TEDIS PROJECT C9

EDI IMPACT: Social & Economic Impact Of Electronic Data Interchange (EDI)

Report prepared in 1994 for European Commission by Ian Graham, University of Edinburgh, Scotland; Prof. Claire Lobet-Maris, FUNDP, Belgium and Dr David Charles, University of Newcastle, England.

The content of this report is the sole responsibility of the authors and does not represent the view of the Commission of the European Communities.


Final Report: Contents

  1. EDI: The Social & Economic Impacts of EDI
  2. Industrial Structures: Networks & Communities
    1. Cost-benefit analysis
      1. Intensity of Transactions
      2. Integration of EDI with internal systems
      3. Elimination of parallel systems
    2. Towards Network Industrial Structure
      1. The hierarchical pattern
      2. Non-hierarchical pattern
    3. Sectoral differences in EDI diffusion
      1. Industrial Concentration
      2. Public Administration
      3. Mediators
    4. Strategic Benefits
  3. Defining Communities: Standards & Telecommunications
    1. Message & Data Standards
    2. Telecommunications
  4. Cohesion & Convergence: Regional & Employment Impacts
    1. Regional Impacts
    2. Direct Employment Impacts
    3. Indirect Employment Impacts
    4. Labour Market Impacts
  5. Policy
    1. Recommendations

EDI: The Social & Economic Impacts of EDI

TEDIS Project C9 was coordinated by the University of Edinburgh and brought together a network of academics from across the European Community to investigate the economic and social impacts of electronic data interchange (EDI). This report summarises the results of this study. It is split into three main sections: an analysis of the economic impacts of EDI on industry, an analysis of the context in which electronic trading is developing, and an assessment of the regional and employment impacts. Finally, the report discusses the implications of EDI for public policy, concluding with recommendations for action by the Commission.

The last ten years has seen a steady growth in the use of EDI within Europe, with organisations using standardised electronic messages transmitted over telecommunications networks to replace conventional paper-based trading. However the pattern of this growth has been uneven, with more rapid take-off in some countries, for example the UK and Netherlands, and also in some sectors, notably in automotive, transport and electronics. There is also evidence that the level of EDI activity is higher within the United States than in Europe.

EDI is a new technology linking information technology across organisational boundaries. However there has been no single technical breakthrough which has suddenly made electronic trading possible. Rather, EDI implementation flows from the marshaling of diverse resources - telecommunications networks, computer software, harmonisation of business practices, standardisation of business data - to form communities trading electronically. To explain these patterns of development and to assess their significance for the future competitiveness of European industry it is necessary to understand the forces shaping the growth of electronic trading, and its economic effects and social impacts.

Electronic trading requires organisations to co-operate, agreeing the form of electronic messages and collectively committing themselves to invest in the technical and organisational adjustments necessary. EDI thus differs from many other information technologies in that organisations cannot implement it in isolation. In practice this requires negotiation among enterprises which may have little history of co-operation. Early implementors of electronic trading stressed its potential as a strategic information technology, used as means of gaining advantage over competitors. However the current rhetoric of EDI increasingly views it within the concept of "partnership" between trading partners. EDI implementations therefore have a symbolic value to user firms, being a result of closer relationships and demonstrating their commitment to a long-term relationship, but not necessarily economically justifiable. At a wider level EDI is also important for the European Community: the growth of pan-European electronic trading is a manifestation of the European Single Market, while conversely the formation of localised communities represents a tangible barrier to a single market. The discussions of the strategic and symbolic importance of EDI have obscured the key questions influencing the development of the technology in the future: what are the economic benefits of trading electronically, to whom do they accrue and what will be the wider social impacts.

One perspective is to view EDI as a socially neutral technology, in which the existing relationships between organisations are "edified"; documents previously transferred by mail being transferred electronically. This narrow vision predicts negligible impacts on industrial structure and modest employment implications as routine clerical operations are eliminated at the interface between firms and in the postal system, but leaving the internal operations of organisations untouched. The alternative paradigm is to view EDI as an enabler of more radical changes in the relationships between organisations. EDI may be viewed as a technology which, by reducing the cost of transferring information, may fundamentally alter industrial structure, for example, by replacing intermediary brokers with an electronic market, or by dispersing operational control through the supply-chain. Interest in this radical conception is closely related to the modish concept of "business process redesign", which focuses on the use of IT internally as a central element in radical change.

In either case, to fully understand the processes by which EDI is evolving it is necessary to widen the analysis from the single organisation to consider how competition and co-operation across networks of organisations is shaping EDI, including users, software developers and telecommunications service suppliers. Standards for electronic messages and data, guidelines for their use and possibly the provision of network services must be agreed across networks of organisations. These networks may be informal or involve the creation of organisations to co-ordinate some of the activities across the network, which we have termed "mediators". The structure and scope of mediators vary, both sectorially and geographically, and are both key factors in determining the rate at which electronic trading spreads through the community and its form. An understanding of the activities of mediators is critical when assessing actions designed to encourage the growth of EDI and when planning actions to encourage the take-up of EDI in regions, as there is a choice between aligning with an existing wider community, attractive where trade is extra-regional, or encouraging the establishment of a local mediator, attractive where trade is largely intra-regional

This report summarises the results of our studies into the industrial impacts of EDI, including the key role of mediators in shaping electronic trading and the impact of electronic trading on competition. The report then considers in more detail the particular impacts and policy issues elicited by EDI in the area of standards and telecommunications. The particular issues related to regional economic development are addressed. The evidence for EDI having direct and indirect impacts on employment are considered. Finally the report considers past policy initiatives and makes recommendations for future Commission actions.

Industrial Structures: Networks & Communities

Cost-benefit analysis

Cost/benefit analysis is a critical issue in EDI diffusion. However hard data on the cost/benefit impacts of EDI is scarce and the picture is affected by many early implementors of EDI implementing EDI as a "strategic investment", where the expected benefits are intangible, for example competitive advantages from improved service. However these benefits accrue to "first-movers" and are short-lived, being eroded as competitors respond. As Scott Morton of MIT noted after their Corporation of the Nineties study: "If creating a business network provides some non-differential benefits, but as others emulate the system the newcomers can establish parity of performance, then we regard such a business network as having a neutral effect", concluding that "with few exceptions the initial justification for IT supported networks seems to have been based on operational cost savings".

There has been a tendency for studies of early EDI implementations to exaggerate the importance of intangible benefits due to the immaturity of the systems studied. "First-movers" gain short-term competitive benefits and the trading communities may not have grown to the size at which operational cost savings become significant. However, qualitative analysis of TEDIS case studies gives some valuable insights into the economic motivations of EDI adopters. First of all, TEDIS cases indicate that EDI is mostly used for operational cost savings through reducing clerical processing of transactions and also as an element in wider supply chain management policies, for example "quick response" or "just-in-time" supply.

The TEDIS case studies also show that these benefits are unequally distributed between EDI partners. These inequalities are explained by intensity of transactions, the integration of EDI with internal systems and the elimination of parallel systems.

