1 Introduction
The Aries Pensions Club, founded in 1997 to continue work begun at City University ten years earlier, has brought around fifty private sector organisations together to build the Aries Pensions System and achieve IT solutions to other shared business problems. This article is about how this unusually successful initiative has been achieved, and some of the key success factors for the collaboration.
2 How did it all start?
The original model was the ARIES (Alvey Research in Insurance Expert Systems) Club, part of the Alvey Programme between 1985 and 1987, which built two prototype expert systems in the domains of commercial fire risk and equity selection. The project was funded approximately one-third by the Government, through Alvey, and the remainder by twenty-three companies each contributing £10,000.
Some of the key people decided, when the project ended, that the work of spreading awareness of expert systems in the insurance industry was worth continuing. The development work moved to City University and the Aries at City Club was formed, with fifteen companies agreeing to support a one-year programme to develop a system for training life assurance underwriters and a personal financial planning demonstrator system. The Aries Team of researchers (including both directors of Aries Pension & Insurance Systems Ltd) was established. The total funding, £105,000, was this time supplied entirely by the private sector.
The new work in 1988 was also completed successfully; particularly the life underwriting system TULIP, because this modelled a task which we found was performed in a very similar way in all companies. Personal financial planning, on the other hand, was a domain in which companies were more likely to vary in their approach, notwithstanding the requirements of the Financial Services Act which came into force in that year. TULIP's relatively greater success repeated the Alvey experience, where the fire risk development (also an underwriting task) proved more satisfactory than equity selection, and provided an important early lesson in domain selection. Collaborative projects depend on a shared mutual understanding: more will be said about this later.
3 Pensions legislation: a suitable case for treatment
The Aries Team believed there might be other problems in the financial services industry which PC-based expert systems could solve. In Spring 1989, the UK legislation governing pension schemes was ramifying rapidly. So was the business, not least because of the 'personal pensions revolution' which had begun less than a year earlier. Pensions technical expertise was becoming increasingly scarce compared with the growing numbers of pension administrators. There was a problem facing all pensions providers: how to ensure that they complied with the increasingly complex and ever-changing legislation and Inland Revenue and DSS requirements? Perhaps a computer system which capitalised on the twin advantages of IT over the human brain, in information storage and retrieval and speed of processing multiple data items, could help?
At the time, the Aries Team knew something about pensions (Gary Chamberlin has an actuarial background) and contacts in the industry provided sufficient encouragement to develop this idea. An outline proposal was presented and with guidance from those interested, a six-month pilot project funded by five life offices and one major company pension scheme was established. The objective was to build a knowledge-based system to guide pensions administrators through the minefield of pension transfers. A thorough analysis of this particularly complex area, it was suggested by one expert, would afford a credible basis for further development. He was right, for the project - which still continues today - has grown into the Aries Pensions System: the first commercial working system in this field.
4 The basic Aries model: how and why it works
The Aries funding model relies upon a number of companies each being persuaded to contribute an equally modest sum, by the twin promises of delivery of proportionately greater benefits and an active role in directing the project to achieve those benefits. It has been found in practice that a minimum of six to eight Club Members are necessary for a viable Aries project. Aiming for fewer sponsors would require the contribution from each to be pitched at a more prohibitive level; or alternatively, the benefits to be scaled down to a much less attractive level.
Having identified a problem and after some research and consultation with experts, a proposal is outlined including a sketch of the contents and scope of the proposed computer system. The number of companies which might reasonably be expected to support the project is estimated and divided into the calculated total cost of development, to produce the price of participation. Thus a viable solution to their problem can be offered to each prospective collaborator for a fraction of the total cost. Careful judgment of the market is required, of course. 'Oversubscription' allows further development work for enhancement of system functionality and content, and possibly lower maintenance charges than would otherwise be necessary.
Typically, a successful proposal will address a shared, industry-wide problem: ideally, one to which a common solution is attractive because a bespoke system offers few commercial advantages in relation to the much greater cost of development. Environments which are particularly burdened by legislation and regulation - such as the provision of private and company pensions - are promising targets, for two reasons. Every player has to comply with the same set of requirements, so there is no competitive advantage in succeeding; hence any way of sharing the cost of compliance is worth examining. The second reason is the problem of managing complexity and maintaining consistent and reliable interpretation: each new piece of pensions legislation is in general additional to, rather than in place of, earlier legislation. The 1995 Pensions Act alone added 187 pages to the pile.
Once sufficient commitment has been obtained to launch the project, a management committee, on which every participating company is entitled to a seat, is established. Advised by the Aries Team, the Committee decides the programme of work on the design, functionality and content of the system. The Aries Pensions Club Committee, for example, typically meets about five times a year. Further collaboration is engendered through the continuous involvement of domain experts who provide the knowledge and verify the system, and of a user community in each Member company. It is not enough to be business-led and problem-driven: to be successful in developing effective solutions, the approach must be consistently user-centred. Being user-centred requires understanding of the context in which the task is performed.
5 Development of the Aries Pensions System
Initially, five companies provided experts, and knowledge acquisition proceeded iteratively with each expert commenting on a draft or 'paper model' which incorporated their collected insights. The resolution of apparent conflicts often revealed that different terms were being used to describe what was essentially the same concept. Sometimes too, especially in the early days, it involved eliminating false distinctions between an 'official' term; the industry's common equivalent for it; and a particular company's own 'product'. In verifying the draft model, quite frequently a correction would be made by just one expert, but then receive collective assent from all the others in the next round.
Users of the Aries Pensions System (APS) text seek supporting information on what to do in the context of a particular event in the 'life cycle' of an individual's membership of a pension scheme: for example, when the member joins, changes jobs or opts out of the scheme, wants to transfer to another scheme, retires, or dies. Thus an event-driven approach to design has been adopted, by which APS provides information on the relevant checks and procedures after the minimum number of data inputs (mostly menu selections) by the user. A dynamic index offers an alternative way to find the answers to specific enquiries.
Textual guidance is actually now just one of four main areas of functionality offered by APS. The system provides fully comprehensive calculations of the maximum permissible contributions and benefits for scheme members, covering all types of UK-approved schemes. The calculations and text are backed up by an extensive reference section which includes a full technical glossary, guide to abbreviations, and a database of statistics. APS in addition provides users with a browser and search engine for the Practice Notes (IR12) and many other key pensions documents.
The system was originally designed for users who are first-level supervisors with two or three years experience of pensions: people who are capable of distinguishing questions about the rules of particular pension schemes and products, for which APS does not cater, from matters concerning the general tax and social security legislation applicable to pension schemes. APS defines the permissive boundaries - the 'floor' and the 'ceiling', ie the minimum requirements and the maximum allowances - within which pension schemes operate. From the beginning, the design emphasised ease of learning and ease of use via menu-driven access to material held in text files.
In the course of the project the user community in general has broadened, mainly upwards to include more knowledgeable users. This reflects the fact that pensions legislation has become so complicated that few people can always be confident in their own minds. 'Knowing what you don't know' is perhaps the first requirement of an expert. Those who do know which documentary sources to consult for the answer to a query often find that APS, which 'knits together' the key facts from various sources, offers a quicker route. Thus APS has come to be regarded by many as such an authoritative source that it is being used by the experts themselves. In some companies the complete in-built benefit calculation routines are being used either as a check on, or in place of, in-house systems.
APS was designed from the start to run under DOS on standard 640KB IBM-compatible PCs to address the widest possible range of user machines. The original system, using an expert system shell was later rewritten in an in-house application development environment built in Pascal. This facilitates APS's integration in the varied computing environments of the Aries Members, including networking. Conversion to Windows (3.11, 95, 98 and NT) will be complete by May 2000.
From the initial six sponsors, the APS project expanded through two major phases between 1989 and 1992 to first eighteen and then thirty-five companies. The majority were (and still are) life offices, plus an ever-increasing number of actuarial and benefit consultants, larger IFAs, third party administrators and major pension funds. Following reconstitution of the Aries Pensions Club in the private sector in 1997, a new wave of expansion saw membership expand to 46, in spite of the losses suffered as a result of convulsions such as takeovers and mergers in the industry. Since 1994, the continued commitment of each Member to the APS project has been decided annually by their decision whether or not to pay the mutually agreed maintenance fee.
The aims of the APS project have been met, success varying with vigour of promotion within the sponsoring company. In some, the system has not percolated beyond the pensions technical support area, while experts in other companies have encouraged its wider use by quoting directly from it (and naming their source) when answering telephone enquiries. Several companies have run APS training sessions for pensions administration staff, with the aid of case study materials and a 'friendly user guide' produced by the Aries Team. Elsewhere, some new user groups have discovered APS accidentally - and been delighted to find a reliable and very accessible way to check their calculations of maximum benefits and contributions.
