. So while congestion charging has generated useful amounts of revenue for investment in London’s transport system, it has been disappointing as a means of reducing congestion – because of the high potential demand from drivers willing to pay the current charge of £11.50. It seems likely that a substantially higher charge would be needed to make a significant impact on congestion in a city like London where many have high incomes.

When road users are asked why congestion is a problem, their main concern is the uncertainty of journey time. While it is difficult to reduce general delays arising from congestion, new technologies are able to lessen uncertainty. The route planning offered by smartphone apps such as Google Maps and Waze, and similar in-vehicle devices, takes account of traffic congestion in real time and proposes less congested routes, making better use of the road network. These apps also estimate journey time, which helps decide the best time to start a trip.

Performance of the route guidance apps would be enhanced by collaboration with road authorities. Transport for London’s 2017 Business Plan announced a new partnership with Google on its Waze technology that will see Google use TfL’s open data, while TfL uses the Waze crowd-sourced data on road conditions to help manage traffic around incidents and road closures.

These new technologies are helping us make better use of the road system within the limited time we can allow ourselves for daily travel.

 

 

The Mayor of London published a draft Transport Strategy in June. The consultation period has just finished. Generally, the draft is sensible in its intentions, emphasising the important aims of healthy streets and people, good public transport, and accommodating a growing population and economy.

One stated objective is to achieve 80% of all trips by walking, cycling or public transport by 2041. However, while reduction in the share of journeys by car is desirable, it would be very ambitious to achieve this aim.

Car mode share has been falling in London, from 50% of all trips (driver and passenger) in 1993 to 36% in 2015. I projected  that on existing policies it would fall to about 30% by 2040. This is happening because the capacity of the road network prevents traffic growth, and population growth therefore results in a decline of car mode share. However, a reduction to 20% car mode share of trips could be difficulty for the following reasons:

  • Walking in London has remained at 24% of all trips consistently for the past 20 years. Walking is the slowest mode, other than for very short journeys, and could therefore be difficult to increase.
  • Cycling is growing, but from a low base. However, it is difficult to get people out of cars onto bikes. Copenhagen has excellent cycling infrastructure and a strong tradition of cycle use, which comprises 30% of all trips. However, car use at 33% is not very different from London. Walking and public transport use are much lower than in London. Crowding on public transport is likely to be reduced by promoting cycling, but there may not be much impact on car use.
  • Growth of bus use will tend to be constrained by traffic congestion, and growth of rail use by crowding and the cost of investment in new routes.
  • Reducing carriageway available for cars, for instance by allocating more road space to pedestrians and to bus and cycle lanes, would tend to reduce car use, but would not reduce congestion and would be detrimental for goods deliveries.

 

 

Transport for London has refused to renew Uber’s licence as a private hire operator after expiry of its current licence on 30 September, on the grounds that Uber lacks corporate responsibility in relation to a number of issues that have potential public safety and security implications (more detail available). Uber is appealing against this decision. Thus far more than 700,000 people have signed an online petition in support of Uber.

The Mayor, Sadiq Khan, has endorsed TfL’s decision, but has recognised the popularity of Uber and has said that there is a place in London for all private hire companies that play by the rules and that he wants to help support innovative businesses in London and to create a vibrant and safe taxi and private hire market. The Mayor is consulting on his draft Transport Strategy, which has the ambition to reduce car use to 20% of all trips by 2040. One reason why car ownership and use in London is on a downward trend is that there are alternatives to the private car, Uber and the like as well as buses and trains.

3.5m Londoners use Uber via its smartphone app and more than 40,000 licensed drivers work for the company. While there have been criticisms of a number aspects of aspects of Uber’s operations, it is evident that this innovative service meets a real mobility need. Its virtues include the clear location and timing of the arriving vehicle with a named driver, the ability to rate the driver after the trip, no need to pay by cash or card, and applicability of the same app throughout London and to all cities that Uber serves. These attractions, together with relatively low fares, account for the success of Uber in competition with black cabs and local minicabs.

TfL’s refusal to renew Uber’s licence on grounds of safety and security seems at odds with the popularity of the service amongst Londoners. Uber must have a fair chance of modifying the decision on appeal to the court or in negotiation. But if not, others will seize the opportunity to step in. The drivers and their cars are there, as are the smartphones waiting to be loaded with new apps. There are other platforms in London such as Gett and mytaxi, (both for black cabs at present), and it is reported that Uber’s main competitor in the US, Lyft, is thinking about coming to London.

