COUNCIL FOR THE CENTRAL LABORATORY OF
THE RESEARCH COUNCILS: ACCOUNTS 1997-98
THE PROCUREMENT AND INSTALLATION OF FAMIS
7. The Council's first draft of a specification for
their new accruals-based accounting system was completed in December
1994, but the final version, which formed part of the Council's
invitation to tender, was not finalised and issued to potential
contractors until May 1996, some 18 months later. The Council
accepted a tender from Coopers and Lybrand which was based on
Oracle software and Data General hardware. That tender was accepted
despite concerns expressed within the Council, including by their
then Chief Executive, about system security and because the contractors
had not been able to demonstrate to the Council that the proposed
hardware/software combination was capable of operating. The Council
sent a letter of intent to Coopers and Lybrand on 6 September
1996, having received confirmation from them that they would have
all essential features of the new system operational by 1 April
The letter of intent enabled work to proceed but the final contract
was not agreed and signed until 20 January 1997. The agreed contract
price of £544,000 covered implementation, configuration of
the new system and certain data migration services.
8. The Council ceased to operate their old cash-based
accounting system on 31 March 1997 and introduced the new FAMIS
system on 1 April 1997. They did not follow the generally accepted
practice of running the two systems in parallel for a set period.
This would have enabled them to satisfy themselves as to the integrity
of the new system and provided a back-up in the event of the new
system not operating properly. The Council told the Comptroller
and Auditor General that such an arrangement would have incurred
very significant additional costs that might have amounted to
as much as £2.5 million. However, the Comptroller and Auditor
General considered, and the Council have now accepted, that a
more reasonable estimate would have lain in the range £0.75
million to £1 million.
The Council told us that the £2.5 million estimate had been
made retrospectively in November 1998 during the preparation of
the Comptroller and Auditor General's report.
We asked the Council what estimate they had made initially which
led them to rule out parallel running. They told us that before
the project began they had expected to have to pay some £500,000
to £750,000 for running the mainframe in addition to internal
staffing costs, and that in view of these costs they had not considered
it worth pursuing parallel running any further.
However, they admitted that with hindsight they should have parallel
run as this would have been the only way to have tested the system
9. We asked the Council if they had undertaken a
serious appraisal of the effectiveness of FAMIS before it went
live. They told us that they had undertaken pilot exercises, in
which users had participated, on each module of the system. Some
hardware problems had been found during these tests. These pilot
tests had not, however, tested the system at full load.
The Council therefore had taken the risk of introducing the system
on 1 April 1997 without the benefits of full scale acceptance
testing or parallel running.
10. The Council went on to tell us that the hardware
supplied had been inadequate for the task. It had been too slow,
and unable to handle the wealth of financial information required
to run their business. In addition, their own PCs had been antiquated
and the supplied software had had many bugs which required patches.
All of this had caused delays.
The specification for the system had required that it should be
At the time of signing the contract, however, the Council had
been aware that the software being supplied would not satisfy
this essential requirement and that they would have to rely on
a later upgrade.
The negotiations with Coopers and Lybrand had also left unresolved
the issues of VAT budgetary control and fixed asset accounting.
The Council told us that these issues had not been specified and
locked into the contract.
They had since concluded that the supplied version of the fixed
asset module could not support the production of the Council's
accounts in accordance with Treasury requirements (they had to
resort to the use of a manually-prepared register for 1997-98),
and that the VAT accounting capability could not be fully supported,
although there was sufficient information present to enable the
Council to compile their VAT returns.
11. In view of the compressed timetable for procuring,
installing and implementing the system and the significant problems
that became apparent during the contract negotiations and pilot
testing, we are astonished at the Council's decision not to parallel
run the new system with its predecessor system. We consider that
they gave insufficient consideration to the risks associated with
this decision and that their analysis of the case for parallel
running was seriously inadequate. Moreover, we note with concern
the Council's admission that their recent and retrospective £2.5
million estimate of the cost of parallel running was two and half
times in excess of the likely actual cost. We find it unacceptable
that the Council did not assess fully and accurately the likely
costs of parallel running at the time the decision was made, thus
preventing management from taking that decision on a properly
12. We are surprised that the Council took the risk
of placing a procurement contract for a computer system knowing
that it was not demonstrably Year 2000 compliant, that the fixed
asset module did not meet the Treasury's requirements for fixed
asset accounting, and that VAT budgetary control would not operate
as required. We note that the uprated version of the software
is expected to meet these requirements and look to the Council
to verify that this is the case well before 31 December 1999.
2 C&AG's report, paras 11-14 Back
para 15 Back
report, paras 16-17, and Q66 Back
7 Q114 Back
report, para 18 Back
10 Q2 Back
11 Q89 Back
32, 91 Back
13 Q103 Back
report, paras 27-28 and 33 Back