Archive Section

The Locomotive & Carriage Institution
Home Centenary 2011 Continental Visits Lectures Newsletters History Visits Pictorial Visits Narrative Seminars
Railway Safety - What Does Society Expect?

John Cartledge, Safety Policy Adviser, Passenger Focus

1st April 2008

For any of you who are not familiar with the current pattern of consumer representation in the rail industry, it may be helpful for me to begin with a word or two about Passenger Focus, the organisation under whose auspices I am here today. Its been in existence for more than 50 years under various names, latterly as the "Rail Passengers Council" until it was recently rebranded in an effort to make it sound less pompous and more approachable. Its one of a family of official consumer bodies set up by Parliament to champion the interests of the users of the various state-regulated industries, which include energy supply, telecommunications, water, air transport and the post office, as well as railways.

Passenger Focus has a number of roles. It investigates rail users grievances when complainants have been unable to get acceptable answers from train companies direct. It is consulted by the industry on the whole range of matters which affect its passengers times, fares, reliability, station amenities, vehicle design, ticketing, information, etc. It monitors service quality and highlights shortcomings in operators performance. And it represents the users in consultations with the government, parliamentary committees, local authorities, regulatory agencies, the European Union, and any other bodies whose activities affect their interests.

It has no formal duties regarding safety, but it is actively interested in safety issues when and to the extent that these are of concern to passengers. That means that Passenger Focus has participated in official inquiries into serious accidents, and that it serves on the various safety advisory groups which bring together the rail companies, the trade unions, the manufacturers, the regulator and the government to help shape the industrys strategy in this sphere. We are strictly a lay body, not railway professionals, or safety experts. It will probably be painfully evident from the rest of my remarks that this is the case, so please be forbearing. But what Id like to use this opportunity to do is to raise some questions which have long occupied the minds of safety policy makers in the rail industry, and discuss their implications.

In order to do so, it is necessary to begin by briefly summarising the body of safety laws within which the industry and its regulator operate. This is a mixture of rail-specific and general legislation. The industry-specific legislation tends to be highly prescriptive, and to lay down exact technical requirements. Some of it is venerable, having been enacted in the century before last. A law which makes it mandatory to have block signalling, the interlocking of signals with points, and continuous brakes on passenger trains, is an example of this. Other parts are very recent. Laws which require there to be automatic systems which intervene if trains pass signals set at danger, and centralised locking of doors on passenger trains, have only come into effect in the present century. But the number of such prescriptive laws is few, and given the rapid pace of technological change, it is impossible for detailed legislation to be enacted fast enough to keep pace with new developments in railway equipment and operations - although there are still some people in and around Brussels who clearly think otherwise.

So, for the most part, the industry must set and update its own technical standards, and justify these to its safety regulator. It does so within the framework of public law which applies equally to all industries. Such legislation has by its nature to be couched in general terms, setting goals rather than specifying the means by which they are to be achieved. The over-arching duty which the law imposes on all employers (under the Health & Safety at Work etc Act of 1974) is that of conducting their businesses in such a way that risk (whether to employees, clients or the public at large) is reduced to a level which is "as low as reasonably practicable". This is commonly referred to as the ALARP principle, and it is enforced through the criminal courts.

Penalties for failing to comply can be severe. After the high-speed derailment at Hatfield in which four people died as a result of a failure to replace a piece of track known to be suffering from metal fatigue, the infrastructure operator (Railtrack) was fined 3.5 million and Balfour Beatty, the company to which it had contracted the track maintenance function, was fined 7.5 million (on appeal) (plus several hundred thousand pounds worth of legal costs). Directors of both companies also faced manslaughter charges, and although in the event they were acquitted, this experience not to mention the reputational damage suffered by both organisations will undoubtedly have helped to concentrate the minds of all railway managers on the importance of being able to demonstrate that they are discharging their safety obligations fully.

The challenge that safety decision-makers therefore face is that of deciding what risk control measures (whether technical or procedural) will, in any given set of circumstances, satisfy this ALARP test. Clearly, the phrase does not mean that the railway must be run in a manner which is totally risk-free, because this would only be possible if no trains ever moved a course of action which would have the perverse consequence of displacing passengers onto other modes of travel, all of which would expose them to greater risk than they face in the course of rail journeys, because rail is easily the safest mode of overland surface transport yet invented.

