Is DRT the Answer? Depends what the question is
DRT is touted as the solution for reducing costs and increasing quality. But it cannot do both.
At the start of this month, Milton Keynes became the first local authority to replace its tendered bus network with Demand Responsive Transport. As with the majority of DRT applications (ViaVan in London, ArrivaClick in Liverpool, PickMeUp in Oxford, etc), the technology underpinning this application is provided by Via - an American startup providing similar technology to DRT solutions globally.
What’s it for?
In assessing whether this solution is likely to succeed, we first need to be clear what problem it is trying to solve. Luckily the council minutes are explicit; for Milton Keynes DRT is a purely financial measure. The council expects bus demand to return to just 80% of its previous level (and council income to remain suppressed) and so feel the need to reduce the costs of subsidising buses.
This puzzles me because one of the characteristics of DRT implementations across the board has been a tendency to lose money. Indeed, the poster-child for mini-bus style DRT was Chariot, a very similar model to Via for which Ford paid $65 million and then shut down two years later. Slide was launched by RATP in Bristol but never reached breakeven and closed in 2018. To my knowledge, not one of the Via-powered solutions in the UK has ever achieved breakeven.
So it seems odd to assume that a model that has only ever generated losses will be more likely to save money than one (buses) that has at least proved an ability to make money in some circumstances.
Chariot was fired
The assumption that the model is the reason why Milton Keynes’ tendered routes lose money is also odd. After all, the entire point of tendered routes is that they’re non-commercial. If they made money, they wouldn’t exist. As the minutes rightly say, the problem is low demand:
Historically demand levels in Milton Keynes have proved difficult for operators to sustain commercial fixed stop routes. As a result, although there are a small number of wholly or almost wholly commercial services, the majority have been subsidised either fully or partly via contracts with the Council.
This results in an average cost to the taxpayer of £2.38 per trip. The business case appears to have been written on the basis of 80% of post-Covid demand being retained.
80% of previous demand being retained suggests that they’re expecting demand of around 1 million trips per year going forward. The paper says the DRT solution will reduce costs to £2.1m from £2.8m, which suggests an expectation of future subsidy of around £2.20 per single trip (down from £2.38 currently).
That seems reasonable to me. Like most people, I never travelled on the previous Milton Keynes tendered bus network, but my understanding is that it was pretty lightly loaded. With most buses operating at below seven passengers per trip, the reduction in operating costs from big bus to people carrier would indeed nudge down the subsidy.
So far, so logical.
The flaw in the plan?
But there seems to be something of a lurking flaw in the plan: demand will not be static. The DRT promise is a highly compelling one: be picked up anywhere in Milton Keynes and taken to anywhere else in Milton Keynes, direct, for £2.50 (£3.50 at peak times).
By comparison, Skyline (one of Milton Keynes’ leading taxi firms) charges £3.70 minimum for the first mile and £1.10 per mile thereafter. So the maximum DRT price will be less than the taxi minimum.
That is a truly superb offer for something that looks and behaves rather like an Uber, and would doubtless prove very attractive.
However, the problem with using tiny vehicles is that the costs scale up proportionately with demand. If £2.20 subsidy per passenger is required for the passengers forecast in the business case, roughly £2.20 subsidy per passenger will be required for all passengers - even if demand quadruples. That means a significantly higher subsidy than budgeted will be required. A bit like Deliveroo during Covid, Milton Keynes risks the problem that it’s created something that people really want but which loses money every time it is used.
So how on earth are Milton Keynes going to deal with their excess demand? There appear to be two options: one is to accept the cost of resourcing up to meet demand, and the other is to not fulfil all demand that materialises.
The former could be done by acquiring more dedicated vehicles but this requires significant capital expenditure which will need paying for. Or the service could be supplemented with unbranded taxis, but that means that MK Connect becomes a straightforward subsidy for taxi rides by individuals but with groups (who could, potentially, provide a useful bulk source of income) booking with a cab firm direct for the same experience.
Nowhere does the paper suggest that they expect to increase subsidy if DRT proves popular. And, indeed, there are early indications that they’re going to go for the alternative - and simply not fulfil the excess demand.
Does it work?
The goal of MK Connect is to enable a lower-cost replacement of the previous tendered bus routes. The key point is a replacement - the cost could be lowered simply by scrapping the tendered bus services altogether. So it is essential for this to work that the new solution actually provides a replacement.
To try to work out if it does, I chose five journeys that had previously been available on tendered routes and attempted to book them on MK Connect. The publicity material promises a walk of 150-250 metres to the pick-up point, and a wait of 10-30 minutes (in town) and up to 60 minutes (in rural areas).