Intensity of Transactions

The primary factor in determining the net balance of costs and benefits of EDI is the intensity of transactions. Where levels of transactions are low, EDI is practicable but not economically justifiable because the operational cost saving will not recover the initial investment. Therefore the benefits of EDI are biased towards large organisations involved in information intensive trading relationships and away from small organisations involved in diverse low information intensity relationships. TEDIS cases show that many of these small organisations adopt EDI because of pressure from large trading partners.

As new members join electronic trading communities the number of electronic transactions will rise for existing members, making EDI more cost effective for them. This insight has led to the recognition that there is a "critical mass" or size of community beyond which the benefits to members exceed their costs . Below this "critical mass" the community is not viable and above it will grow as new members join. Although this is a picturesque analogy it overlooks the fact that electronic trading communities are not nuclear weapons, they are social systems. Electronic trading communities often trade at levels at which they are not cost effective but are sustained by the expectation that they will grow to a viable size.

Integration of EDI with internal systems

The second factor in determining the net cost impact of EDI is the level of integration of electronic interchange with internal systems. Many organisations are nominally EDI users but either use manual re-keying of EDI data or crude human mediated file transfers between the EDI gateway and their internal systems. Also the additional information which EDI links can facilitate is not fully utilised, for example detailed stock and shipping information from retailers is simply printed out rather than integrated with suppliers' production planning systems. Organisations, because of the cost of modifying existing systems, frequently postpone fully integrating EDI into their systems until the systems are due for replacement, in which case EDI functions can be justified. This inhibitor to the rapid diffusion of EDI is seen in financial EDI where the banks have found customers unwilling to embark on expensive modification to existing company financial systems, but are able to justify incorporating EDI into new systems.

Elimination of parallel systems

The third factor is the elimination of the need for parallel EDI and paper-based systems. Even where EDI itself offers cost savings, having to maintain a parallel paper-based system operating less efficiently with lower volumes reduces EDI's net benefits. This cost driver leads dominant EDI traders to exert considerable pressure on partners who prefer to trade on paper to switch to EDI, up to the point at which EDI is made a requirement of trade. A second effect of this cost driver is to make EDI more attractive in isolated well-defined communities in which the level of trade outside the community is small, as the prospective of eliminating paper-based trade is more realistic than in communities which will have carry on significant volumes of trade across the community boundary.

Towards Network Industrial Structure

The formation of Pan-European electronic trading communities is a manifestation of the European single market, whilst conversely the formation of localised communities represents a tangible barrier to the single market. To clarify this question we have assessed the TEDIS 1 and 2 cases studies in order to identify the trading patterns for EDI development. These show two main patterns for EDI development: the hierarchical pattern and the non-hierarchical pattern.

The hierarchical pattern

The hierarchical pattern is based upon the hub and spokes structure of trading relationships, with a dominant central organisation largely determining the form of electronic trading with a collection of weaker partners. Its development leads to the formation of a closed EDI community in which the terms of electronic trade are defined. Here we face the problem of the formation of gaps between EDI communities. These gaps must be closed to make the vision of pan -European electronic trading realistic.

Non-hierarchical pattern

The non-hierarchical pattern is based on a non-hierarchical structure of governance of trading relations between partners. The absence of a hierarchical hub capable of defining the terms of electronic trading is one of the start-up problems of EDI in this type of community. So in order to develop EDI, partners will co-operate to found artificial hubs to mediate the interests of community members. These mediating associations are legitimised by the members of the community and become responsible for co-ordinating electronic trading within the community.

The development of these non-hierarchical communities could lead to inter-community co-ordination problems. However the problems seem weaker here because the EDI terms of the community are defined by consensus, that means by all members of the community according to their various interests in other communities. This process of EDI development is slower than the previous one, but can lead to the development of EDI projects based on true community benefits.

Sectorial differences in EDI diffusion

There are obvious variations in the extent to which EDI has been implemented between industrial sectors; from high levels in automotive, electronics and retailing, to much lower levels in public administration and construction. However sectorial analysis of EDI must be approached with caution. With the exception of a very small number of horizontal applications, most EDI is between organisations in vertical trading relationships: between buyers and suppliers. Therefore, to some extent, almost all EDI is intersectorial, whether between manufacturers and retailers or banks and their customers.

Disparities in levels of EDI diffusion between sectors across Europe are largely explained by two factors: the degree of industrial concentration and the co-ordination needs of the sector.

Industrial Concentration

The degree of industrial concentration is generally expressed in terms of the Herfindahl index, the sum of squares of market shares, which is a measure of the extent to which the market is dominated by the largest enterprises. Some sectors, for example automotive, chemical, retail and electronics, are highly concentrated and dominated by a small number of powerful players, whereas other sectors, for example clothing and furniture, have lower levels of concentration. The greater the degree of concentration, the lower will be the barriers to the formation of electronic trading communities for two reasons. First, high concentration decreases the complexity of agreeing the form of electronic trading within the sector, including message standards and data directories. It is easier for a few major players to shape the implementation of EDI in a sector than a plethora of smaller players. Second, high concentration implies the existence of large organisations who not only have most to gain from EDI but also have the influence to coerce trading partners into using it.

However the apparent concentration depends upon the geographical area being considered. As the boundaries of the community are expanded, the sector becomes less concentrated and more difficult to co-ordinate. This finding has implications for the supporting of communities within Europe, for in many industries, such as banking or retailing, the sectors are generally highly concentrated in national markets, and therefore fertile ground for the growth of electronic trading communities, but not across Europe as a whole. This creates a dilemma in the formation of electronic trading communities. The intensity effect described above means that communities are more viable the wider their boundaries, but the problems of co-ordination mean that the initiation of electronic trading communities is easier in highly concentrated localised communities. In practice this problem is overcome by communities starting among small groups of organisations, then expanding by being open to new members to gain the benefits of a wider community.

Public Administration

Links between public administration and commercial organisations, for example taxation, state insurance and statutory company declarations, form a significant special case. On the one hand the government departments responsible represent a monopoly, so at the local level they do not have to co-ordinate the introduction of electronic trading with any other equivalent body. However, public administrations lack the commercial pressures to seek efficiency improvements through IT innovations faced in the private sector. Also political considerations inhibit public bodies from insisting on trading by EDI, so they face the prospect of operating parallel systems for the foreseeable future. In many areas of public administration the scope for governments to share the development costs of EDI is undermined by the lack of standardisation between states, with moves to standardisation requiring major changes in procedures and legislation. It is significant that the most highly developed public administration EDI application across Europe is in customs declarations, where information and processes are largely standardised.