Just in case any of you are not familiar with these comparisons, this chart shows the relative fatality risk per passenger kilometre of the various alternatives as multiples of the figure for the railways. The other public transport modes are in the same order of magnitude, but car travel is more than 20 times as dangerous, and the self-powered modes (cycling and walking) about 300 times as dangerous, while motor cycling is how you hope your worst enemy likes to get about.

Shipping is not shown on this chart, because comprehensive data for that mode are hard to come by, but its record is probably not dissimilar to that of rail. It is possible to normalise casualties by time or by trip, instead of by distance. If you do this the relativities shift somewhat, but rail is always in first or second place. And this remains true even if you factor in the risk arising from the access leg at each end of the rail journey, in order to make comparisons on a door-to-door or whole trip basis.

So, in managing their safety risk, if rail operators are required to reduce it to the level which is "as low as reasonably practicable", they have to ask themselves what this phrase actually means. Fortunately, there have been some legal rulings which have helped to clarify the position. In essence, the courts have said that a balance must be struck between the safety benefits of any particular measure and the resource costs of providing it (or as the judge put it, the "sacrifice" required to do so). The test of reasonable practicability is set by whether or not these costs are grossly disproportionate to the reduced quantum of risk (i.e. the benefits) which can be secured by incurring them.

This is a rational economic judgement, because safety however important is only one of many desirable outcomes which rail operators must seek to deliver. Like everything else, safety has an opportunity cost, because resources devoted to improvements in this field are thereby pre-empted and become unavailable for investment instead in greater frequency, higher reliability, improved comfort, or any other competing commercial objective (unless, as is happily sometimes the case, a development which delivers greater safety also delivers other benefits too). In order to make such a judgement, it is necessary to be able to quantify the costs and the benefits on a common scale which means, in practice, to price them.

On the cost side of the equation, this is or ought to be a relatively uncontentious process. A mature industry should have well-developed methods for calculating the costs of introducing new equipment, with higher performance characteristics, and/or of changing working practices and procedures. The railways do not have an unblemished record in estimating such costs accurately, but guidance from the Office of Rail Regulation is available on how this should be done. I am well aware that there is plenty of scope here for debate among economic analysts about amortisation periods and discount rates, not to mention the question of whether to use unit costs or social costs (i.e. whether or not taxation is included), but I hope that most of you will forgive me if we dont embark on those tonight.

To my mind, the greater intellectual challenges lie in calculating the benefits in order to determine whether or not they are sufficient to justify incurring the costs. Some of the benefits of improved safety accrue from avoiding disruption to the network, damage to equipment, mobilisation of the emergency services, and (perhaps) the medical expenses incurred in caring for the victims of accidents, the insurance costs of compensating them, and the legal and administrative costs of accident inquiries. But here the law of diminishing returns applies. If further marginal increments in the railways safety performance can only be achieved at ever-greater cost, it is often difficult to justify incurring this purely on the basis of these directly measurable benefits (i.e. the accident costs avoided) alone. Most of the benefit of risk-reduction comes in the form of casualties prevented, and to quantify this, it is necessary to place a monetary value on the fatalities and injuries which would otherwise be caused.

This is a two-part process. The first part involves estimating the scale of the risk in question. Techniques of quantified risk assessment have advanced rapidly in the British rail industry in the past decade. The Rail Safety & Standards Board now has a well-developed and internationally respected model which identifies all of the hazardous events to which railway operations give rise, and uses a combination of the statistical record and expert judgement to assess the probability of their occurrence and the likely scale of their consequences. Excluding suicides, which are numerous and a serious problem for the industry, but are non-accidental and outside the scope of this talk, the model currently predicts the average annual fatality risk to be:







General public






[RSSB, 2006]

It is noteworthy that more than three-quarters of all of the fatalities are suffered by people who are neither travelling nor working on the railway. Most are trespassers, but some are motorists - or pedestrians - using level crossings. Passenger fatalities occur mainly at stations, not on trains, while workforce fatalities are mostly suffered on or near the track.

But (happily) most casualties are not fatal. To take proper account of the total quantum of risk, such a model must also incorporate injuries. This presents a number of problems. Not all injuries sustained on the railway are necessarily reported to (or recorded by) the industry, e.g. some of those suffered by trespassers, or by passengers who are the victims of assault. Some are psychological rather than physical (shock or trauma), and harder to identify and classify. Some injuries result in permanent damage while others are only temporarily disabling. Nevertheless, enough is known about the injury toll to make it possible to include an estimate of it in the model. Injuries are categorised as either major or minor, and until yesterday one fatality was equated to ten of the former and 200 of the latter, allowing fatal and non-fatal injuries can be combined to give a single measure of risk. Using these weighting factors, the model showed the current average annual risk profile (measured in "fatalities and weighted injuries", or FWIs) to be:







General public






[RSSB, 2006]

Using this measure, total railway risk can be seen to be much more evenly divided between the three categories of victim than risk of death alone. The passengers share is inflated by the large number of slips, trips and falls which they suffer on platforms and stairs, in which alcohol often plays a contributory part. The workforce share is inflated by the usual range of minor industrial injuries and (regrettably) by the assaults to which some staff, such as ticket inspectors, are vulnerable.

I should mention, in passing, that these figures relate only to the main line rail network, and dont include the metro, light rail and heritage sectors of the industry. But these are also subject to the same regulatory regime, and must approach these issues in the same way, although their risk profiles differ in detail. The London Underground, for example, is comparatively untroubled by trespassers or misuse of level crossings (it only has one of those, which is a footpath crossing), although unsurprisingly it does suffer many more injuries occurring on escalators.

I said that the injury weightings applied until yesterday, because from today following extensive research - the Rail Safety & Standards Board (RSSB) has introduced a new classification. Fatalities are still equated to 10 major injuries, but the category of minor injuries is sub-divided. Reportable minor injuries and class 1 shock or trauma are still weighted at 200 to a fatality, but non-reportable minor injuries (that is to say, those that are not required to be reported to the safety regulator under the RIDDOR regulations) and class 2 shock or trauma are devalued to 1000 to a fatality. Class 1 shock or trauma relates to witnessing a fatality or a train collision, derailment or fire, while class 2 shock and trauma includes everything else, such as verbal assaults, witnessing physical assaults, or witnessing non-fatal accidents and incidents. The effect will be to reduce the overall level of risk by around 40 FWIs a year, almost entirely from the shares assigned to passengers and staff. That's an overnight reduction of roughly 20%, no mean feat when you recall that last years white paper sought a reduction of only 3% between now and 2014!

[RSSB 2008]

But whatever the weightings, once the distribution, scale and cause of casualties (FWIs) is known, the second part of the process of quantifying the benefit side of the equation comes into play. This entails placing a monetary value on the reduction in their number which would be secured from any measure taken to mitigate a particular risk. Cold-blooded though this approach may appear to some, to coin a phrase there is no alternative if the industrys resources are to be utilised in such a way as to maximise the return in the form of enhanced safety accruing from such expenditure. So it is necessary to arrive at a benchmark value for each FWI which can be averted.

There are a number of sources from which such figures can be sought. For example, insurance policies contain scales of compensation paid for various types of personal injury. Or courts award damages to victims of accidents caused by the negligence of other people. But neither of these is wholly satisfactory for this purpose. Insurance payouts are capped according to the scale of premium charged, and court awards are usually designed to compensate survivors for the economic loss arising from an injury so that more is likely to be paid where the victim is a young person in work with several dependents than where he/she is an retired elderly person with none.

Instead, the appraisal of investment in transport safety improvements is usually based on a "value of preventing a fatality (or weighted injuries)" (VPF) derived from research into peoples "willingness to pay" (WTP). There is a substantial body of literature describing the methodology employed, and it is not necessary to rehearse the details here. In essence, it involves asking a large sample of respondents how much extra they would be prepared individually to pay for safety measures offering a small reduction in risk (e.g. eliminating a one-in-a-million chance of being killed in the course of a journey). Aggregating the answers over large groups of people makes it possible to arrive at a mean VPF. The Department for Transport publishes such a value for use by highway authorities in calculating the benefit of road safety projects, and revises it periodically. The current figure is about 1.6 million, of which about 1 million arises from these rather intangible "human costs". [Department for Transport, 2007]

Assuming for the moment in the absence of hard evidence to the contrary that the VPF for the railway is the same as that for the roads, i.e. that rail users lives are worth the same as those of road users, it becomes possible to decide whether or not a particular safety improvement offers acceptable value for money. A scheme which offers the realistic prospect of reducing the FWI toll by ten at a cost of less than 10 x 1.6 million, i.e. 16 million, will have a positive rate of return and will clearly be justified. Given the inevitable element of uncertainty about such predictions, and the desirability of incorporating a margin of error (of, say, +50%) on the side of enhanced safety, a scheme offering the same level of safety improvement at a cost of up to (say) 24 million may win approval. But if the cost per FWI averted is two or more times the VPF, then it is likely to represent poor value for money, and the disproportion may in the eyes of the law be "gross" and therefore unjustified.