For two, MK Connect provided me with a great quality replacement for the previous bus service. They were:
1) Stantonbury campus to Central Milton Keynes, previously provided by tendered bus 18:
2) Emerson Valley to Bletchley, previously provided by tendered bus 24:
For one, it gave me a direct service - but required a 600 metre walk at the start.
Woburn sands, the swan, to central milton keynes, previously provided by tendered bus 18:
But for two it failed to provide a service at all, and defaulted to the local commercial bus route - even though both required 1.4km of walking.
Indeed, in both cases, they required walking that totalled more than half the length of the journey requested.
Conniburrow to Bradville, previously provided by tendered bus 23:
Shenley Wood to The Hazeley Academy, previously provided by tendered route 28:
This latter one was particularly unimpressive: an 80-minute journey on two different buses involving 1.4km of walking to do a trip of 2.25km in total that could be walked (by someone able-bodied) in 30 minutes and which used to have a direct bus.
The 150-250 metre walk to the nearest pick-up point was achieved on only 2/5 trips and for a further 2/5, the customer was required to walk around four times this distance - in order to then get a local bus.
As a customer, an infrequent bus service is poor - but at least you can plan for it. The problem with a DRT solution that offers a nil return is that you’re literally stranded (and let’s assume that a 1.4km walk is equivalent to being stranded for many of the users for which the tendered routes were originally created).
That’s why Uber became such a cash sink (as a reminder, back in 2019 Uber lost $5bn in a single quarter); because fulfilling all requests required a density of drivers that couldn’t be achieved commercially. Uber as a private business knew that it needed to pay for the costs of incentivising drivers to fulfil every order - even at the expense of billions of dollars. As a public service MK Connect looks likely simply to not fulfil excess demand.
DRT doesn’t make money
The fundamental problem is that DRT flies in the face of the laws of transport physics. Pretty much every transport service on earth (airlines, ferries, trains, buses) operates on the basis that the fully-loaded vehicle generates a significant profitability surplus, which is thus able to cover the times of lower demand when a fully-loaded vehicle doesn’t happen (or, in the case of revenue managed demand, when seats are sold at below cost price).
But a seven-seat Milton Keynes Connect is capped at 7 x £3.50 = £24.50. Assuming, for sake of argument, each peak seat is occupied twice, that means that the vehicle earns £50 in a peak hour. Assuming the driver needs to earn at least £12 per hour and we need a few quid for fuel, that means there’s just £35 of surplus to cover the off-peak periods when fares are lower and demand less. Remember, this is a universal service, operating out into the villages around Milton Keynes at a flat fare. By contrast, a double-deck bus can earn £500 in its peak hour; which generates over £450 of profit to cover the quiet times.
That’s not to say that DRT isn’t a great product but it means that it will always require a substantial public subsidy.
Indeed, thinking back to when I was founding Snap, two key things strike me. Every time we modelled it, we could only make the service profitable with big vehicles. And we couldn’t do so by offering anywhere to anywhere. While Snap was designed to act a bit like an intercity DRT service, all our modelling suggested that it was impossible to every pick up and every drop off, and we needed to start on key corridors and build up.
Based on the modelling we undertook, a universal service using small vehicles simply didn’t stack up.
For Milton Keynes to have introduced it as a cost-saving measure therefore feels ambitious.
“I wish to register a complaint”
I’ve not used MK Connect but I did try Oxford’s PickMeUp and Liverpool’s ArrivaClick before they were both withdrawn. (Yes, a pedant could point out that Liverpool’s has not been withdrawn, just scaled back to Speke. But given how tiny the operating area now is, it has, in practice, been withdrawn). Both were absolutely superb. The vehicles were smart, the drivers were incredibly friendly and the app was functional (though not as slick as Uber’s). I suspect, therefore, that the service delivered by MK Connect (to those users offered a ride) will also be outstanding.
Pick Me Up was superb
But there is a wider, interesting point. Unlike Oxford and Liverpool, where Via were a technology provider to an operator, in this case, they appear to be the “operator” themselves. Instead, Via are delivering the service through a network of independent drivers operating under taxi licences.