Mediators

Where there is no single organisation influential enough to initiate electronic trading unilaterally there is a need for a mechanism or institution to mediate between organisations, negotiating the form of EDI and co-ordinating its implementation. These mediators derive their legitimacy from the organisations which support them. The most widely known examples of mediators are ODETTE in the automotive sector and Edifice in electronics. The negotiation role involves agreeing the scope of the trading network, which primarily concerns the transactions to be included, the data to be exchanged and the geographical boundaries of the community. The mediator must then marshal and co-ordinate the resources required. Here the primary activities are agreeing message and data standards, ensuring the availability of suitable telecommunications services and software, and providing consultancy and support to members. Ancillary activities include liaison with other mediators and national and international agencies, for example EDIFACT message development groups. Where there is a well-developed market in support activities, the mediator can either leave their provision to the market or sub-contract them to specialist firms, but where there is not a viable external market the mediators have to undertake these activities themselves. Therefore the organisational substance of mediators varies widely. At one extreme, in UK banking IDX, the Interbank Data Exchange, is based on "minimum co-operation, maximum competition", and is simply a committee, a set of standards and a service level agreement. The development of software and the choice of telecommunications networks and standards for use between bank and customer is left to the market, allowing the banks to develop and market EDI services in line with their idiosyncratic strategies.

Successful EDI initiatives are characterised by the strong involvement of future participants of the EDI network, whereas in less successful initiatives the co-ordination efforts are undertaken by third-parties and not controlled by users. Where users do not come together spontaneously, industry associations prove to be very effective foundation. There are two explanations for this. Firstly, if potential users control the co-ordination process the elements of the future EDI network can be adapted to user needs much faster than in a third-party development process, since in the latter case results are checked against user needs only afterwards. Secondly, if companies invest resources in the co-ordination process they are more likely to use the EDI network when set up. Nevertheless, the early involvement of potential users does not guarantee the success of an EDI initiative. Since industry associations are very efficient in co-ordinating the development process their involvement is frequent in EDI initiatives. Under these conditions successful diffusion of EDI takes place only if the institutional structure of the industry prevents the competition of different industry associations. This could be brought about either by a hierarchical structure of industry associations or by a low level of institutional differentiation. In contrast, a highly differentiated institutional structure with several strong industry associations could effectively hinder the diffusion of EDI. A possible explanation for this result is that since industry associations command a high level of legitimacy among their members, the competition of industry associations increases the uncertainty about which group might finally triumph over the others thus, forcing potential users into a wait-and-see position (even if the isolated network would totally fit their needs).

To maintain the legitimacy of the mediator it is important that they are independent and not under the control of any single organisation and also that in their relationships with software suppliers or telecommunications service suppliers they do not cede control of the trading community, for example by allowing a software supplier to own the rights to required software or be tied to a single VAN. These strategic requirements can lead to complex institutional arrangements. For example, the structure of electronic trading in the UK life assurance sector seems at first baroque; with the life assurance companies forming a jointly owned mediator, (Origo), the independent financial advisors who sell the products forming their own mediator (The Large Intermediaries Group), and Origo and AT&T forming a company to provide the physical network (The Exchange). However this structure ensures that no organisation can exploit electronic trading for their own competitive advantage to the detriment of others. The life assurance offices were aware that in the earlier viewdata systems having two network providers offering parallel systems was costly, both in network fees and the costs of interfacing with two systems, but that using a single system owned by a network provider would leave the network provider able to dictate terms. However the mechanism also had to ensure that the expertise of network providers could be drawn on and the needs of intermediaries were built into the process.

Strategic Benefits

Some early implementations of electronic trading were based on its strategic use to gain competitive advantage: a single powerful organisation using proprietary standards and closed networks to tie customers to it and squeeze out competitors. A small number of strategic applications of electronic trading appeared in the US in the 1980s, notably McKesson in pharmaceuticals and AHSC in the hospitals supplies sector, generating a disproportionate amount of interest. They were seen as the vanguard of a wider shift towards the use of IT as a source of strategic advantage rather than just a means of improving operational efficiency.

Membership of an electronic trading community provides efficiency advantages. This evokes competition issues where some organisations are prevented from joining the community due to the cost of participation or the cost of harmonising business practices. Although these barriers are most significant where a single organisation controls the network, similar issues arise where a group of organisations co-operate to form a network and then close it against further entrants or use their control against the interests of users who join the network later.

The potential impact of EDI as a barrier to entry will be increased by the development of more complex inter-organisational linkages requiring greater investment in equipment, training and changes in procedures. This issues are being confronted by defence industry contractors faced by requirements for CALS compliance in US defence contracts. The CALS (Computer-aided Acquisition and Logistic Support) programme was initiated by the US Department of Defense in 1985 to improve logistic support through integrating the data generated during the design, manufacture and use of weapon systems. The programme includes the development of standards for the sharing of technical text and graphics data across networks of organisations. Requirements for CALS compliance by the US Department of Defence places a large cost hurdle in the path of European defence companies seeking DoD contracts. The cost of CALS compliance has been estimated as being in excess of ECU 2 million for a large defence contractor. As CALS and similar non-transactional electronic communication are implemented across a range of sectors, (for example construction, automotive and petrochemicals) many organisations will face a dilemma: investing in the technology to stay in the game or withdrawing from this market. These forms of complex electronic trading will seriously disadvantage small and medium sized enterprises, particularly where they already lack IT infrastructure.

The case of airline reservation systems demonstrates the two key issues in the competition impacts of EDI systems: ownership and openness to new users. In the AHSC, McKesson and Sabre examples competitive pressure opened up the systems to become a market for products beyond the system owner, but ownership of the system still provides advantages and opportunities for distorting the market which are very difficult to regulate.

Electronic commerce can provide the linkages to embed concerted anti-competitive practices. For example, it makes possible the electronic transfer of pricing information or the electronic creation of an artificial market between enterprises These practices are inherently covert and when carried out electronically become even more difficult to detect and regulate. By their very nature these practices are rarely described in the literature, but one case in which it is implied that the users became aware of nature of the system was Prodex in UK petroleum refining. The system was designed to automate oil product exchanges between oil companies, where the decision to adopt a common trading platform between community members rather than establish a centralised clearing house was due to concern that a clearing house would breach UK competition law.

It is now recognised that "strategic information systems" such as AHSC or McKesson were an aberration and that increasingly the development of electronic trading networks will require groups of organisations who have traditionally been in competition to co-operate over the development and co-ordination of electronic commerce. The number of markets in which there is the concentration of sellers and potential strategic advantages to make the sole development of proprietary systems effective is small.

The operational advantages offered by EDI and the investments required to build electronic trading networks mean that EDI will impact upon the nature of competition in markets. The key issue is ensuring that communities are open to late entrants at an equitable cost. The issue of openness is related to the use of open standards and access to any network infrastructure. The development of closed proprietary electronic trading systems for competitive advantage by single enterprises has always been rare, albeit interesting to IT management theorists, and is likely to become rarer. The expected mode by which electronic trading communities form will increasingly be through organisations which compete in the market coming together with groups of their suppliers or customers. However this still affects competition if their association is closed to other competitors.

In policies for developing EDI use there is a clear requirement for ensuring that communities are open and that property rights in the network are not abused to discriminate against users. The commission has recognised the need to avoid to avoid abuses of computer reservation systems for airlines. Similar issues of misuse of data, system bias and cost of access will increasingly occur in a wider range of electronic trading.

Defining Communities: Standards & Telecommunications

In essence "membership" of an electronic trading community is defined by organisations using the same data and message standards and having access to software embodying the standards and the inter-organisational telecommunications connections to allow them to trade.

Message & Data Standards

EDI requires standardisation of information between organisations, even if only implemented between two partners. The standardisation includes both the meaning of the message content and the structure of the message itself. In rudimentary implementations of EDI, the agreement of these could be isolated in a small group of trading entities, with the message content and form mirroring existing paper based transactions. However, the potential benefits of network externalities leads users to open their systems so that they may be used by wider groups of users. The content of messages determines the scope of electronic trading. Any message development process balances the pressures for universal standards with the needs of local trading communities. The CEC has, through its support for the UN/ECE EDIFACT standards process, been a leading promoter of global standards.

The last ten years has seen the EDIFACT message development process grow from an informal body of largely self-appointed participants into a large global bureaucracy. The growing scale of the process and its growing geographical scope led rapidly to difficulties in co-ordination, which manifested themselves in message development almost grinding to a halt with few messages progressing to become United Nations Standard Messages. The underlying difficulties related to conflicts between the needs of participants: between users who wanted stable messages and others who wanted messages developed rapidly to meet business needs; between users who wanted messages for use in closely defined regional communities and users who wanted universal global messages; between users who wanted messages rapidly available to allow the rapid edification of their existing trading links and users who wished to take longer to develop messages which would enable the re-engineering of inter-organisational links. Any standards development process has to reconcile these competing pressures.

The increasing formalisation of the EDIFACT process has led to the development of procedures to co-ordinate activities between continental groups and to more closely define the activities of actors in the process.

The changes in rapporteurs' procedures specified in TRADE/WP.4/R983 are predicated on the belief that standards should be global. R983 states:

"It should be understood that the benefits of having a single international standard outweigh the drawbacks of the occasional compromise."

Having a single international standard imposes costs on users because message functionality is compromised, the time to develop the message is increased and it makes the message development process remote from the user.

The standards being developed have moved beyond a simple automation of paper-based transactions, towards the development of messages which facilitate new models of trade. The creation of messages embodies the mode of their use; the moves within EDIFACT to use formal data modelling techniques is an attempt to align business practices within electronic trading communities.

Opinion on the further development of standard messages is split between people who believe that the emphasis should be on the development of generic messages, in which a large proportion of segments are omitted within specific user-communities, and people who believe that a wider-range of application/community specific messages should be developed. In either case, the development of semi-formal message sub-sets within trading communities is growing.

The emphasis on the use of formalised information modelling is a sign that EDI is moving away from messages which simply embody the process of existing paper transaction, into messages which facilitate more complex innovative inter-organisational integration. As EDI develops into areas which have not already had to be standardised to conform to paper-based inter-organisational transactions, the variety of business practices which the standardisation process has to reconcile across any geographical area will become more diverse, so the difficulties of developing and implementing global standards will increase. The "occasional compromises" will become more frequent.

The publishing of global standard messages does not prevent isolated trading communities evolving their own sub-sets of the global messages. Global standard messages do not necessarily lead to a global trading community. There are sound reasons why regional communities should use the same syntax and data directories, and why global messages may be useful in applications related to inter-continental trade; it is less clear what the benefits are of having a global message where the potential users are trading within isolated "island" communities, even if the process is ostensibly identical.

The Western Eurpean Edifact Board is based on a process of national representation, supplemented by representation from recognised sectoral bodies. This pattern has been reinforced by national 'edifora' taking on the role of channelling their national industrial requirements into the process and nominating national representatives. The structures created to feed national representations into this process act to strengthen an identification with a national EDI community rather than a Europe-wide community. This structure nurtures EDI chauvinism, in which participants identify with their national EDI community rather than a pan-European community. This parochialism leads to the development of national message sub-sets. National EDIFACT sub-sets and the idiosyncratic business processes they embody both represent barriers to the European Single Market. The UK national forum UKCEDIS has been established explicitly to form a central body to feed a coherent view of UK EDI standards requirements into the international standards setting arena. This structure of national representation presupposes that there is a coherent national view and that mechanisms can be determined to find out what it is. More crucially this structure views the standards setting process as a hierarchical process in which requirements are passed up the hierarchy to national, then European and finally global levels, leading to the process alienating most practitioners from the process. For users the process becomes a black box into which requirements are fed and eventually standard messages emerge.

All policy bodies relating to EDI have recognised the importance of open, stable standard messages drawing on published data dictionaries which meet the needs of users. This is the basis for TEDIS support for the Western European EDIFACT process. However the EDIFACT process has to be able to reconcile the conflicting needs of users; not only users who are active within the process but the entire European EDI community. The process has so far failed to satisfy these needs.

The proposal for the message development groups to build data models for input into formal message development techniques implies that message development is a rational scientific process. However participants suggest that the process is not free from organisational politics. The costs of attending all message development group meetings are high which acts as a barrier to many taking a full role and leads to the suspicion that many participants aim to develop products derived from the standards or are there to obstruct developments in other countries.

Consequently some European organisations with an interest in new messages develop them outside the EDIFACT process, albeit still submitting them into the process as a fait accompli. The UK Customs & Excise's developed the SEMDEC message because they did not expect the INSTAT message to be released early enough for software suppliers to have products available prior to the requirement for traders to submit single market trade statistics. In utility billing it was agreed by the potential users that the EDIFACT syntax should be used but that message development should proceed through EAN as this would permit pan-European involvement, but avoid the bureaucracy of development through EDIFACT.

The barriers to the European single market posed by local sub-sets should be targeted by bringing national or sectorial communities developing similar messages into contact. It is crucial for the development of the EDIFACT process that the national agencies actively communicate EDIFACT activity to interested organisations and that the EDIFACT board ensures that activities are communicated to international sectoral bodies. In principle the structures to achieve this exist, but many participants do not view the communicating of information down the hierarchy as their primary role.

Telecommunications

Electronic trading requires telecommunication services linking users. EDI services are still an immature market, particularly in relation to voice telephony, but the policy issues faced in wider telecommunications policy are echoed when considering the market for value-added network services: deregulation, interconnection and security.

TEDIS has launched a series of projects in these fields. These actions are largely concentrated on the "Interconnectivity Platform and One Stop Shopping". TEDIS has supported the establishment of two Interconnectivity Platforms (EuroEDI and Concorde). However, the main problem with interconnection is the need for clear and precise information. It would thus have been relevant to study first the demand side (users) and the supply side (VANS) of interconnection and assess the scale of interconnection problems before designing a global, European interconnectivity platform.

The Registration Authorities project appears to answer a real need because addressing problems are important in some sectors, for example the automotive sector. For intersectorial transactions the problem is even more complex. TEDIS has been aware of users' needs in the telecom area but has not always chosen the most appropriate solutions. However, sometimes they have led a more prospective role, especially with the study of the potential use of ISDN for EDI applications.

Concerning EDI and deregulation, the impact of the telecom deregulation on EDI (i.e. the "Services Directive", the "ONP (Open Network Provision) Framework Directive" and the "ONP Leased Line Directive") is not clear.