Provided that the scale of a particular risk and the costs of averting it have been properly assessed, then a decision not to proceed with a particular control measure on the grounds that the safety benefits could only be achieved at grossly disproportionate cost becomes tenable and legitimate. If an accident subsequently occurs, which this measure (if implemented) would have averted, the railway operator has a legal defence against any charge of having failed in its duty to control the risk at a level which is ALARP. Of course, over time the risk profile is likely to change, as are the costs of possible preventative measures. So safety management is a dynamic process, and it is necessary periodically to revisit the options and the costs in order to ensure that their relativities have not changed sufficiently for a different conclusion to emerge. The change, if any is indicated, may be in either direction. In other words, additional controls that were not previously justified may now have become so, if risk has grown or costs diminished. But it is equally possible that control measures that were previously justified will have ceased to be so, because the risks against which they were designed to offer protection have declined or the costs of mitigating them have grown. Such a situation is sometimes called "reverse ALARP", and presents a cost-conscious rail operator with the challenge of persuading the regulator that an existing safety measure can be removed. Operators are understandably cautious about attempting this, but there are instances in which it has been done. London Underground, for example, persuaded both the fire brigade and the safety regulator that it was right to remove fire extinguishers from the passenger compartments of some of its trains, because experience showed that they did more harm than good.

Much of what I have said so far is likely to be familiar intellectual territory for any safety professionals amongst you, and I am grateful for your tolerance in allowing me to trudge laboriously across ground over which you speed effortlessly in the course of your professional lives. My excuse for wearying you with this is that unfortunately the process is not as simple as I have so far described it, and it has been necessary to paint the background scenery in order to contextualise a discussion of the problem. Because, in this country at least, railways are not just another industry. Roughly half their total income is provided by the state rather than charged directly to their users, and the government takes a close interest in their affairs. Indeed, virtually all passenger trains are run by companies which have won time-limited operating franchises issued by the Department for Transport, and all aspects of railway policy and performance (including safety) are subjected to constant parliamentary (and media) scrutiny. The Government has recently published a white paper on their future, an accolade of uncertain value and one with which few other industrial sectors are honoured. This may be irritating to railway managers, but it is a fact of life with which they are compelled to cope.

The rail industry must therefore take due account in its decision-making of political susceptibilities, which are informed by the concept of "societal concern". A definition of this has been offered by the Health and Safety Executive, the public agency which oversees and enforces occupational health and safety legislation. Societal concerns, it says in its policy statement on Reducing risks, protecting people are

"the risks or threats from hazards which impact on society and which, if realised, could have adverse repercussions for the institutions responsible for putting in place the provisions and arrangements for protecting people, e.g. parliament or the government of the day. This type of concern is often associated with hazards that give rise to risks which, were they to materialise, could provoke a social-political response, e.g. risk of events causing widespread or large-scale detriment or the occurrence of multiple fatalities in a single event. Typical examples relate to nuclear power generation, railway travel, or the genetic modification of organisms."

[HSE, 2001]

To railway professionals it may seem unfair to be told that the potential political fallout from railway accidents is comparable to that which might arise from a leak of radiation from a nuclear power station or the escape from a laboratory of organisms which are capable of causing genetic damage to the human species. But it has certainly been the case in Britain in recent years that when multiple-fatality accidents have occurred on the railway (fortunately a rare event), they have excited political and media interest far in excess of that aroused by the continuous toll of death and injury on the roads.

This preoccupation is not a peculiarity unique to our political class. The European Union has recently promulgated a directive on rail safety setting out requirements for (inter alia) accident reporting and investigation, and the safety authorisation of operators, which is binding upon all of its member states. The directive requires that

"Member states shall ensure that railway safety is generally maintained and, where reasonably practicable, continuously improved giving priority to the prevention of serious accidents."