Here’s the relevant section of the terms and conditions. It’s long but it’s worth quoting in full (their block caps - not mine:
YOU AGREE THAT VIAVAN DOES NOT PROVIDE TRANSPORTATION SERVICES AND IS NOT A TRANSPORTATION CARRIER. WE MERELY ACT AS THE COMMERCIAL AGENT OF THE THIRD-PARTY TRANSPORTATION PROVIDER, DRIVER, OR VEHICLE OPERATOR IN OFFERING TRANSPORTATION SERVICES AND IT IS THE RESPONSIBILITY OF THIS PARTY TO PROVIDE THE TRANSPORTATION SERVICE WHICH MAY BE RESERVED THROUGH OUR SERVICE. VIAVAN OFFERS INFORMATION AND A MEANS OF ACCESS TO SUCH THIRD-PARTY TRANSPORTATION SERVICES, BUT DOES NOT AND DOES NOT INTEND TO PROVIDE TRANSPORTATION SERVICES OR ACT IN ANY WAY AS A TRANSPORTATION CARRIER, AND HAS NO RESPONSIBILITY OR LIABILITY FOR ANY TRANSPORTATION SERVICES PROVIDED TO YOU BY SUCH THIRD PARTIES.
Don’t know about you, but I kinda get the sense that Via don’t really want to take much responsibility.
And it gets better:
VIAVAN MAY INTRODUCE YOU TO THIRD-PARTY TRANSPORTATION PROVIDERS TO PROVIDE YOU TRANSPORTATION, BUT VIAVAN DOES NOT ASSESS THE SUITABILITY, LEGALITY, OR ABILITY OF ANY THIRD-PARTY TRANSPORTATION PROVIDER AND YOU EXPRESSLY WAIVE AND RELEASE THE COMPANY FROM ANY AND ALL LIABILITY, CLAIMS, OR DAMAGES ARISING FROM OR IN ANY WAY RELATED TO THE THIRD-PARTY TRANSPORTATION PROVIDER. YOU ACKNOWLEDGE THAT VIAVAN DOES NOT MONITOR ANY THIRD-PARTY TRANSPORTATION PROVIDER’S ON-GOING COMPLIANCE WITH ANY AND ALL LICENSING AND/OR PERMITTING RULES AND REGULATIONS, AND VIAVAN WILL NOT BE RESPONSIBLE FOR ANY LIABILITY, CLAIMS, OR DAMAGES ARISING FROM OR IN ANY WAY RELATED TO A THIRD-PARTY TRANSPORTATION PROVIDER’S FAILURE TO MAINTAIN A CURRENT LICENSE AND/OR PERMIT. VIAVAN WILL NOT BE A PARTY TO DISPUTES OR NEGOTIATIONS OF DISPUTES BETWEEN YOU AND ANY THIRD-PARTY PROVIDERS. RESPONSIBILITY FOR THE DECISIONS YOU MAKE REGARDING VIAVAN’S SERVICES RESTS SOLELY WITH YOU. VIAVAN WILL NOT ASSESS THE SUITABILITY, LEGALITY, OR ABILITY OF ANY SUCH THIRD PARTIES AND YOU EXPRESSLY WAIVE AND RELEASE THE COMPANY FROM ANY AND ALL LIABILITY, CLAIMS, CAUSES OF ACTION, OR DAMAGES ARISING FROM YOUR USE OF THE SERVICE, OR IN ANY WAY RELATED TO THE THIRD PARTIES INTRODUCED TO YOU BY THE SERVICE.
Just in case it wasn’t clear, later on we get this:
THE QUALITY OF THE TRANSPORTATION SERVICES SCHEDULED THROUGH THE SERVICE IS ENTIRELY THE RESPONSIBILITY OF THE THIRD-PARTY PROVIDER WHO ULTIMATELY PROVIDES SUCH TRANSPORTATION SERVICES TO YOU. YOU UNDERSTAND THAT BY USING THE SERVICE, YOU MAY BE EXPOSED TO CONTENT THAT IS POTENTIALLY DANGEROUS, OFFENSIVE, HARMFUL TO MINORS, UNSAFE, OR OTHERWISE OBJECTIONABLE AND THAT YOU USE THE SERVICE AT YOUR OWN RISK.
And don’t think about taking their driver to the small-claims court if he nicks your phone:
These Terms and the relationship between you and ViaVan shall be governed by the laws of the Netherlands
Now, maybe I’m wrong, but I suspect that when it says that “You agree that ViaVan does not provide transportation services”, that is precisely what most users of MK Connect think it does provide. Indeed, the council’s publicity material is pretty explicit:
The new ‘MK Connect’ service will be provided by Via.