An important question is whether a single platform would tend to be dominant on account of economies of scale and scope – spreading the overhead costs across the greatest number of users and providing the fastest response times. Or whether stable competition between a number of providers could be achieved, as they compete for both clients and drivers by offering attractive terms to both. Lyft has been growing market share in New York at Uber’s expense. The New Economics Foundation has proposed ‘Khan’s Cars’, a mutually-owned, publicly-regulated alternative to Uber.

Another question is the sustainable level of fares for services provided by these platforms, which would need to be sufficiently high to attract drivers but low enough to be competitive with conventional taxis and minicabs. The app-based platforms such as UberPool allow customers to share journeys with other going in the same direction at lower fares – a source of competitive advantage as well as a means of increasing vehicle occupancy, which is helpful for efficient use of the road network.

For the longer term, the possibility of driverless vehicles offers scope for substantial cost reduction for robotic taxis. This could allow population of the vacant space between high capacity, low fare public transport offering stop-to-stop or station-to-station service, and low capacity, high fare taxis offering door-to-door journeys.

 

 

 

 

 

 

 

The National Infrastructure Commission has sought evidence on how the deployment of intelligent traffic systems could help optimise the road network. I sent a response, found here Metz NIC traffic management 1-9-17

My argument is that we need to move beyond traffic management using traffic signals to management involving collaboration between public sector road authorities and the private sector suppliers of digital maps and route guidance apps, such as Google Maps and Waze. These apps have become very popular for turn-by-turn route guidance that can take account of, and help avoid, traffic congestion and provide estimates of journey time before setting out. These features help tackle the main problem of traffic congestion which is journey time uncertainty.

I would expect that collaboration between public and private sectors would improve both the experience for road users and the efficiency of network operations.

In late July the Government published its Plan for tackling roadside nitrogen dioxide (NO2) concentrations, to reduce these below the statutory limit. Publication was met by considerable criticism that the Plan lacked urgency and effectively dumped the problem on the most affected local authorities, which would be required to implement Clean Air Zones (CAZ).

An accompanying 155-page Technical Report included, on its final page, new estimates of the economic benefits from reducing damage to health through measures to reduce NO2. Remarkably, these new estimates were very substantially below those that had been published in the Technical Report of May 2017  that accompanied the consultation preceding the Plan. For instance, the previous estimated health benefit of a further 21 CAZs was £3.6bn, but now is £620m – an 80% reduction.

This huge reduction was attributed to new advice from the independent experts of the Committee on the Medical Effects of Air Pollution (COMEAP), which had found it difficult to disentangle the impacts of specific pollutants, in this case NO2, from that of the whole mix of traffic related pollutants.

COMEAP’s previous advice recommended a central coefficient of 1.025 per 10ug/m3 NO2, which means that for every 10ug/m3 increase in NO2 concentration, the increase in mortality risk would be 2.5%. The revised advice now recommends a coefficient of 1.023 for traffic-related pollution; but COMEAP also recommends that when measures are primarily targeting NO2 emissions this coefficient should be adjusted to account for possible overlap between the direct impacts of small particulates and NO2. Applying their recommended adjustment, the resulting coefficient used for the analysis is 1.0092.

My enquiries of the Government’s Joint Air Quality Unit elicited the confirmation that overall the updated NO2 damage costs for road transport are approximately 80% lower than those used at consultation. This splits into roughly 60-65% resulting from the revised COMEAP advice and the remaining c.15-20% resulting from the other updates such as new dispersion modelling and population data. The JAQU confirmed that the reduction in the road transport NO2 damage cost primarily reflects a reduction in the estimated mortality impact associated with NO2 alone.

There has been widespread interest and concern about the health impacts of roadside air pollution over the past year and more. So it is surprising to learn of such a substantial downward revisions of official estimates of health impact. However, the relevance for policy is as yet unclear.

Current policy is driven by the need to avoid exceeding statutory limits for NO2 concentration laid down in a EU Directive that applies uniformly to all regions of the Community, in which context the scale of health benefits from remedial measures is not relevant. However, Britain is to leave the EU, which may open the possibility of regulation of air quality based on UK targets set to reflect the balance of benefits in relation to costs. A down rating of health benefits would then be relevant, particularly given the expected reductions in pollutants from improved vehicle technology and from the introduction of electric propulsion. Whether the Government would be willing to propose targets to reflect UK conditions remains to be seen – it may depend on the outcome of the Brexit negotiations and on a judgement about the politics of air pollution.

 

A version of this post was published in The Conversation on 30 August 2017

https://theconversation.com/no-not-as-bad-as-we-thought-83056