[European Union, 2004]

The term "reasonably practicable" is one which, for the reasons I have sought to explain, will have a familiar ring to the British railway industry. The unstated assumption underlying the emphasis on "serious accidents" seems to be that it is for some reason more important to seek to prevent (say) ten fatalities occurring in a single event than the same number of fatalities in ten separate events. Why this should be so is not explained, and I know of no ethicist who has been able to offer a justification for it. But if there is this bias in public policy objectives it has significant implications for the railways because the accident profile of different transport modes is quite distinctive.

A study published a few years ago by Professor Andrew Evans of Imperial College reported that in the 35-year period up to 2001, there were 178,226 people killed in road accidents in Britain, compared with 270 in train accidents. [Evans, 2003]

Now, of course, the volume of traffic by road was greater, and the figure includes pedestrians, but when normalised against distance travelled, the roads were several times more dangerous. By any objective measure, it is road safety which should give rise to the greatest "societal concern". However, of these fatalities, 85% of those which occurred in train accidents were in multi-fatality events, whereas on the roads this was true of 13% of deaths. Fatal train accidents are far fewer than those on the roads, but when they do occur, the casualty toll is generally greater. It seems to be this which makes them less commonplace and more newsworthy which in turn excites a political reaction.

This phenomenon was well described some years ago in a paper by June Bridgeman, who was then the Director of Road and Vehicle Safety at the Department for Transport :

"There is a far higher awareness of accidents on public transport than on private transport, and an imperfect appreciation by individuals of the relative risks from other situations. Where people place themselves in the care of operators, they expect a higher standard than they are prepared to accept for situations in which they regard themselves as having personal control. Public concern stemming from recent major disasters and associated public inquiries held to establish cause, have generated large scale demands for investment and other measures with considerable cost implications.

"Catastrophes hold a particular grip on public imagination, totally out of proportion to the risks involved Today people no longer believe in fate, or the wrath of God, but want to know the human errors involved. They want to apportion blame beyond the individual to the liability of an employer or an institution .. who ought to have foreseen and prevented the errors. And they want to be compensated in financial terms for pain, grief and suffering as well as economic loss."

[Bridgeman, c1989]

Most people in this audience would immediately recognise a list consisting of Ypres, Vimy Ridge, the Marne, Mons, Verdun, Passchendaele, the Somme, Jutland and Gallipoli as being major battles of the First World War. These names are infamous because the casualties which were caused at them were numbered in hundreds of thousands (there were more than half a million casualties at Passchendaele alone). But most of you would probably also recognise a list consisting of Southall, Ladbroke Grove, Hatfield, Potters Bar, Great Heck, and Ufton Nervet as being the locations of all of the events giving rise to multiple passenger fatalities that have occurred on Britain's railways in the past decade. The fact that the total death toll in all of these railway incidents taken together was less than the toll on Britain's roads each week merely helps to illustrate the double standards which seem to be in force.

It is a compliment to the railways that their superior safety record makes lapses from it, when exceptionally they do occur, the more remarkable and therefore the more likely to attract comment. But it is a backhanded compliment, because such comments are likely to be critical, and to be accompanied by calls for remedial action. Such action will have a cost and if the railways are expected to bear it, they will be placed at a commercial disadvantage relative to their competitors. Is this just? And if they are already meeting the ALARP test set by law, what obligation are they under to heed such calls and take such action?

This inconsistency in the level of importance ascribed to safety lapses by different modes was nicely illustrated by two items which I found in heavyweight (i.e. non-tabloid) newspapers.

One read:

"Four people died in an accident that closed a section of the M25 yesterday A lorry collided with a Volkswagen car, police said. The lorry driver suffered chest injuries and was taken to hospital The M25 was closed at junction six anticlockwise, causing long tailbacks. Crash investigators carried out checks on the vehicles involved and examined the road surface for damage."

[The Guardian, 2006]

You will note that the effects of this accident on traffic flow seem to be almost as newsworthy as the fact that four people lost their lives. The other news story read

"The televised adventures of Thomas the Tank Engine feature too many crashes and could be leaving children terrified of going on a train, a psychologist said yesterday. The little engine and his steam-driven friends have been entertaining children on television for almost 20 years. But according to a psychology lecturer at Exeter University, the sheer volume of accidents in the TV programmes could have an adverse effect on young minds.