Telecommunications deregulation will eventually reduce tariffs of network provision by VANS and public telecommunications operators (due to increased competition and the possibility of leased lines capacity resale), improve the quality level of provided services and networks (due to ONP and increased competition), and favour VANS emergence and the creation of a genuine market between VANS. However national variations in the speed of deregulation impedes the development of EDI.

VANS and interconnection appear to be one of the most important telecom aspects of EDI. Even if EDI is only a part - albeit an increasing part - of VANS activity, VANS may be important in EDI diffusion. As the EDI diffusion rate is different from country to country, the potential of the EDI market varies from country to country. The VANS suppliers are more or less the same in each country but their strategies vary between countries, due to the requirements and the power of the national users, the competition in the telecommunications and VANS markets and the legislation of the country.

The central issue in the creation of a genuine market in VANS is the growth of interconnection. Currently there is not a unified European market in value added network services and the market is still immature. Initial entrants into the market targeted niche markets which state telecommunications operators had either overlooked or ignored. This enabled the VANS to lease telecommunications capacity and for a small capital outlay be profitable with low transmission volumes. By reaching agreements with major users VANS were effectively able to gain monopoly franchises in trading communities. However these local monopolies are generally unstable. Firstly trading companies recognise that the costs of EDI can be reduced by introducing competition between VANS and secondly new entrants to the VANS market will aggressively attack the most lucrative monopolies. It has been claimed that the decision by the UK retailer Argyll to tie in with IBM rather than with the dominant VAN INS was a strategic decision to break INS's monopoly, even though in the short-term it imposed the additional cost of linking into two networks on suppliers. Although this claim implies strategic perspicacity bordering on clairvoyance, the effect of Argyll's decision was to open up UK supermarket retailing to a wider range of VANS. IBM were later joined in the sector by AT&T and BT. To overcome the problems for manufacturers supplying a range of retailers of having to subscribe to a range of VANS, the UK government has supported the foundation of an EDI clearing house allowing suppliers to communicate via all networks through a single link. Once the VANS market in retail grocery involved more than one VAN it became inevitable that retailers would see the potential benefits of trading through more than one network, with the competition making price and service quality critical. Once three service providers were competing directly, interconnection between networks became inevitable, as it is strategically in the interests of the second and third placed VAN to interconnect to attack the dominant VAN. Their interconnection then forces the dominant VAN, possibly reluctantly, to interconnect with the others. One effect of the interconnection between VANs is that it has made the DTI supported clearing house appear redundant. The prime effect of interconnection in the UK is that EDI services have become a commodity with fierce competition on price, subject to achieving acceptable service quality levels. Competitive pressures force the VANs to overcome any technical barriers to interconnection.

These changes in the VAN market pose serious strategic problems for the VANs which entered the market to attack market niches. Having to buy telecommunications capacity off telecommunications networks, who are increasingly becoming their competitors, to provide what is basically a commodity service, they are in the position of a car rental company who are forced to rent their cars from Avis and Hertz. For companies with ownership of telecommunications networks EDI can be very aggressively priced to generate marginal network traffic. In the UK VANs like INS have seen their competitive edge wiped out in twelve months and appear to have redefined their strategy towards greater reliance on their software activities. It is argued that these VANs can develop new value added services. However, it is hard to identify what these services might be. The one niche with potential for growth is in interactive EDI services replacing videotex systems, but to provide real-time services it is of even greater value to have ownership of the network infrastructure than in 'store and forward' conventional EDI. Our prognosis is that the provision of EDI services will become dominated by the international telecommunication operators created in the wake of deregulation, for whom EDI services will be just one small offering in a full portfolio of telecom services, with limited scope for other operators in highly specialised niches.

The speed of these changes depends upon the speed of effective telecommunications deregulation and the formation of genuine competition. In the interim there may be justification for actions to facilitate the provision of telecommunication services more directly. The first solution is to reduce the complexity of this problem by simplifying the existing range of alternatives and making the companies' technical choices more coherent. The second solution is a more general one and consists in encouraging the interconnection of VANs.

The reduction of the complexity of the is often undertaken by the industrial associations influencing their members' choice of VANS. However, this is not the case in all sectors. The role of the industrial associations varies from sector to sector and from country to country. This approach may lead to technical incompatibilities between sectors if the chosen technical solutions are defined on too restricted a sectorial basis.

Interconnection between VANS is problematic. The availability of interconnection is often not clearly explained and the interconnection tariffs and the structure of these tariffs vary between VANS. Moreover, there is a conflict between users' perceptions of current interconnections (Very few existing and working interconnections) and those of VANS (all VANS are providing interconnection). Interconnection is now largely a commercial problem but the commercial aspects are rarely analysed.

Interconnection is not a problem in each country, in every sector or for every company. The importance of interconnection for users depends on several factors including the position of individual companies in relation to suppliers and customers, the industry sector, the degree of experience of EDI, whether EDI is seen as an integral and strategic part of future business methods, and the level of international trade.

When considering the telecommunications aspects of EDI, there are gaps between companies and between sectors. Industrial associations try to reduce the complexity of the problem by providing information to their members and by recommending service suppliers. This is not necessarily the best solution as there is no single solution for users. They have different - and sometimes contradictory - needs. To overcome this problem, some interesting initiatives exist such as the one developed by GTF in France: users will be supplied with a folder describing technical solutions - in terms of telecommunications, standards or VANS - appropriate to their characteristics and needs. Interconnection seems to be a better way to overcome gaps. Concerning interconnection, it is evident that the main obstacles to wide use of interconnection are commercial and legal. However, TEDIS actions relating to interconnection have concentrated on the technical issues. It would be valuable to investigate these commercial and legal issues: responsibility, billing, commercial strategy of VANS. Furthermore, it appears that interconnection agreements between VANS will only succeed if all interested parties are on the same footing and will share equal benefits.

Cohesion & Convergence: Regional & Employment Impacts

The diffusion of EDI has clear potential social impacts. Firstly, that EDI will develop most rapidly in well-defined local communities of large enterprises, raising the fear that EDI will accentuate uneven levels of economic development across Europe, disadvantaging regions dependent upon high levels of SMEs, low levels of IT in organisations and poor telecommunications infrastructure. Also the cost savings from EDI imply labour displacement. The following section considers the regional and employment impacts of EDI and considers whether the issues they raise are sufficiently significant to justify specific policy initiatives.

Regional Impacts

To date there has been very little discussion or research about the effects of EDI on the process of regional development, although some funding for EDI supporting activities has come from regional policy sources at a local or Community level. There is however an assumption that EDI may be able to assist in overcoming problems of peripherality in the less favoured regions of Europe, and that without some form of public action, EDI will be slow in being adopted within those regions.

Whilst there is little literature on EDI and regional development, a considerable body of research examines the role of supply chain developments in the growth and dynamics of regional economies. In particular, recent research has focused on two issues: firstly the performance and dynamics of "industrial districts", typified by local networks of small firms engaged in intense intertrading and information networking, and secondly on the local development prospects for just-in-time based supplier agglomeration around major assembly plants. Such research has countered previous studies on the effects of internationalisation, where branch plants have developed increasingly international sourcing strategies with few local suppliers, the so-called 'cathedrals in the desert' syndrome. The key question in terms of EDI research is whether EDI adoption will affect the nature of supply chain links in ways that will reinforce or weaken these particular trends or models, and if so will regionally differentiated adoption rates disadvantage particular European regions.