Do councillors in Milton Keynes realise that their ‘provider’ makes it absolutely clear that they do not ‘provide’ the service and take no responsibility whatsoever; right down to whether or not the drivers are ‘harmful to minors’ or ‘unsafe’. I suspect that most users believe that the driver who picks them up in a branded vehicle is an employee of Via, which is a subcontractor of the council.
Indeed, there seems to be something of a hole where an operator ought to be. The app logo looks like this:
However, if MK Connect is operated in partnership with the Council and powered by Via, who is MK Connect itself? That logo appears to give the impression that there is an MK Connect organisation that you could hold accountable. But, as far as I can tell, there isn’t. There is a hole where the operator should be.
There’s a definite communication and duty-of-care issue here. Communication because it looks more like a public transport service than the terms suggest that it is. Duty-of-care because you literally have to accept unsafe drivers as a condition of using the service.
Pragmatically, while the whole exercise was created to cut costs, that doesn’t mean that using unsafe drivers is a good way of doing it. We could cut the costs of the NHS by not using doctors.
It feels like Milton Keynes council may have fallen between two stools: they’re still responsible for subsidising every journey (which means it’ll be a disaster if it proves popular) but to get a small subsidy reduction on existing levels of demand, they’ve given their voters a service which expressly promises to be “objectionable”.
Whose gig is it?
The resourcing model also creates a gig economy question. The recent legal case made it absolutely clear that Uber drivers are not self-employed but are in fact workers of Uber. But Uber chose to interpret the ruling narrowly: an Uber driver is now a worker when they’re carrying people but the same driver in the same car is deemed self-employed when carrying KFC on Uber Eats. It obviously can’t last but they’re holding out as long as they can.
Like Uber Eats (as opposed to Uber), Via appear to be delaying implementing the Supreme Court ruling. But it is, of course, clear that these drivers are workers and not self-employed contractors. They drive branded MK Connect vehicles in set operating hours charging fares set by MK Connect. They are workers.
Milton Keynes council have redacted all the financial information, so we can’t see how the contract works. But I wonder whose workers. After all, the branding is all MK Connect. Presumably, the MK Connect brand is owned by the council? The service is “Powered by Via” but it seems ambiguous from the outside as to who is responsible for paying the sick pay. Right now, I suspect no-one is paying it at all. But the law is clear that someone is responsible for paying it, and eventually will have to start actually paying it. Who is it?
DRT may not be the answer
Because DRT looks like Uber, it is often assumed that it will be as successful.
But the genius of Uber was that they eliminated the friction in accessing a market that already existed and was already huge. The minicab market was vast and about as friction-filled as you can get. The Uber app eliminated the stress (“where’s the bloody cab number? I thought it was on the sideboard”), uncertainty (“he’s just around the corner mate”) and inconvenience (“can you go via a cash machine”).
Demand Responsive Transport has existed in more or less the form offered by MK Connect for decades. The reason why it has been restricted to patient transport or community transport is because it is a cost-inefficient mode. Changing the booking process from a telephone service to an app achieves all the same friction benefits as Uber achieved with minicabs, which is why the service is generally so good. But it’s not alchemy: it doesn’t change a lossmaking model into a profitable one. It changes a loss-making model into a much easier-to-use loss-making model.
This contrasts with genuinely new transport modes such as E-scooters, for which the jury is still out on the economics: they may be absolutely transformational to urban transport.
(By the way, I’m talking about a UK context here. Minibus-based DRT is a mainstream transport mode in countries that make up a global population majority, including all of Africa, much of Asia and even parts of Eastern Europe. These countries have lower labour costs and less competition; giving DRT operators the flexibility to keep customers waiting. I suspect that the promise of an immediate universal service would make those services unviable as well)
So I don’t believe that DRT is likely to be the answer to the question of how to cut the cost of delivering bus services. It’s inherently more expensive. If that is the question that has been set (as it has been in Milton Keynes) then either cost or quality will have to be sacrificed.
The Milton Keynes deployment is a kind of lottery. If you press the button in the right place and right time, you’ll get a great service. If not, you’ll get nothing. And the gig economy operating model pushes risk on to the users that they’re not aware they’re taking on.
But with the long-term subsidy, DRT could be an outstanding solution to the problems of public transport in low-density cities designed for cars.
Maybe the next experiment is for someone, somewhere to attempt a large-scale DRT solution as a genuine attempt at a public transport service - with appropriate long-term investment. They should have no expectation of profitability and they should not attempt to palm the user off with ‘unsafe’ drivers. I’ve no idea if it would work - that’s why we need the experiment - but they should not go into it expecting it to save money. It won’t.