"The Independent Television Commissions expert on how children react to programmes said Thomas the Tank Engine is aimed at a pre-school audience who tend to be more likely to see the programme as reality. They haven't learnt to disconnect what they see around them from what they see on television"

[The Times, 2003]

You may think that as it is highly improbable that these children will often encounter steam engines in their lives (unless world oil depletion results in a reversion to the technology of Watt, Trevithick and Stephenson), they are unlikely to relate the adventures of a collection of anthropomorphic locomotives to their own journey experiences. But the main point I'm trying to convey is the relative seriousness with which these two stories were perceived to deserve to be treated. The first occupied seven column centimetres on page ten of the paper. The second was given 51 column centimetres (including a picture of the fictional engine in question) on page one. What subconscious factors caused the editors to deem the second story to merit seven times as much space as the first, and ten times more prominence? What does this say about the relative level of interest by their readers in road and rail safety?

If it is true that large-scale accidents arouse more societal concern than small ones, you might imagine that when they occur, they would have the detectable consequence of deterring fearful passengers from travelling by rail. But unlike terrorist bomb attacks, and concerns about personal security, i.e. the risk of becoming the victim of crime in the course of a journey there is little or no hard evidence of any such effect. The only multi-fatality rail disaster that does seem temporarily to have suppressed demand was the fire at Kings Cross Underground station, and not a train accident. Equally, you might imagine that a higher level of aversion to such events (the so-called scale effect) would be reflected in a greater willingness to pay for their prevention. But, again, the evidence does not support this. When asked by researchers whether a given sum of money should be spent on preventing (say) five fatalities in one accident or ten fatalities in ten accidents, members of the public invariably opt for the latter.

Until recently, the rail industry used a "multi-fatality multiplier" to allow for this scale effect in its cost:benefit appraisal of safety improvement schemes. But it has now been actively discouraged from doing so by the Treasury, because there is nothing to show that this reflects the revealed preferences of its users. [HM Treasury, 1996] What the public seem to be saying is that they wish to see such accidents prevented or at least reduced, and that they expect the industry to do so, but that they are not willing to pay more, either through fares or through taxes, to achieve this result. But how else they believe such improvements should be funded is an unanswered question.

In a paper on Societal Risks commissioned by the Health and Safety Executive, two academics (David Ball and Peter Floyd) alluded tactfully to another factor which may help to explain the industry's preoccupation with this issue.

"There is very little evidence for differential risk aversion by the public where this is based on numbers of fatalities. However, there is a hint that elected officials, senior administrators and risk managers may be averse to major accidents Whether this is attributable to enlightened self interest or to a professional assessment of the wider consequences of such events for society is less easy to discern."

[Ball and Floyd, 1998]

To complicate things further, the scale of the outcome of an accident is not the only characteristic which may make it necessary to modify the application of the willingness-to-pay approach in its simple form. Recent research has shown that there are a number of other contextual factors which can influence the publics perception of the seriousness of lapses in the railways safety performance, and thus societys judgement of the competence of the industrys leaders. These include:

The "helplessness" effect. There appears to be greater aversion to events over which the victims perceive themselves to have no control, which is typically the case in a train accident. Car drivers are willing to take greater risks because they believe that they are capable of taking avoiding action, even though this is frequently an illusion.

The "culpability" effect. Some accidents are rightly or wrongly seen as "freak", i.e. as being wholly random and unpreventable by anyone. Others are regarded as foreseeable, and within the control of an identifiable authority. When the latter occur, blame is ascribed to those who are held to be responsible. Railways are seen as a closed system, with an identifiable command structure, and therefore offer accident victims (or those who purport to speak for them) a specific target where culpability is alleged to reside.

The "mistrust" effect. This stems from an inherent lack of confidence in the integrity and competence of the industrys leaders, leading to a belief that they cannot be trusted to ensure the safety of their passengers. As a rule, this mistrust is placed entirely on senior managers, while the professionalism of (e.g.) train drivers is held in the highest regard.

[Wolff, 2002, Risk Solutions, 2006]

None of these effects is amenable to easy quantification but if the industry is to be responsive to societal concern, it cannot disregard them, and it must find some way of taking them systematically into account in prioritising its safety programmes and objectives. This may lead it into situations in which it has to make difficult judgements of a quasi-moral nature. As already indicated, a large proportion of risk to the public is faced by trespassers, who have no legitimate (or at least no legal) right to be on railway premises. And a large proportion of risk to passengers is faced by those with what is delicately referred to as "impaired competence" i.e. they are under the influence of alcohol or other drugs. There is little or no public concern at their fate. Does this entitle the railway to be indifferent to it too?

