Taurus & Crest case - why did Taurus fail and Crest succeed?

For the London Stock Exchange to run effectively it must have the right technology put in place. However, this has been a problem for the LSE and has greatly damaged its reputation around the world since the early 1990's.

In 1986 the "Big Bang" changed the way the stock market operated. There were less regulations to follow and deals were now performed through automated systems instead of face-to-face on the trading floor. This increased the trading volume of the LSE dramatically but they now had to improve the business behind the scenes. After a deal was struck the share certificates and money still had to change hands, which could take weeks using the LSE system at the time. The LSE quickly decided to develop an automated system to tackle this problem.

In 1989 there was a proposal to develop a system called TAURUS (Transfer and Automated Registration of Uncertified Stock). This system looked to automate the settlement business transactions. It would also aim to dematerialise stock certificates and instead stock ownership would run on a computer database. If TAURUS were successful it would have saved the LSE millions per year and decreased the risk of purchasers going bankrupt before settlement. However, the implementation of TAURUS did not go according to plan.

The main factor that resulted in TAURUS's failure was that the design team tried to adhere to too many needs. All firm members would have to use TAURUS so each firm demanded that the system was designed to allow them to trade using their individual methods. This was on top of the huge size and complexity of system needed to deal with so many transactions. This lead to the design team trying to design a system that carried out the standard method but also performed differently for the multiple firms. In addition the government wanted their say in the design of TAURUS, which resulted in the design team being issued a 150 page legal document of complex regulations to follow. The end result was the TAURUS tried to meet to many needs and ended up meeting none. Delays occurred as the system was constantly getting redesigned because of outside pressure. Delays obviously lead to mounting costs that did not look like slowing. During this time other bourses were successfully implementing automated systems so eventually the project was terminated in 1993.

Another mistake made by the design team was that the software they choose to operate TAURUS needed extensive modification to meet all the different demands. Due to all these modifications TAURUS was basically relying on unproven software, which is obviously very risky. This also added to the extra costs and delays that lead to termination. The size of the project team was also criticised as they had hundreds of staff working on the design. With so many staff it is important to have rigorous structure to prevent coding mistakes this was not done in the case of TAURUS.

With the failure of TAURUS the problem of implementing an electronic processing system was still there. The Bank of England now looked to solve the problem and regain the credibility of the LSE. Learning from the mistakes made by the LSE the Bank of England set up a small ten-man task force and within 3 months they had come up with a new system called CREST.

The CREST system looked to follow a minimal approach using tried and tested software. CREST was designed independently from the LSE so unlike TAURUS the design team had no need to adhere to the different firms needs. Their minimalist approach meant they designed CREST to perform only 2 processes, which made up nearly 90% of all transactions of the LSE. This compared to the 21 attempted by TAURUS. They also had a smaller design team of just 20 with 4 more experienced members supervising the whole project. This resulted in all modifications being closely examined before getting turned into code.

CREST went live on 1996 on schedule and within budget and it is still being used today. Other stock exchanges have tried to buy the system as a result of its effectiveness but it is not for sale. Keeping to a minimal approach and using tested software meant the system was implemented with only a few problems that were solved. TAURUS tried to be too many things to too many people, which resulted in its failure. Constant modification lead to delays and increasing costs resulting in the LSE terminating the project.

References

TAURUS and CREST, Failure and Success in Technology Project Management by Christopher H Head

http://www.practicallaw.com/7-100-4206

http://commentisfree.guardian.co.uk/daniel_davies/2006/09/dont_just_do_something_stand_t.html