Rail 789: Voodoo economics and driverless car myths

Railway economics always smacks of smoke and mirrors. However, the report by Peter Hendy on the investment programme is more like deep fog and blind alleys. Commissioned to report on the prospects for Network Rail’s enhancement, Hendy has somehow contrived to salvage virtually all of it, despite the £2bn – and rising – overspend on a single project, the Great Western electrification programme.

Hendy was lured earlier this year from Transport for London, where he had been commissioner for nearly a decade, as chairman of Network Rail to replace the hapless Richard Parry-Jones who had been so invisible that few rail managers I talk to had ever met him. Hendy’s immediate task was to sort out Network Rail following the revelations that the £38bn investment programme for 2014-9 – control period 5 as it is known in the industry – had gone completely off the rails (I thought you eschewed railway clichés? – ed; Sorry, but for once nothing else works CW). In particular, just a few months into the five year period, the enhancement programme, worth around £11bn,  was proving impossible to deliver either on time or on budget despite having been given the go ahead by both the Office of Rail and Road and ministers so recently. Indeed, both Network Rail and ministers contrived to ignore the problems of the enhancements, despite mounting evidence, until after the election, which meant that Hendy was forced to conduct his review quickly to ensure that the programme did not collapse entirely.

Hendy’s 43 page document, which is the first result of his appointment, says that Network Rail has undertaken a thorough review of ‘the cost and deliverability of the enhancement plan’ with updated cost estimates and delivery dates. Would it be churlish to ask just why this had not been done before? The core business plan has been re-examined and the whole lot considered, once again, by the Network Rail board (gosh, really!).

And guess what. All is pretty much rosy in the Network Rail garden, despite the fact that the cost of the Great Western electrification has gone further north than the Arctic Circle. According to Hendy, ‘While the vast majority of projects are being delivered on time and on budget, there are a small number of projects for which the forecast cost estimates are significantly higher than originally assumed, particularly the electrification projects and several projects where the scope was poorly defined at the outset’. Ah yes, but there are two problems with that statement. First, it is the biggest projects that have gone awry and secondly it is rather like a surgeon claiming he had a success rate of 75 per cent because the patient recovered well after the first three operations but died after the fourth. Mark Carne the Chief Executive of Network Rail tried the same line when I met him in the summer but It won’t wash.

I asked Hendy on the day the document was published, why in an industry using modern technology, had costs for electrification gone up an estimated six fold (a figure courtesy of Roger Ford from Modern Railways) compared with the East Coast that was electrified a quarter of a century ago. Surely, I suggested, given progress in technology, which  means there is the need for far fewer boots on the ground, we should be paying less – or at worst the same – amount for putting up masts now. Answer, came there none. Hendy said that there had been a loss of experience as there had been no major electrification in the intervening period and that we should ‘not be looking at the past, but the future’. I tried to press him but it proved pointless. He also said that a fundamental mistake had been that the project was ‘not managed in a holistic manner’ and that ‘one of the challenges was that it was being designed and delivered simultaneously which is undesirable’. Instead, it should have been fully planned and designed, then implemented. None of this, it seemed to me, was exactly rocket science and begs the question of how Network Rail, now almost 15 years since taking over from Railtrack, is still learning such basics.

Hendy’s report is very much a top down analysis, but I suspect that to understand Network Rail’s high cost base – and this is not the first time that it has been so apparent that everything Network Rail touches seems to cost more than expected – is the entrenched risk aversion in the organisation. Of course accidents must be avoided, but at times Network Rail practice seems to be based on a failure to properly assess risks. What else could explain the fact that the piles being driven to support the masts are being dug to more than 5 metres when 3 metres has been the industry norm. Or that because no one knows precisely the location of existing signalling wires, many holes are being dug by hand – yet equipment to find buried wires exists. That is why rather than the promised 18 masts per night being erected, the team operating the special misnamed ‘high output’ train often only manages two or three.

Oddly, Hendy’s report does not mention one factor which he could have used as part of the excuse: the fact that nowadays in the franchised environment compensation has to be paid to train operators if they cannot have access to the track, something which does dramatically push up costs (but not, as has happened with Great Western Electrification from £800m to £2.8bn).

Hendy’s report does not go into this kind of detail when, surely, the overspending of Network Rail on projects, and its bureaucratic GRIP process should have been considered as part of his remit. After all, the one way to ensure that the future programme can go ahead is, surely, to ensure that costs are not excessive. Instead, we have a solution that involves more spending.

However, we can all rejoice that the gap has been plugged. Well this is where the fog and blind alleys comes in. Apparently, Network Rail has been given permission to borrow a further £700m, above the previous limit, and will sell some £1.8bn worth of non-core assets to pay for the overspend.

That begs a lot of questions. First, how come Network Rail has not considered selling these assets previously if they are ‘non-core’. The reason is that assets are, well, assets. That means they are currently earning a rate of return – let’s say conservatively something like 10 per cent. So if Network Rail is selling assets worth £1.8bn, it will lose some £180m in annual revenue that will have to be made up somehow. Or perhaps the organisation can just lose that sort of sum without any consequence, a demonstration of just how unaccountable it is. There will, too, be some slippage into Control Period 6, but precisely what will be delayed has yet to be announced.

In a way, all this is wonderful. The railway has got off scot free and the unsustainability of its finances has not led to any substantial cuts, with just a few schemes being kicked into touch. As Mrs Thatcher once said, ‘Rejoice’. Despite the budget for the Department for Transport being slashed by 30 per cent over the next four years in the Comprehensive Spending Review, there is little talk of belt-tightening for the railway. It seems Xmas has arrived early for the industry. However, it does not take a Mystic Wolmar to predict that despite all this, there is trouble ahead as the figures simply do not add up. You read it here first.



Driverless cars myths


One of the arguments deployed by anti-HS2 campaigners is that driverless cars will soon obviate the need for trains. We will simply sit in our own personal vehicles, reading the paper or more likely playing with our mobile devices while a robot car (which is a better name for the concept) whisks us to our destination.

I attend a lot of conferences where it is considered conventional wisdom that these robot cars will soon take over the world and that we must prepare for them. Oddly, just as I was writing this piece, a press release about a new report from a consultancy called Juniper Research on driverless cars plopped into my inbox. The headline read ‘self driving cars to take off by 2021, approaching 20 million on the road by 2025’.

Despite huge amounts of government and private money being invested in the concept, robot cars are nowhere near being either technically or financially available. The technical problems are legion: the Google cars that have run up hundreds of thousands of miles in California are restricted to 25mph and are unable to be used in fog, heavy rain or snow (not that there is much in San Jose). Moreover they have had 11 accidents so far, none of which has been the fault of the robot car but, interestingly, most have been rear end shunts. That’s probably because they stop and start unpredictably as their sensors detect danger that a human driver would ignore.

Let’s take another issue. Humans driving cars are able to pull out onto busy roads with a stream of traffic by gradually edging out and finally being allowed out. A robot car would detect danger permanently and never edge out, therefore causing jams or having to be overridden.

Then there is the cost. The sensors being developed are exceedingly expensive and these cars need lots of them. Another misconception. There is an assumption that these cars will not be individually owned but rather, like the London Santander bikes, will be hired out by the hour or so. This is an odd assumption and is predicated on people changing their attitude towards car ownership, something for which there is no evidence.

And finally there are major legal hurdles to overcome – who will be to blame in an accident? The software manufacturer, the car designer, or the owner?

The report cost £990 so I decided that I would not bother to examine it further but I can state categorically that its conclusions are bunkum. I am happy to bet with anyone that in 2015, there will not be 20m driverless cars on the world’s roads – or indeed, I suspect, any except those being tested or used experimentally. Good luck to Juniper Research, but I hope that most people will be sensible enough not to shell out a grand for its ‘research’.

So while there may be all sorts of reasons why HS2 should not be built or will be an expensive white elephant, robot cars are not one of them.


  • John A

    Christian, I look forward to the opportunity of discussing driverless cars with you! Taking up just one point, for now, pulling into a busy stream of traffic. A computer algorithm would be able to risk assess the move with precision, even calculating how close the car behind would come to your own vehicle and how much the car would have to decelerate/brake. But granted, the sensors do appear to be a big problem however if humans can identify objects in heavy rain, I think computers with the right digital image processing will be able to do the same. (Note that low level light cameras now have far better night vision that humans.) The real advantage of driverless/autonomous cars over human driven cars is that the style of driving of the former is/will be entirely predictable and I assume determined by national regulators. As an aside you don’t need segregation for cyclists if there are only driverless cars on the road.

  • Paul Holt

    (Penultimate paragraph) given that it is now 2016, that is a very safe bet.

  • Christian Schmidt

    If they are operated like London Santander bikes, then who will ensure that they are at the right place?
    Many years ago there was a presentation at a STAR conference suggesting turning the Glasgow South Suburban System into one of automated 6-seater pods. I probably didn’t endear myself to my bosses, who were waxing about the ‘taxi-like’ service quality this may offer but asking how this would work in the peak hour (and doing so repeatedly when no answer was forthcoming…)

  • TorontoisRoy

    The big problem with driverless cars is that there little new road space being constructed (which is a good thing in itself). So driverless cars will be in slow traffic and jams like every other vehicle. As such, being bumper to bumper, there is no increase to road vehicle capacity.

    Driverless cars are this decade’s Hydrogen Cars – fancy technology that doesn’t change the fundamentals.

    Moreover, few can read in a car due to headaches, but one can easily read and work on a track. Plus rail vehicles are much more efficient in moving people per square metre, and per unit of energy.

  • LondonCalling

    You are spot on about driverless cars. Talk about not thinking something through !