6 The benefits of Aries
The Aries project has aimed to make APS the first port of call for pensions staff seeking technical guidance in response to an enquiry from a less-experienced member of staff, salesperson, or broker. User adoption has been assisted by consistent attention to making the system's advice both concise and relevant. By moving expertise downstream, closer to the original enquirer, the supervisor is enabled to answer more questions directly and empowered to deliver a better service to the customer. For the company, consistency and cost of the service can be better controlled with this uniform delivery mechanism.
It has been essential to the success of the project that senior managers regard APS as sufficiently reliable to deploy it as a partial alternative to their in-house technical experts. Users must also trust the knowledge delivered by the system: otherwise they will simply ignore it and continue to ring the experts. It must be emphasised here that APS has never been intended to replace these people, but to relieve the pressure on them by replicating their collective understanding about the ground rules. Unusual, particularly complex, or previously unheard-of cases will still be dealt with by the experts.
Trust in the system is in turn dependent on the Committee's ability to rely on the knowledge acquisition and verification programme, conducted by the Aries Team in conjunction with the experts. Over the past decade, many of the experts who have contributed have become well-known to each other and mutual reliability established in their insight and judgement. The Aries Team, including Gary Chamberlin and Ian Neale who began the project in 1989, has also grown in professional stature.
Apart from the Aries Pensions System itself, an important benefit for Members of the Aries Pensions Club has proved to be the forum provided for sharing knowledge and experience. The opportunity at Committee meetings to exchange information informally has been valued by the individual company representatives. Many do not have the opportunity to sit on the various influential industry bodies, such as the Association of British Insurers and the National Association of Pension Funds and their committees, and can feel rather isolated in their own companies. Committee members tend to be middle and senior pensions technical managers.
A number of conflicts in interpretation of the regulations between companies have been discovered, and usually resolved by consensus or by reference to the appropriate authority such as the Pension Schemes Office (PSO). As in any legislative domain, considerable uncertainty often exists over the interpretation of requirements - even, it has to be said, inside the PSO, who appear to have provided differing guidance on different occasions. Where agreement cannot be reached on a subject which all agree APS should cover, a 'refer to senior manager' note to the user is inserted in the knowledge base.
7 Other projects
Several other projects have been successfully launched on similar principles. In 1991, the Aries Team found that pensions technical training was a problem for many companies. The increasing availability and popularity of pension transfers, for example, had meant that pensions administrators now needed to know the basic legislation and practice governing other types of pension scheme besides those with which they were familiar. At the same time, traditional training methods were increasingly expensive and unreliable in their results.
As a result, seven Aries Member companies have collaborated with the Aries Team to fund and develop a very comprehensive library of sixteen computer-based training modules covering pension transfers, using software written in-house by the Aries Team. This Transfers Training System (TTS) was followed by a second, and even larger, project to develop a computer-based training system in the domain of Revenue Maximum Benefits calculations. Other developments have included a technical training system on Equal Treatment and the UK's first comprehensive CBT programme for Trustee Training.
While still at City University, Aries participated in the Integrated Cooperative Workspace (ICW) Project, a collaboration between universities and the IT industry co-funded by the Engineering and Physical Sciences Research Council and the Department of Trade and Industry. Using infrastructure developed by industrial partners for computer-supported collaborative working, a Demonstrator was built to show how the pension transfer process might be supported in future. This considerably extended the principle of collaborative working, beyond 'groupware' running on a local area network. The ICW Aries Demonstrator modelled an industry-wide collaboration between a final salary company scheme, an independent financial adviser, a life office providing personal pensions, and other organisations including employers, the DSS, and the PSO.
8 Conclusion: 10 Critical Success Factors for collaboration
The Aries Pensions Club has been a notably successful collaboration in the private sector over the past eight years. From this experience, the lessons learned might be expressed in terms of ten critical success factors for similar collaborative projects:
- The problem can be defined in terms common to a number of organisations.
- Collaborators have a common interest in the proposed solution.
- The development team has business credibility.
- A fixed-price entry fee, representing a fraction of the full cost of the defined benefits.
- Equal opportunity for members to participate in project direction and management.
- Involvement of all problem owners, including willing domain experts and users.
- Use of paper models to verify results of knowledge acquisition.
- Project remains business-led and problem-driven.
- Solution system design is user-centred.
- System runs on a variety of industry-standard platforms and with minimum constraints.
