Examination of Witnesses (Questions 1-19)|
WEDNESDAY 22 MAY 2002
1. Good morning, ladies and gentlemen. Can I
declare the formal session of evidence open this morning. The
Committee is doing a one-off evidence session on the Child Support
Agency and we are pleased to have as witnesses the Chief Executive
Doug Smith and his team, which consists of Mr Vince Gaskell, who
is the Child Support Reform/Change Director, and Mr Mark Neale,
who is from the Department for Work and Pensions and is the Children
and Housing Strategy Director. Thank you, gentlemen, for coming
and thanks for the written memorandum which is valuable. I think
we should say at the start that the Committee had already planned
an up-date session with you just now, at this stage in the year,
before we had notice of the Secretary of State's decision to delay
implementation of the new reform system. The inevitable question
to start with is what can you tell us about the delay?
(Mr Smith) I think, first of all, it
is worth noting that the delay is to one part of the Child Support
Reform implementation and the Agency has implemented some of the
early phases. Child Support Reform is a large and complex project
that involves new legislation, new IT systems, the way the Agency
is organised, the way the staff go about their jobs and embraces
more than simply the IT changes. Focusing on IT, we knew from
the outset that this was going to be a large and complex IT development,
in that it is a brand new IT system we are looking to put into
place to support the new legislation. We did try to ensure that
we did learn the lessons from other large developments, so we
did put in place a comprehensive approach to risk management,
which Vince Gaskell will talk about. We did ensure that we had
external reviews of the project at regular intervals and we did
ensure we built contingency into our plans. The early development
work on the project, the IT aspects of the project, went well.
They went broadly according to plan. As the size of the system
grew, as the whole build grew, we started to see some delays creeping
in to the development and testing. That was containable. We expected
to see some delays, that is why we built some contingency into
the arrangements. EDS remained confident as we went through the
build of the system that it would be completed and tested in time
for our expected go-live of April this year. Around the back end
of last yearvery much at the back end of last yearwe
started to become aware of the difficulties that EDS were having
in integrating various elements of the system to work together
in an integrated way. In December jointly with EDS we agreed to
replan the final phase of activity leading up to April on a different
basis, so that we worked together and we did the final stage of
testing in parallel rather than sequentially, to support a go-live
in late April. That also required us to scale back with some reluctance
our plans to live-test the system in our offices before go-live.
At the back end of last year/early this year we both recognisedwe
and EDSthat there were risks attached to that approach.
At that stage we believed that the balance of probability remained
that we would be able jointly to deliver a system in April this
2. Can I just stop you a moment? None of us
are computer experts on the Committee, but you are talking about
risks and I understand that the complexity in a project of this
scale is actually about the management of the introduction and
in the scaling up of the process. However, do I understand you
to mean a software and application kind of problem, or is it a
hardware problem, or is it an interface problem with other parts
of the agency system? Can you pin down for us in language that
lay people would understand what the actual problem is that is
causing you all these new risk assessments and analyses?
(Mr Smith) The issues around the turn of the year
that were concerning us the most were how you put the different
components of the system together.
3. Within the CSA?
(Mr Smith) Within the CSA.
4. So it is not an interface problem with Income
Support or any other aspect of the Benefits Agency?
(Mr Smith) I will come to that because down the track
we did discover some interface issues around the interface with
the Benefits Agency and Jobcentre Plus. The issues around the
turn of the year were largely about how you actually bring together
the chief components of the system and put together the mainframe
at the centre, the boxes on the desktop, the communication links,
the telephony and how you interact the telephony with the IT and
how you integrate our work management system with the IT.
5. How was that missed? I had rather assumed,
with the information that was available to us in the public domain,
that the work was proceeding apace and problems arose only when
you really, towards the end, started linking up, almost at the
completion of the project. It would appear to me that the risk
of problems in an internal configuration of the system could easily
and should have been, surely, foreseen.
(Mr Smith) Yes, they were foreseen and this was not
something that came unexpectedly. We knew that as we brought the
different components togtherand Vince can speak in more
detail around this areathere were going to be issues, and
we allowed time for those to be resolved as we went into the first
quarter of this year. So this was not a surprise to us. Inevitably,
with a large, integrated system, bringing the different components
together, bolting them together and operating the system in an
end-to-end way for the first time, issues are going to emerge.
That is why we built time in to the timetable. We expected to
see some of those issues and, true enough, they emerged. However,
those were not the key issues, at the end of the day, around why
we deferred it. This was an issue around the end of the year that
I referred to. It did push out and shorten the amount of time
we may need between the time that EDS have broadly dealt with
those problems and go-live when some other issues came up, like
the interface, that had to be addressed within a short time-frame.
6. So was this EDS' fault?
(Mr Smith) I do not think it is an issue of fault.
I really do want to stay away from fault. This is a joint project
that we are developing together. This is an issue of fact, though:
as we brought the various components together they took longer
to integrate than we expected within the plan that we agreed.