In essence there are two main policy issues. Firstly, if there are supply chain effects that disadvantage the peripheral or declining regions of Europe, then what forms of regional intervention are needed to offset the problems and ensure that firms in those regions can make use of EDI to gain better access to European markets? The second issue relates to EDI diffusion policy, and is independent of the first, and poses the question as to the best means of delivering EDI awareness policy, and whether regional initiatives are most successful in reaching their target audience.

The fundamental problem with these questions at present is that the problems identified - supply chain restructuring or lagging diffusion processes - depend to some extent on an existing level of EDI diffusion and its impacts on business relationships. Given the slow and partial adoption in certain countries and sectors across Europe this presents two main problems for analysis. Analysis of regional differentiation of new technology adoption normally requires some significant SME adoption before the regional differences become clear. And in terms of supply chain effects, most users of EDI are still in the early stages of administrative applications and have not begun to use EDI in a strategic way. Consequently the cases where EDI is used to underpin real regional restructuring are few, and so the analysis has to be prospective.

It is also important to recognise that EDI strategies are not implemented in a vacuum, and that EDI is usually a means to an end rather than the key driving force behind supply chain restructuring and any geographical shifts in purchasing. Thus many of the more active EDI users in the manufacturing sector are using EDI as part of a "lean production" strategy. Similarly in retailing EDI is associated with a desire to cut inventory, reduce costs, and strengthen the power of the retailer on the supply chain.

Examples of potential regional consequences of supply chain restructuring, and the application of EDI, can be seen in the car industry where extreme "just-in-time" localisation strategies are being implemented. In a number of new plants, such as Nissan in North East England, and Fiat in Southern Italy, EDI is being used to integrate suppliers located close to the car assembly plant, such that lead times can be measured in minutes rather than days. However the same firms are also developing EDI-based logistics systems that permit longer distance suppliers to achieve just-in-time deliveries several times each day through consolidated deliveries. It is clear that EDI can be used to assist a variety of different strategies, and that it is the existing strengths and weaknesses of different regional industrial competencies that will probably determine the geography of benefits. Thus, although EDI itself does not have a particular spatial bias, it is likely to reinforce the advantages of core regions without countervailing support for the disadvantaged areas.

A second issue is in the public procurement area, where the LFRs, as with all regions, are being required to liberalise their public procurement, where they previously maintained policies favouring local suppliers. Whilst the issues are the same in all regions, the LFRs are most vulnerable to liberalisation as their export record tends to be poor and the competitiveness of local firms also tends to lag. If EDI is to be a requirement for public procurement in some regions then LFR firms will need to be supported in adopting EDI in order to assist them to gain a fair share of public contracts. This will be especially important for the SMEs currently dependent on local public contracts.

A final issue concerns the successful orientation and management of local EDI initiatives. A variety of different models of EDI initiative have been adopted around Europe. These vary in terms of scale, sectoral orientation, implementation organisations, and funding. There is a need to examine carefully the most appropriate organisations for different tasks and especially the interaction between EDI awareness and diffusion policies and those initiatives to support wider technical development in small firms. It is likely that the most appropriate regional organisations for promoting EDI will either be sectoral associations or general purpose technical assistance agencies.

Direct Employment Impacts

The evidence on the direct employment impacts of EDI is largely anecdotal, drawn from case-studies of EDI implementations in individual organisations. One study which directly addressed the organisational impacts was COST 320 ('COST 320: The Impact of EDI on Transport', CEC, 1993). This project studied the impact of EDI on the transport sector in fourteen European countries, collecting data through a literature survey, national reports and a survey of 692 firms involved in transport, including shippers and providers of transport services.

The transport sector provides a good opportunity to gain insights into the organisational impacts of EDI for two reasons: EDI is relatively well developed; and EDI may be expected to have an impact on the complex structure of shippers, agents, stevedores, brokers, forwarders and carriers which has evolved based on paper-based communication, with the possibility of some activities being transferred between organisations or eliminated.

When asked to specify the main implementation problems, the most frequently cited in the COST 320 survey were "technical start-up problems" (57%) and "interfacing EDI software and in-house systems" (36%), with "organisation/procedural start-up problems" only the third most frequently cited (33%). This suggests that the rhetoric of EDI implementation being an organisational rather than technical issue understates the technical issues. Alternatively, it may be that the increasing availability of standard EDI interfaces may obviate the technical problems faced by these early users of EDI.

In the COST 320 study most respondents disagreed with the statement that EDI leads to job losses (61%), however 26% agreed, the majority being large companies and shippers. Only 13% of companies admitted that EDI had led to a reduction in clerical staff, nearly balanced by 10% who said they had to take on staff. However, of the companies who did not admit to making reductions, 20% of the total forecast that they would make reductions in the future and 16% admitted that turnover had increased without the need to engage additional staff. It may be deduced from this that the gross employment impacts of EDI may be disguised by EDI being implemented in companies increasing their market share and also postponed as companies gain sufficient experience of EDI and volumes of EDI traffic to adequately assess its impact on workload. In the COST 320 study, when considering only those companies who exchanged in excess of 10,000 messages per month, 33% admitted to a reduction in staff and only 20% claimed that EDI had not and would not affect employment levels. This evidence implies that there is a threshold level around 10,000 messages per month which must be reached before EDI affects employment levels.

The direct employment impacts of EDI will be modest. There will be some displacement of routine clerical labour due to the adoption of EDI. In the initial stages of EDI adoption, when electronic and paper-based systems are operating in parallel, labour savings will be small. Such savings may well increase as EDI applications mature and become extended in scope (for example in those cases where invoicing or delivery notes are dispensed with). However, when the level of transactions rises over 10,000 per month, the displacement will start to result in internal restructuring. The paper-based management of inter-organisational relations is still a labour intensive process in many sectors, for example insurance and utility billing, so it is in these organisations processing in excess of 10,000 transactions per month that EDI will have its most significant employment impacts. It is probable that the use of more sophisticated integration of EDI with internal systems will bring this threshold level down.

When EDI is mature and widespread, it will yield improved efficiency and labour productivity in purchase and order administration. The extent of direct labour savings from EDI will probably be small compared to other types of automation (and will probably build up very slowly). Labour savings will be unequally distributed between trading partners, with intense users seeing reductions in the numbers employed in clerical functions, medium sized organisations seeing the balance of work tasks shifted away from routine processing and small firms seeing no impact.

The slow build-up and limited extent of labour displacement arising as a direct result of EDI mean that there will be little requirement to adjust employment policies in the light of EDI.

Indirect Employment Impacts

The indirect employment impacts of EDI are very hard to estimate, but will probably be greater than the direct impacts. They will result from restructuring of supply chains arising from a combination of EDI and various other initiatives in supply-chain management (e.g. Just-In-Time supply, Quality Auditing of suppliers etc.). Broadly it can be anticipated that these changes will concentrate industrial structure and the employment effects will occur in firms that fail to adopt EDI. SMEs that have limited internal technical and managerial expertise and resources, and which as a result do not anticipate the importance of EDI, will be most affected. The effects of these developments are not pre-determined, but will depend in a substantial degree upon the broader economic and industrial policy contexts. There will also be powerful regional effects.

Additionally EDI linked to business process redesign leads to the possibility of some industrial restructuring. The organisations most threatened by these changes are intermediaries which arbitrage between buyers and sellers. The wide span of electronic markets may remove the intermediary's role, bringing buyers and sellers into direct contact. An example of this was seen in the Easigo electronic market for surplus oil field equipment in which the electronic market allows buyers and sellers to interact directly, in contrast to the traditional situation where a broker would buy the surplus equipment and then use his knowledge of the market to find a buyer.

Labour Market Impacts

There is little evidence that training issues and skills shortages will be a major factor in the adoption and use of EDI. The training needs identified in EDI projects fall into two categories: technical skills and business skills. For most users the technical issues related to EDI are resolved through the use of bundled pc/software/modem packages which can be interfaced with existing systems.

Recognising that EDI implementation is to a large degree an organisational change leads to the identification of skill requirements in managing the process of change and organisational development. The importance of these skills will increase as EDI develops from the automation of existing inter-organisational links to the redesign of those links, with wider organisational impacts. Currently the main training issues identified in case studies of EDI flow from a narrow view of EDI as a technical IT issue; the wider impacts on business practice are overlooked.

There is no evidence that the lack of particular technical skills pertaining to EDI is a barrier to small firms, but it is a subtle distinction between arguing that they find EDI difficult because they have no IT skills and because they have no IT. The barrier to EDI posed by skill and expertise shortages will be most severe where general IT skills are deficient.

Changes in the nature of jobs are frequently reported without hard evidence of head-count reductions. This pattern may be ascribed to a genuine shift in responsibilities from routine processing to other areas or it may be a temporary phenomenon arising from the use of "natural wastage" to re-balance resources and thus avoid redundancies.

The changes observed in roles with EDI are almost invariably in-line with the concept of "partnership" supply propounded widely in the management literature. This involves a shift from adversarial low-trust relationships between organisations, based on fierce price competition, to co-operative relationships, where resources are directed to developing long-term relationships. Examples of these changes were identified in the TEDIS II case studies. In the study of the Italian pasta manufacturer Barilla's EDI links with distributors it was found that EDI significantly affected the role of buyers, moving it away from dealing with Barilla on a day-to-day basis, towards the monitoring of quality and the negotiation of terms. Similar organisational adjustment was seen in Germany in the purchasing department of Bosch-Siemens, and in France in the sales department of Brun Passot .

The reaction of professional groups to EDI is a major influence in shaping how EDI is adopted. EDI activities span across functions in organisations, so there are alternative departments which can lead EDI implementation, from IT, finance, sales, purchasing or logistics. EDI has been found to take-off most rapidly where the benefits of EDI accrue to the function responsible. In Scottish & Newcastle, the Scottish brewers, responsibility for developing EDI with the retailers resides in the credit management function. This has provided an impetus for S & N to go beyond the EDI receipt of orders, the main benefits of which fall to the retailer, to the computer generation and electronic distribution of invoices, the benefits of which accrue mainly to S & N through lower credit management costs and a higher proportion of invoices paid on time.

Policy

The potential of EDI to increase the efficiency and effectiveness of a wide range of industries gives national governments and the European Commission an interest in fostering electronic trading. However EDI is a network technology, so it is not a matter of targeting individual firms and persuading them to adopt EDI, but of identifying groups of organisations which trade intensively and persuading them to collectively implement EDI, and identifying the technical and infrastructural barriers to the formation of active communities. Before national or Commission resources are directed at supporting EDI, it is necessary to justify that a growth in electronic trading is economically important and that there are barriers to the formation of electronic trading which policy initiatives can overcome.

The variable rates of adoption of EDI between sectors and between European regions is partly explained by the relative economic benefits of EDI and partly by the existence of non-economic barriers to the formation of electronic trading communities. Support for EDI awareness must be based upon an identification of sectors and communities in which electronic trading will generate economic benefits but is being inhibited by non-economic barriers and be directed to overcoming these barriers. The barriers are lack of technical expertise, insufficient understanding of the business impacts, the non-existence of mechanisms for the formation of EDI communities, inadequacy of network infrastructure and unavailability of agreed standards.

Of these the critical barrier is the lack of mechanisms to bring potential electronic trading communities together. In general this requires the establishment of mediators to negotiate and co-ordinate. This stage is particularly problematic where a viable trading community requires co-operation between organisations whose traditional interactions are competitive. The resources required to form these trading communities are not solely restricted to trading organisations, as links with software developers and network providers have also been key in successful communities. Once viable communities are built, the community itself is the most effective body for identifying deficiencies in expertise and infrastructure, dependent upon them having access to information about existing standards, software and communication channels.

Within the community the activities should be directed to overcoming the identified inhibitors. Therefore the nature and structure of activities amongst communities will vary. Mechanisms should also be developed for transferring expertise from developed users to less developed users, both between sectors and between regions. Similarly the needs of communities vary between regions, with mediators only needing to co-ordinate message and data standards in countries where the market in telecommunications services, software and consultancy is well developed, but needing to undertake these tasks where the services are not available.

Allowing communities to self-select and apply for funding to underwrite their activities biases against potential communities where the main barrier is the absence of mechanisms for community formation and biases in favour of communities based around existing industry groups. There is also a potential bias in favour of proto-communities being supported which satisfy political requirements rather than being economically efficient. The successes of the Dutch VEDI programme highlight the benefits of clearly identifying the beneficial outcomes from EDI in specific sectors and then being proactive in bringing the required actors together.

Early EDI programmes, such as the UK Vanguard programme, focused on basic awareness, providing information on standards, examples of best-practice and reports on the viability of EDI in a range of communities. The Vanguard programme may have facilitated the diffusion of EDI in Britain by raising awareness, but the low circulation of its outputs raises doubts about this. It clearly failed to catalyse new EDI trading communities. More successful have been programmes targeted at the formation of communities (e.g. VEDI). The CEC TEDIS programme represents a concerted use of all these tools, ranging from support for message development through the creation of case-studies to the sponsorship of inchoate communities.

The TEDIS programme embodies the wider industrial policy objectives of the CEC to support the European Single Market and use technology to enhance community cohesiveness through disseminating EDI in regions in which it is less developed and targeting the particular needs of small and medium sized enterprises.

Recommendations

Electronic trading is a potential source of advantage for European industry and a route to increased efficiency and effectiveness in public administration. The Commission has an interest in ensuring that EDI is expeditiously developed and diffused across applications. This study has shown that the diffusion of EDI in Europe is inhibited by three factors.

Firstly, the mechanism by which EDI is diffused is through communities co-ordinating electronic trading, based on industrial sectors or specific applications, co-operating, either informally, through the aegis of existing industry associations or by establishing new institutional mediators. These organisational networks are built upon the foundation of existing links, embodying existing perceptions about the community to which each organisation belongs; the result is electronic trading developing in national communities rather than across the European Economic Area.