Whether it deserves it or not, the rail industry's safety performance has demonstrably been an issue of public interest and, in some quarters, of public disquiet. At the Ufton Nervet inquest last October, there was a determined attempt by the relatives of some of the deceased victims to put the rail industry on trial for its alleged unconcern at preventing their fate, even though this disaster was directly caused by the homicidal behaviour of a car driver and was not, strictly speaking, an accident. But the evidence suggests that most members of the public have at best a sketchy understanding of the industry's true safety performance.

For example, a survey in 2003 revealed that passengers believed that the most frequent "incidents and accidents" occurring on the system were derailments, collisions, landslides, falls from carriages, and injuries due to overcrowding. [Rhind and Thomas, 2003] In fact, these are all infrequent and together account for less than one tenth of the risk on the railway. Instances of the last two categories are virtually nonexistent. In another recent survey, when people were asked to name the mode of transport in which they believed that travellers are least likely to be injured in an accident, more than twice as many nominated cars as nominated trains, even though the opposite is true by a wide margin. [CfIT, 2001] And when a motor insurance company asked drivers whether they felt safer in cars or in trains, cars won by a similar two-to-one margin. [Green Flag, 2006]. So the rail industry clearly has a major task before it, if it is to win the publics hearts and minds, or simply the publics serious attention to the facts.

The most systematic measure of public attitudes to safety on the railways can be found in a MORI tracking study commissioned by RSSB, and I can offer you a few of its headline results from 2006, the latest set of findings to have appeared. The respondents sampled included both users and non-users of the railways, with only 7% using them at least once a week, which reflects the national average. Unprompted, only 10% offered prevention of accidents as one of the industrys two or three priorities for action, putting it eighth on a list headed by reducing fares. 66% thought theyd be killed or seriously injured if they were in a train accident, compared with only 28% in a car accident where they were the driver. When told that on average 7 people a year are killed in train accidents , 56% expected the figure to be higher. 53% thought that privatisation had increased the risk of railway accidents (which is simply not true, and even cynics must recognise that accidents are unlikely to be good for profits).

The net proportion who thought that the motor industry is good at implementing safety measures was 63%, compared with only 41% who thought that of the railways. When asked about what should be the railways top priority for safety investment, the largest vote (33%) went to preventing a child from being fatally injured whilst playing on the line, compared with only 19% to preventing a train collision resulting in a passenger being fatally injured. Quite how an operational railway line could ever be made safe to play on is an interesting question the surveyors unfortunately did not explore. 69% thought that the rail industry should spend whatever it takes to eliminate all risks of accident on the railways. When asked how this should be paid for, 46% said the Government and 42% said reduced profits of train companies compared with 7% who said higher taxes and 7% who said higher fares. Its not clear where they thought the Government gets its money from other than from taxes, and given this result you may be worried that 55% agreed that the public should be involved in deciding how the industry should invest in rail safety.

Encouragingly, when asked about the best way of spending public money to reduce death rates generally, 17% said road safety compared with 4% who said rail safety. 54% said the rail industry only listens to public opinion when forced to do so by pressure groups (which presumably means theres a continuing role for Passenger Focus), and 59% said that the industry takes too little account of the opinions of the public (1% of the public said it takes too much account of its opinions). 84% trust train drivers to keep them safe while travelling, but only 45% trust the people running the railways. Net trust in news programmes as a source of reliable information about railway safety runs at 77%, net trust in the railway industry for this purpose at 14%, and net trust in the government at 33%. Theres a lot more in this study which deserves attention, and I commend it to you.

[RSSB, 2006b]

But, does any of this matter? I think it does because, as a passenger, I have a personal stake in the industrys success. The Chair of the National Consumer Council (Deirdre Hutton) gave evidence to the public inquiry into safety regulation and assurance in the rail industry which took place after the Southall and Ladbroke Grove collisions. She said :

"Public perception is a crucial factor in the acceptability or otherwise of regulatory systems. In the UK now there have been a number of examples where the public have ceased to trust the system in place and what is obvious is that public displeasure, which is often slow to gather pace, can have serious consequences and is powerful enough to drive change.

"Two areas, food and financial services, illustrate the power of public perception in bringing about change when trust has been lost.

[This was a reference to the then-recent outbreak of "mad cow disease" and to false claims made by firms selling personal pension plans.]

The key factors were common to both areas and included:

[Hutton, 2000]

Every one of these characteristics can or could until recently - plausibly be held to be shared by the rail industry too. What action did Ms Hutton recommend?

"The approach of the regulator has to be underpinned by a general compact with the public about the level of safety which is required and of risk which is acceptable. This will not remain static and will change over time or in the light of experience. It would be important to remember that the public do not approach statistics in the same way as scientists or economists : the 1 in 100,000 is potentially a brother or child or parent.

"This kind of discussion can only be held in language that the public might use in everyday life This also suggests that the regulator must on a continuing basis have an easy and effective method of ensuring that the voice of the public in putting their case is heard every bit as clearly as the various voices of the industry."

[Hutton, 2000]

Mr Hutton’s remarks were directed at the industry's safety regulator. But I believe that they are no less relevant to the industrys decision-makers themselves. It should not simply be left to a regulator to impose standards or set targets. A regulator can ask questions, give guidance and when necessary apply sanctions, but ultimately only the railway can take ownership of these problems and deliver solutions to them. A mature, self-confident and competently managed industry should be capable of doing so, and of demonstrating that it has appropriate systems in place for this purpose. It cannot ignore concern about its performance evident in society at large, and it must find effective means of engaging in a dialogue with its users and the wider public about its safety record and its strategy for improvement. The danger it faces, if it fails to do this successfully, is that political demands may be made of it which do not properly reflect the true pattern of risk, and that it will incur opportunity costs which prevent it from delivering either the most cost-effective quantum of safety, or other service enhancements which rank higher amongst its users priorities, or both.

My purpose in coming here is really to learn, rather than to lecture. The railway industry cannot be alone in grappling with these difficult managerial challenges. So, please tell me, how are they handled in other transport sectors? What guidance can they offer the railways in seeking solutions? How do they develop this dialogue? Who should the railways debating partners be? Which media should they be using? How should they be framing the questions? What is society entitled to expect of them? Does society really know? If it does, are its expectations credible and deliverable? If it does not, how can they help it to come to a coherent understanding of the options and the arguments? What, in a nutshell, does the test of "reasonable practicability" really mean? And how can the railways be sure that theyve passed it?


Ball and Floyd, 1998 : "Societal Risks", David J Ball and Peter J Floyd for the Health & Safety Executive

Bridgeman, c1989 : "Three million dead and injured", June Bridgeman, paper for seminar on bus and coach safety

CfIT, 2001 : "Public Attitudes to Transport in England 2001", Commission for Integrated Transport

Department for Transport, 2007 : "Highway Economics Note No 1, 2005"

European Union, 2004 : "Directive 2004/49/EC of the European Parliament and of the Council of 29 April 2004 on safety on the Communitys railways"

Evans, 2003 : "Transport fatal accidents and FN-curves : 1967-2001", Andrew W Evans for the Health & Safety Executive

Green Flag, 2006 : "Drivers dupe themselves into feeling safest at the wheel but are terrified as passengers", Green Flag Motoring Assistance press release 22.5.06

HSE, 2001 : "Reducing Risks, Protecting People", Health & Safety Executive

HM Treasury, 1996 : "The Setting of Safety Standards - a report by an interdepartmental group and external advisers"

Hutton, 2000 : "Ladbroke Grove Rail Inquiry Part 2", witness statement by Deirdre Hutton for Health & Safety Commission

RSSB, 2006a : "Profile of safety risk on the GB mainline railway, Issue 5", Rail Safety & Standards Board

RSSB, 2006b : "Public Attitudes to Safety on the Railways, General Public Research", Ipsos MORI for the Rail Safety & Standards Board

RSSB, 2008 : "Proposals for the weighting of Major and Minor injuries", Rail Safety & Standards Board

Rhind and Thomas, 2003 : "Passenger Surveys", Daniel Rhind and Lauren Thomas, Cranfield University

Risk Solutions, 2006 : "Modelling Societal Concerns", Risk Solutions for the Rail Safety & Standards Board

The Guardian, 2006 : "Four killed as lorry and car collide on M25", The Guardian, 9.8.06

The Times, 2003 : "Thomas the crash engine leaves children fearful", The Times, 11.3.03

Wolff, 2002 : "Railway Safety and the Ethics of the Tolerability of Risk", Jonathan Wolff for Railway Safety