7. But it has wrecked your annual plan, has
it not? All the planning assumptions are out of the window, as
(Mr Smith) It has wrecked our ability to deliver Child
Support Reform from April, and that is a big issue for us and
it is hugely disappointing. What is not wrecked is our ability
to maintain our level of service to our clients.
8. It is a substantially reduced level of service.
You had been busily writing to your customers earlier saying "Don't
worry, reform and paradise is round the corner". It was a-win-win
situation because the people who had custody of children would
get benefits, in fact, because of the change in the maintenance
premium with a £10 increase, but so would the non-resident
parents have, on average, reduced levels of maintenance to pay.
So everybody was expecting and waiting, and with some anticipation,
on this new system being rolled out. All of these expectations
are now as nothing. We have no idea when we are going to get access
to these advantages.
(Mr Smith) Indeed. The Agency was equally waiting
to be able to confer this advantage, and so we are disappointed
that we cannot deliver.
9. Just come back to some of the technical issues.
You say it is internal to the Agency. Does that mean that you
still have to do the interface issues once you have resolved the
internal problems? Mr Gaskell can probably tell us a bit more.
I am really keen to get on the record what exactly it is that
you still have to do, because it is only then that we can have
any kind of assessment about how long it is going to take to fix.
(Mr Gaskell) The way that the system was being developed,
really, by joint agreement between us was in what we call a series
of builds or chunks. As Doug indicated, the early work on those
builds went broadly to plan. As they added the later builds to
the rest of the system that increased the complexity of the system
inevitably and we started to hit problems in terms of the amount
of lapsed time to do the testing on those later builds. As those
builds start to come together, which they did towards the back
end of last year, a number of other things then need to happen.
First of all, this system would not have operated on the old infrastructure
we have in our offices. So to help reduce the risks of the programme
we had gone for a programme of implementing PCs in all our offices
in the vanguard of the Department's Early Office Infrastructure
and that was completed to time in December. So all our offices
were equipped with those PCs but that did not materially affect
the testing. What we needed to do at that point, and with EDS,
was as the software was being developed for the main system to
include the new rules and new accounting and so on that we needed.
It then has to be seen to be working across that new infrastructure
and all the supporting infrastructure that EDS would need. As
Doug touched on, that has brought the new mainframes that EDS
have to put in place, linking those to the network, through to
the servers and into the PCs in our offices. We simulated that
environment in Newcastle to try and make sure that that worked
properly. As those things were starting to come together, as they
should at that point, the testing that was under way to see if
all of those worked together well and the software itself was
working wellparallel stream testing taking place with EDSthe
defects and problems they were finding, which you would expect
to find during a period of testing, were inhibiting progress on
testing. So that in the run-up to the Minister's decision what
we were looking for and would have expected normally is to see
that corner being turned and the number of defects being identified
and rectified going down, and you would expect to see a clear
period when we would expect the testing to end. Both those things
we did not turn the corner onthe number of defects being
found in both the application of the software itself and some
related to how that software was working in the infrastructure
too. You mentioned the interfaces. Clearly, to make our maintenance
premium work this particular application software has to interface
with the Jobcentre Plus systems, Jobseekers Allowance and Income
Support. There were particular problems from the outset in testing
to see if those interfaces work in getting the data across from
the Jobcentre Plus systems and then populating the relevant parts
of the new system for Child Support. There were problems at both
ends. Those were particularly problematic and meant that the testing
particularly on the interfaces was stalling. So the position that
we had as we approached March was we were not turning the corner
on testing, defects were still risingwe were finding more
defects in testing than we were clearingand we could not
see a clear, projected end to testing. Neither, at that point,
did we have the interfaces working. There were one or two other
issues, just for clarity. Populating the part of the system on
managers' work flowthis is the work allocation to individual
members of staffthere were particular problem areas in
there too that we had to sort out, and trying to get the notifications
out to our clients was problematic because of the number of defects
we were finding there. For that reason that was a negative impact
on our clients and a negative impact on our staff with no clear
idea of when we would turn the corner.
10. That is very helpful. Let me make it clear:
we might be asking you some tough questions here but we are absolutely
with you and with the Secretary of State in deciding that none
of this should be rolled out until it is absolutely robust. We
agree with that. We said that in our earlier Report and so nothing
that we say this morning should be taken as contradicting that
decision. However, that does sound slightly worrying because it
sounds to me as if you have not actually got the internal system
debugged even to the extent of being able to approach the interface
problem and the Legacy systems. That is what I felt was always
going to be the hardest bit of it. I cannot understand why the
internal work that could and should have been anticipated is causing
so much difficulty. Is it a question of scale, as so many of these
PFIs are? Bench-testing the applications and the systems on a
small scalehas that been done? Have you demonstrated on
a small scale that the applications and the system are robust?
Does it work at any scale?
(Mr Gaskell) I can answer that in two levels. First
of all, we wanted an extended period of what we call live-testing
and we wanted that to begin in December. We were not able to achieve
the level and degree of live-testing because of the number of
defects that were being found in testing. That was inhibiting
progress, but we did have a limited start to live-testing in December,
and have progressed since then. So both ourselves and EDS have
got an understanding on two things from that: first of all, which
parts of the system are not working in a live environment and
we have also got to understand how that interacts with the changes
to the business process and the way of working. That is still
working at our site at Newcastle. The second thing we have done,
because, jointly with EDS, we were concerned about the still rising
number of defects being found, and this inhibiting progress, we
were therefore unable to form a clear judgment because we had
not got sufficient progress in testing, that the system itself
across the infrastructure would perform at the required service
levels, even with this period of live-testing. Crucially, EDS
were not able to get to that point in testing to be able to fully
demonstrate that. So what we have done with them, working jointly,
is to understand what the pattern of defects was pointing to,
both across the interfaces and within the software itself, to
see if there were any particularly worrying areas that pointed
to a potential fundamental flaw in the system or whether it was
just going, in the elapsed time needed, to work through those
defects and complete the testing. That has led us to conclude
that at this point there is no fundamental flaw in the system
itself. These are newly recruited, high-calibre, external IT experts
that have helped us within the Department, working jointly with
EDS, to form that view. That is based on the testing that has
been done to date. The result of that about what it shows is a
particular pattern of defects. So that has given us a degree of
confidence, but clearly you do not get that full degree of confidence
until you are sufficiently through testingparticularly
user-acceptance testingto be fully satisfied that the system
11. So it is a question of refinement rather
than redesign, at this stage?
(Mr Gaskell) At this stage the indications are that
that is the case.
12. Who set the timetable? It seems to have
been very tight, to start testing in December to start in April.
I am not an IT expert but that seems to me pretty tight. Who set
(Mr Gaskell) The timetable we agreed with Ministers
and with EDS. EDS were under no particular pressure to come up
with a particular timetable and that was set some while ago.
13. How long ago?
(Mr Gaskell) The timetable was agreed, I think, in
January/February, if my memory serves me correctly, in the year
2000 and the contract was signed with EDS for this particular
system in August.
14. Why did it take eight months?
(Mr Gaskell) To do what?
15. From January/February to August?
(Mr Gaskell) I do want to go back to testingI
do not want to lose that pointbecause we did not just start
testing in December. If I can just cover that point about the
period between January and August, one of the recommendations
from I think it is the Public Accounts Committee in the past has
been that there needs to be a period up-front with your supplier
to work through jointly and to reach an understanding about what
the business requirements are, so that they can put together a
firm and clear plan for delivery and a period where we can understand
how they are going to meet those requirements.
16. Is that what is known in the trade as a
(Mr Gaskell) There was a gateway review pre-contract.
17. I will come back to that in a moment.
(Mr Gaskell) Can I follow up a separate point, because
it would be a wrong conclusion to draw that we started testing
in December. One of the things that we did to help manage the
risks associated with the programme is we went into this jointly
with EDS with our eyes open, knowing this was a large-scale and
complex programme. We tried to explore ways with them about whether
there was any other way to develop the system and the approach
that we came up with was a joint one agreed between the Department
and EDS. Following that, what it allowed us to do is as each of
the chunks of the systemor the builds of the systemwere
coming through it allowed them to carry on their system-testing
and their technical testing of the system and for us to be able
to do some user-acceptance testing on those parts of the system.
That pattern of testing has proceeded not since December but well
before then. The precise start date for that I cannot remember
but it would have been sometime in 2001, if my memory serves me
18. I thought Mr Smith said the testing that
was done was changed and it was not going to be done sequentially,
as you originally planned, but done in parallel. You now seem
to be suggesting that it was being done sequentially.
(Mr Gaskell) To begin with we started to do this sequentially,
based on the build approach. As the system grew and its complexity
grew problems started to emerge around about spring to summer
of last year around our ability to stick to the current planned
timetable. Consequently, they are things that we incorporated
in the contingency measures that we built into the plan. However,
by the time all the components started to come together at the
back end of last year, as Doug has already indicated, that led
us to do the further re-plan. Rather than do things sequentially,
to try and secure an implementation in April we agreed to change
the approach to the previous form of testingwhich is the
ideal way to do thisand to try and do some of that activity
in parallel. Hence the revised approach.
19. Can I come back to the gateway reviews.
My understanding is that the new procurement rules require that
the Office of Government Commerce set up some so-called gateway
reviews. Can you tell me what they amount to and whether the project
went through any gateway reviews?
(Mr Gaskell) Yes, we have gone through gateway reviews.
In the early days of this particular programme they were not called
that, they were called the McCartney Reviews at that point. We
went through a pre-contract gateway review. However, the pattern
of those reviews was still being establishedit was very
embryonic daysand at that point there was no clearly agreed
pattern for those gateways.
1 The Department's project to upgrade/replace its IT