Secondly, the barriers to electronic trading vary across the European Economic Area . These may be expressed as a hierarchy of needs for electronic trading:

1. intra-organisational IT systems;

2. a market in value added network services;

3. mediators to co-ordinate EDI within defined communities;

4. mechanisms to bridge between emergent EDI communities.

The relative significance of these barriers varies across Europe. In Greece and Portugal the low levels of IT usage limit the economic attractiveness of EDI, whereas in the UK and Netherlands the issues increasingly focus on developing seamless electronic trading which is not sector dependent.

Thirdly, the emphasis on national communities means that it is difficult to achieve "critical mass" in smaller European states. This barrier is particularly severe for Ireland and Denmark, where potential EDI users are disenfranchised from national communities in their larger neighbours.

The goal of European Commission policy towards EDI should be to achieve the "Electronic Single Market", in which organisations across Europe can interchange electronic transactions seamlessly, without barriers due to disparate standards or interconnection between networks. This study has investigated the potential actions for overcoming the barriers identified and achieving this goal.

The TEDIS programme has funded projects to develop pan-European EDI communities, so one policy scenario is for Commission resources to be directed at supporting a wider range of these communities. However this policy is not viable for overcoming the particular problems of EDI growth in Europe. First, setting the boundaries of supported communities around the EEA increases the co-ordination costs, hindering the formation of a critical mass. Second, the potential number of communities is almost limitless, all of whom would have a valid claim to support. Third, due to the heterogeneity of most European sectors, the needs of firms in most sectors are not identical across Europe. Finally, directing resources at sectorial communities does not overcome the need to breach barriers between sectors.

The mechanism by which communities grow is through the aegis of mediators (co-ordinating agencies), which are the collaborative institutions or sets of agreements co-ordinating electronic trading within communities. The most effective means for EDI to grow in Europe is for electronic trading to be initiated in small locally based communities trading intensively, where EDI meets the users' business needs and can quickly become economically viable. The advantages to existing members of new members should ensure that the communities grow. However vigilance should be maintained to avoid the communities becoming closed to new members or used anti-competitively by the original members. The key policy issue is to ensure that electronic trading communities are open to new entrants at an equitable fee. An international community which is closed to new applicants represents a de facto concerted practice to distort trade. These issues are of particular concern in relation to the regulation of electronic markets, where ownership of the system confers power to distort the market.

The specific obstacle to the growth of electronic trading communities which can be tackled is the growth of parallel communities by providing mechanisms for promoting the harmonisation and alignment of electronic trading practices between local and sectoral communities.

As communities grow they will inevitably impinge upon rival communities. The danger of these overlaps resulting in competing networks can be exaggerated. The increasing tendency of EDI communities to eschew dedicated telecom services and the growing importance of operational cost savings over strategic competitive benefits both reduce the barriers to alignment between communities. This has been seen in the alignments between ODETTE and Tradacoms and the wider EDIFACT community. For alignment and harmonisation to progress smoothly there is a need for mediators developing local EDI communities to be kept aware of parallel developments elsewhere. This information facilitates joint standards development, avoids duplication of efforts and eases the problems if the communities impinge in future. To a large extent this need should be achievable through the Western European EDIFACT message development process, as to develop a message without using the EDIFACT syntax is becoming increasingly rare. However to be effective it requires mediators to submit their messages into the process and for the procedures which distribute proposed messages to interested parties in each country to be reliable.

The development and use of standard messages are the key activities for defining electronic trading communities, and therefore the harmonisation of messages between communities is central to the vision of the "single electronic market". For this reason the CEC has been a leading supporter of the EDIFACT message development process, seeking to develop global standard messages. However, the EDIFACT message development process has failed to meet the requirements of European EDI users. The remoteness and bureaucracy of the process is leading organisations seeking to develop messages to bypass the EDIFACT process and develop messages outside it. This increases the risk that competing message standards will be developed in parallel. Even where United Nations Standard Messages are used, the moves for local communities to develop sub-sets based around implementation guidelines or amendments to the standard represents a further danger of incompatibility. The European Commission's key requirement is to prevent the formation of incompatible standards within Europe as these represent barriers to the single European market. The need is to ensure that all players in Europe involved in message development are aware of other parallel activities, whether inside or outside the EDIFACT process. This can most effectively be achieved by ensuring that national and sectoral links into the EDIFACT hierarchy are genuinely two-way. The existence of network externalities should ensure that parallel activities will align, both between national communities and between sectors.

The availability of reliable and competitive telecommunications service suppliers is a critical requirement for the growth of electronic trading. Increasingly, through the maturing of the market, value added data services are becoming a commodity, where users can choose between alternative suppliers on the basis of service quality and price. In this mature VANS market, competitive advantages are enjoyed by suppliers owning data networks, squeezing out niche suppliers (e.g. GEIS, IBM, INS). The growth of these niche suppliers has been as a result of PTTs failing to innovate to satisfy the nascent EDI market. Telecommunications deregulation in the UK and the US has forced the dominant telecommunications suppliers to compete in all areas of service provision. Increasing deregulation across Europe will lead to greater competition between VANS. This competition accelerates the take-up of EDI because it forces transmission charges down and gives the suppliers an incentive to market EDI services aggressively. However, one critical issue in the commodification of EDI is interconnection between networks. Competitive forces are driving EDI networks towards interconnection, as users do not want to be tied to a single network's user-base. This study has shown that the technical and business uncertainties surrounding interconnection are acting as a barrier to the use of interconnection but that these barriers will be overcome due to competitive forces on telecom service providers following full deregulation.

This study has shown that many of the issues associated with EDI are transient problems due to the immaturity of the European EDI community. Many of these issues will be resolved through on-going developments: interconnection by telecom deregulation, EDI's use in medium-sized enterprises by the growth of the market in packaged software, the take-up of FEDI by customers replacing their existing financial systems and the banks developing products to meet the customers' needs. EDI is developing very rapidly, so there is the danger that policy initiatives will be addressing last year's problem. Where there are clear policy implications - in regional policy and sectorial industrial policy - EDI forms one element in a wider milieu. There is a case that the disadvantaging of small enterprises, particularly where they are in peripheral regions remote from strong electronic trading communities, means that the encouragement of EDI may form one element in economic development policies in regions. Similarly the potential impact of sophisticated inter-organisational data exchange may have to be considered in relation to the future competitiveness of some European industries. The most obvious example of this is the potential impact of CALS and related inter-organisational data exchange technologies on the competitiveness and structure of the European defence industry. The European responses to the CALS programme have been ad hoc, reactive and uncoordinated at the European level. The potential of task support EDI to trigger industrial restructuring and impact upon international competitiveness generates a need for focussed programmes to support and co-ordinate the introduction of these technologies in key manufacturing sectors. However we have found no evidence of generic permanent barriers to the adoption of transactional EDI across Europe requiring a pan-European co-ordination mechanism.

Ian Graham, University of Edinburgh / I.Graham@ed.ac.uk