The European Commission has determined that common regulation of safety and standards is to be a crucial element in its drive to create a single market in rail transport. A draft Directive on Rail Safety is expected this year, which will have the explicit objectives of promoting open access and interoperability, as well as achieving fewer deaths and injuries

BYLINE: Michael Spackman

National Economic Research Associates

OUR STARTING POINT for this detailed examination of the way rail safety is regulated was the EC’s remit to ’assess to what extent different regulations or practices have an impact on railway interoperability, in particular on cross-border traffic and on the railway supply market.’

The team was asked to review existing rail safety regulation regimes (RSRRs) and liability standards, assess the impact of differences, and make ’recommendations for a common safety approach to rail transport within the EU’. We were not asked to compare accident rates, although one important conclusion was that reliable comparisons are impossible with currently available data.

National regulation

Despite EU Directive 91/440, which requires the separation of infrastructure from train operations, in most countries the integrated nationalised railway industry (NI) remains the centrepiece. Where there is an infrastructure manager (IM), functions were often devolved back to the NI.

Safety regulation lay formally with a railway inspectorate (RI) or the ministry, but the resources and the initiative remained with the NI. Even so, there were big differences in the distribution of responsibilities among operators, regulators and other institutions.

Only in Finland, Sweden, Denmark, the Netherlands, Germany and Great Britain were active markets in train services encouraged. Here, the IM was usually perceived as having an inherent system safety role including management of the interaction with train operators (TOs).

Some IMs had been drawn closer to government than the integrated NI was. The RI and IM might be merged, or the IM’s original functions of traffic management, safety regulation and infrastructure management separated. Only in Great Britain had the IM become a regulated, private sector monopoly, with self-regulation responsibilities for its own activities as well as those of TOs imposed as a deliberate decision rather than following automatically from its public sector status.

On the contentious issue of ’profits versus safety’ our report concludes that commercial incentives are as likely to be good as bad for the level of safety - given, crucially, a sound regulatory regime. However, it was apparent that commercial pressures have created incentives to shift costs and accountability to others. It was also noted that that some forms of subsidy, such as cost-plus provision of capital, can be bad for the efficient provision of safety.

The national RSRRs were compared in many dimensions, including separation of functions, punitiveness and the use of analysis.

Separation of functions varied widely. External regulatory control of the four key functions of specification, compliance, investigation and prosecution were all carried out by one independent national body in four member states, and by two bodies with differing combinations of functions elsewhere.

The most significant difference was the independence in some countries of accident investigation. The Report recommends independent accident investigation, without going so far as to endorse fully-fledged transport safety boards as found in the US, the Netherlands and some Scandinavian countries.

Punitiveness towards organisations was strongest in Great Britain - a trend which appears to be continuing even further this year. Some other countries were more punitive to individual employees. Punitive measures were virtually unknown in the Scandinavian countries or the Netherlands.

We found no suggestion that greater punitiveness improves the efficiency or effectiveness of an RSRR and noted that, to the extent that punitiveness encourages secretiveness, it can reduce safety.

Investigation of the use of analysis covered questions such as the explicitness of cost-benefit trade-offs. Political pressures to deny any trade-off were omnipresent, but handled in different ways both in law and in public and professional debate.

Explicit valuation of safety benefits was most widely applied in Great Britain and also much debated in Scandinavia. The trade-off may be acknowledged less formally, as in the Netherlands, or publicly denied as in most other countries - despite the fact that in about half of member states costs were formally admissible in defining legal safety obligations.

European regulation

Standards negotiated through the UIC are a major component of acceptance criteria throughout Europe. Railway restructuring is already changing the framework for liability insurance requirements, and there are signs of the international insurance market developing as interoperability progresses. Reinforcing these self-regulated and free market developments, EU legislation is becoming increasingly important, driven by single market policies on procurement and by the Interoperability Directives.

The EC’s High Speed Interoperability Directive (96/48) is in force, and the potentially far-reaching Conventional Interoperability Directive was issued last year in draft. Technical standardisation under the High Speed directive is being led by the AEIF (Association Européenne pour l’Interoperabilité Ferroviaire) representing suppliers and operators. AEIF develops Technical Specifications for Interoperability (TSIs), which are translated into operational standards by the European standards bodies (CEN, CENELEC, ETSI).

Our Report welcomes the establishment of AEIF, but expresses strong concerns about management of the large and complex administrative structure of the harmonisation process. Although harmonisation of EU railway standards focuses mainly on engineering compatibility, safety is often inseparable.

A theme of the Report is that safety issues are integral with interoperability issues such as command and control. But these EU activities have had very little effect on national safety regulation regimes. The railway industry Europe-wide, including safety regulators, displayed surprisingly little curiosity about likely future European safety requirements.

Impacts of safety regulation

In France, Germany, Italy, the Netherlands, Spain, Sweden and Great Britain, case studies were made of new and cross-border services, of acceptance processes for new traction, and of safety regulation for track workers.

A typical response to new entrants wanting to run trains was to require full acceptance of responsibility for costs involved in persuading the IM that the service meets conditions that might be defined only loosely, including infrastructure data that the IM cannot provide. No new operators suggested this was deliberate discrimination, but most believed that established regulatory regimes were a brake on the development of new markets, and sometimes on technical innovation, to a degree not justified by safety.

While the Report notes with approval that the onus for acceptance now generally lies with the manufacturer, rather than the TO, manufacturers criticised many IMs for lack of clarity about what was required, excessive costs, or refusal to be proactive in seeking solutions.

Examination of 14 cross-border freight and passenger flows revealed, as expected, that most obstacles arise from differences in traction power supplies or signalling, but safety regulation issues were important. Where traction crosses a frontier, the safety norm is ’levelling-up’. Each regime requires all of its conditions to be met, in addition to the conditions required by neighbouring regimes, although many of these standards could be harmonised to give worthwhile savings at minimal engineering costs (panel p490).

The case study of track workers was designed to probe decision rules for trade-offs between safety and other factors, some risk being accepted in all countries. The Netherlands, Great Britain and Spain explicitly accepted the trade-off; France, Italy and Germany had formulae which legitimised a pragmatic solution while denying the acceptance of avoidable risk. Sweden was unable to agree on a response.

Scope and style

The study of safety regulation identified two sets of stylised characteristics. The first separates safety decisions, where acceptance of avoidable risk is denied, from operational decisions; approaches risk assessment in terms of a small range of options; merges political and technical judgement; and reaches decisions by specialists in private. The second considers all factors are together; looks at a wide range of options; distinguishes clearly between political and technical factors; and follows due process.

While all applauded the second in principle, some partners in the study team felt that traditional regulation had tended towards the first style. (Britain had seen a deliberate move by British Rail and then Railtrack from the first style to the second, followed by a swing back in recent years.)

To achieve international harmonisation, the study reviewed many approaches including common standards, mutual recognition, and self-regulation. The conclusion was that no single approach merited a dominant role. Many were needed.

Conclusions and Recommendations

The report’s conclusions follow the four main themes of Due Process, Cost Effectiveness, Information, and Management at the European Level.

Due Process

The shift from monopoly nationalised industries to a dynamic international railway market has serious implications for the management of safety regulation. With much wider competition in equipment supply and train operation, transparency and the other features of due process (clear documentation, consistency, specified substantive decision rules, the provision of reasons for decisions, and rights of independent appeal) become much more important.

Without due process, acceptance procedures, in particular, act as serious obstacles to market entry and to innovation.

The study concluded that due process was already largely applied in Denmark, Sweden, the Netherlands, Germany and Great Britain, and partly so in Finland and Austria, but that it could be extended in all Member States. A universal failing was the lack of specified substantive decision rules.

The Report recommends that the EU, through legislation, should encourage the railway safety regulation regimes of Member States to follow due process.

Cost Effectiveness

A central concern of the Report is the need for the inescapable trade-offs between safety and other aspects of quality of service, such as speed, reliability and cost, to be handled efficiently. This it sees as a precondition for the railway industry to develop, with competitive pricing and without unsustainable demands for public subsidy.

The political difficulties are extreme, and not all project partners agreed on the arguments for valuing safety benefits mainly on the basis of people’s considered preferences about risk.

The Report concludes that that there is scope for a considerable further development of quantitative analysis, and it includes a discussion of individual risk, tolerability limits and societal concerns. It also notes that the AEIF has been mandated to accompany its draft TSIs with, in principle, an assessment of the estimated costs and benefits of the technical solutions examined. In practice, this is not effectively achieved.

It also proposes two areas for research. One is into the methodologies used across European industrial safety regulation. The other is a systematic study of international differences in rail safety regulations or procedures which it may be cost effective to remove or reduce, and of the associated costs and benefits and timescales.

The study’s remit precluded any appraisal of the outputs of the Interoperability Directive processes. However, the authors came to feel some unease about whether these were producing standards which will best serve the long term interests of the industry.

With limited awareness in member states of the processes, and the low profile of cost effectiveness, it was not clear that alternative ERTMS options, for example, had been fully considered. The process seemed to be driven by engineering factors alone, with an implicit presumption that cost effectiveness was the preserve of other activities, such as competition policy. The debate in Britain about TPWS versus ATP was seen in Brussels as belonging to another world.

The Report recommends that EU policy should be designed to steer European railway safety regulation towards stronger analysis of costs and benefits and explicit substantive decision rules.


Safety information flows and availability were poor in many areas. Some member states do not publish reports on major accidents, and information about accident investigations is often not available to all those in Europe to whom it would be useful.

Statistical data lacks standardisation in many areas - for example, in the handling of contractors’ employees. It is also becoming increasingly difficult for the UIC to collect standardised data as the old order of national railway monopolies breaks down. Within national railway industries there is a widespread problem of inadequate knowledge by the IM about the infrastructure for which it is responsible, and the dissemination of data where it is known.

There is very limited internet access to EU working documents; indeed the culture is not sympathetic to exposing documents until they are final. This contrasts with the use of the web by the World Bank, for example.

The Report recommends research into the means of developing safety performance and other safety-related data, and European requirements for the maintenance and provision of infrastructure data to actual and prospective train operators or suppliers.

Management at the European Level

The continuing reform of safety regulation, and of railway regulation generally at the EU level, involves powerful established interests, both within member states (such as the still powerful railway NIs), and centrally (such as the standards institutions). It therefore needs to be strongly managed.

The Report concludes that this needs in particular an industry body, which is much stronger than the present AEIF structure, established to implement the interoperability directives.

It sees many tasks for this body, some extending beyond safety but seen as necessary for the effective development of safety regulation: