Examination of Witness (Questions 1040-1059)
SIR BRIAN
BENDER, KCB, CB
6 DECEMBER 2006
Q1040 David Taylor: When the collective
decision was taken to go for the dynamic hybrid do you say that
an analysis was made of the pros and cons but the categories themselves
were merely analysed in a qualitative and not quantitative fashion?
You had a critical path, but it did not seem to be a particularly
rational way to come to a final decision on something as important
as this.
Sir Brian Bender: Ministers were
advised on the greater simplicity of implementing the historic
model and the greater complexity of implementing the dynamic hybrid,
but, to come back to the Chairman's previous question, at no point
were they advised that the latter was sufficiently complex that
it would be very difficult or impossible to do and the then Secretary
of State said, "Well, I want you to do it anyway." The
advice was concerned with the degrees of complexity.
Q1041 David Taylor: Did you attempt
to assess the skills, people and resources necessary to implement
the dynamic hybrid and compare it with what you had on board,
and looking back did those estimates woefully undershoot the true
position?
Sir Brian Bender: I cannot remember
whether you were in the room when I said that one of the issues
that arose subsequently—the Sub-Committee has pursued this—was
the loss of staff and skills along the way. As I think I said
in answer to a previous question, throughout I was saying to the
RPA that if there was a conflict between achieving the efficiency
targets and delivering the single payment system it should come
to me because SPS delivery was the priority and we would have
to work out with the Treasury how to handle it in the returns.
One of the lessons from all this was whether the RPA understood
the productivity, if you like, of its staff using the new system.
Obviously, with hindsight it did not. I have rambled on and am
not sure I have answered your question directly.
Q1042 Mr Williams: I can understand
why politically the decision was made fully to de-couple and go
for a dynamic hybrid, but pragmatically there was the option which
would have complied with the regulation of postponing the scheme
for a year. Was that ever considered and, if so, in what depth?
Sir Brian Bender: Ministers decided
quite early on to go for 2005, so the context in which the advice
was put was the degree of difficulty in going for 2005. Clearly,
they could have decided that it was sufficiently difficult and
the risks so great that it would be safer to go for 2006. They
were not advised that that was an avenue to pursue and, therefore,
they did not cross-examine us on it. It comes back to the advice
given to them.
Q1043 Chairman: I ask about terminology.
You have used the word "risk". Can you quantify that
for us? Putting it in betting terms, what were the odds on getting
there in terms of the different models? Was it 10 to one on delaying
it for a year and then going for a historic name change and eight
to one against for the dynamic hybrid but you would still have
a go because you thought it was a good runner?
Sir Brian Bender: It was not quite
like that. There were two types of risk.
Q1044 Chairman: How did ministers
know what the word "risk" actually meant?
Sir Brian Bender: There were two
ways in which we used that word in this process. The first was,
if you like, just a risk register. What are the big and significant
risks of failure, and are they moving into the worrying and dangerous
territory or back down? For example, at one point the risk of
industrial action in the Rural Payments Agency on the issue of
comparative pay across agencies and core Defra went up the risk
register. There was a risk register on the question whether it
was at an acceptable level and it could be pushed down. Therefore,
there were individual items. There was then a critical path which,
based on RPA, led us at various stages to consider and challenge
its assessment of the probability of making the payments successfully
at a particular point in time. Usually, that calculation was around
60%. At one particular point in the summer of 2005 that fell to
40%.
Q1045 Chairman: But when ministers
had before them the options was there a score to show that the
probability of delivering the dynamic hybrid within the desired
timescale was x%?
Sir Brian Bender: I do not recall
it being quantified at that time in that way; in other words,
ministers were advised that it was more complicated but it could
be done early in the payment window.
Q1046 Chairman: Therefore, ministers
were blind. What they had was a subjective assessment as opposed
to a combination of subjective and objective assessments when
they made their decision.
Sir Brian Bender: They had an
assessment that was subjective but with the elements of what the
particular problems might be. It was not put on the basis that
if they went that way it would be x%; if they went the
other way it would be y%. It was not put to them in that
way at that point as far as I can recall.
Q1047 Chairman: Do you think it should
have been?
Sir Brian Bender: With hindsight,
yes. I know that one of the questions in all this is the extent
to which ministers should have taken the decisions they did. Putting
it in the negative, they were not advised in terms that meant
they made the wrong decision. With hindsight, there is a question
about how the department and agency might have quantified the
set of risks to help them understand better, but I come back to
the point that it was not necessarily the wrong decision because
right the way through until January 2006 there was optimism that
the payments would be made.
Q1048 Lynne Jones: Was there ever
any feeling that irrespective of what advice was given to ministers
as to the difficulty and complexity of administering the dynamic
hybrid they were determined to go down that route anyway?
Sir Brian Bender: I am just trying
to recall the flavour of it. They were attracted.
Q1049 Lynne Jones: Because if that
was the case maybe that was why the messages were not being conveyed
to them.
Sir Brian Bender: They were attracted
by that route. Nonetheless, it is the role of a civil service
to speak truth unto power and expose these issues. Clearly, with
hindsight one of the questions is the culture in different parts
of the organisation that fed anxieties properly up the line. But
I certainly did not experience any fear of exposing these issues
to the Secretary of State. Obviously, what I do not know is what
people felt particularly in the second tiers of the agency in
Reading.
Q1050 Mr Williams: When Accenture
was first engaged it had to undertake the change programme and
deliver the schemes already in place or upgrade the way in which
they were delivered. With the advent of the single farm payment
and choice of the most complex scheme, was this all really too
much? In hindsight we can say that it was all too much.
Sir Brian Bender: All too much
for Accenture?
Q1051 Mr Williams: Yes.
Sir Brian Bender: I do not believe
it was. If I look back at the RPA experience with Accenture, there
were some performance problems early in the first stage where
my sense at the time was that it had not moved quickly enough
from sales to delivery. It had leadership of an inexperienced
team that gave rise to some deficiencies in quality of testing.
We addressed those in a number of ways. The RPA did it. But that
was when I began to have contact at my level with the chief operating
officer of Global Government Services. I then had those contacts
regularly. On a monthly basis I had a meeting with him in addition
to whatever other meetings were taking place in which Accenture
was involved. Nothing that I saw suggested that the overall task
was beyond it, but one of the things in respect of which we kept
on its back was to make sure this was seen corporately in the
company as a sufficiently high priority in terms of its reputation
to put the right resources into it. I remember at one point the
chief operating officer of the company came to the UK and saw
the cabinet secretary. I made sure that that was registered by
Andrew Turnbull with him as well as in a meeting that I had with
him. There is always an anxiety, as some Members of the Sub-Committee
know better than I, about whether an IT supplier has the right
skills which are relatively scarce if it is trying to manage a
number of contracts. There were times when I was worried about
that. What I would say is that when those worries were registered
personally at that man's level he took action on it. I had no
indication that the task was beyond it. Indeed, subject to the
particular difficulties over the maps, it delivered what it said
it would do, albeit sometimes with some slippage.
Q1052 Mr Williams: Although different
elements of the computer system were tested, it became apparent
fairly soon that the whole system could not be tested in a timely
manner. Was there any period then when ministers or the Secretary
of State could have said, "We really want to abort this and
revert to something about which we are more knowledgeable and
have experience of delivering"?
Sir Brian Bender: We are now getting
to the stage after I left Defra, but I will try to join up the
two bits. Individual elements were tested and with various hiccups
they worked, so the central question you are asking is why there
was not a whole system test. One answer to that is that with hindsight,
knowing that the system gummed up, to coin a phrase Helen Ghosh
used in May, plainly we would have done that. But I think there
is an interesting question whether an IT system failure or the
compliance and restrictions checks put into it by the users caused
that to happen. I come back to my earlier point that one of the
interesting things Mark Addison did pretty quickly to release
quite a lot of money was to remove some of those restrictions.
But with hindsight I guess that we should have ensured that there
was an all-system test. It would have had implications for the
date because the only time it could happen would have been around
the turn of calendar year 2005-2006. That was after the time I
was in the department. That would have been quite a difficult
decision to make when there were no signals at that time that
the individual elements were not working.
Q1053 Mr Williams: What stopped a
full testing of the system? Was it that the system was not in
place or there were insufficient validated claims on which to
test it?
Sir Brian Bender: The whole system
could be tested only when the last element was in place, so we
are talking about a period when I had left the department. We
are talking about the autumn of 2005 and going into 2006 with
the validated claims. I think Simon Vry tried to answer some of
these points last week. Trying to piece together what happened—it
was after I left the department—there was no reason to believe
it was necessary. With hindsight, it plainly was. Had it been
done it would have led to some delay in the payments.
Q1054 Chairman: You were having these
nice discussions with the cabinet secretary and laying it on the
line to the top man from Accenture how important it was. You were
in post when the contract was signed and the specification for
the system was drawn up. You were in post in May 2004 when the
Accenture contract with the RPA was renegotiated. Are you telling
the Sub-Committee that at no stage in any of these discussions
was there a line in the contract which said, "We, the buyers,
expect you, the provider of the system, to have fully tested it
to make certain it works within the timescale that you have agreed
to do the work"?
Sir Brian Bender: I do not know
the answer to the question about exactly what was in the contract.
What I can say is that there was a lot of discussion about the
testing of the individual components and what risks might be taken
by release earlier or later.
Q1055 Chairman: When Accenture came
before us it obviously very well schooled in not saying anything
beyond the area of its agreement. It kept falling back on the
line that the system was stable and it had tested the ingredients
that went into it. It is a bit like somebody saying that he has
an idea for a thing that has four wheels which may move but he
has not bolted it all together to see if it will turn into a car
and go forward. Therefore, I should like a check to be made on
whether there was a requirement in the contract that it should
be tested in the round, ie that all the bits worked together,
before the system went live. All I am being told is that there
was a lot of discussion about the testing of the individual components
but not the system as a whole. Mr Williams has been asking why.
I would have thought that a pretty fundamental part of the contract
was that it should work.
Sir Brian Bender: I will talk
to the department and get back to you on it. Accenture was a member
of the programme board and so was part of the wider discussions.
I do recall discussing with the man from Accenture not only whether
it was delivering on time what it had been asked to deliver but
the RPA's capability to implement it, in particular its productivity
in using the Accenture system.
Q1056 Chairman: They are all sitting
there in the RPA and they know that the Secretary of State is
coming to the point of decision. A decision is made to go for
the dynamic hybrid model and a pretty tight timetable is agreed
because ministers are talking about a payment window opening in
December, possibly paying in February. They did not take the full
payment window to June in any public statement. All these guys
are sitting back in Reading along with Accenture thinking, "This
is a bit much. What can we do? We do not want to say we cannot
do it because our jobs are on the line." Mr Neill says that
his bonus is on the line and asks Accenture what it can do to
take some time out of it. Accenture turns round and says that
if it tests each bit and marries it together as it goes along
x weeks can be taken out of the critical path and there
will be a system that it thinks will work because it is the expert.
Was anything like that said?
Sir Brian Bender: I have two comments
on that. First, I recall discussing with the company not only
the release of individual elements but how it would work together
and whether the agency could make it work together. It was as
a result of those discussions that we started getting the Executive
Review Group data on RPA staff productivity and using the system
as it came out. That is not exactly the point you are asking.
To come back to the point that I am trying to make, this is not
simply an IT issue; it is the productivity of the RPA staff in
using the IT, and that was where I think there was a big issue.
Q1057 Chairman: This is one of the
points about the relationship with Accenture which I find fascinating.
Accenture are very expert people with considerable world expertise
in the way that complex interactions occur between the IT systems
that they are designing and the environment in which they will
operate. I find it unbelievable that they seemed to maintain a
Trappist-like silence in commenting on what they saw going on
around them which reflects entirely on the point you have made
about the productivity of the system. Did Accenture interact to
your knowledge with anybody in the RPA, the Executive Review Group
or CAPRI to express a scintilla of doubt that what was going on
around them would not happen to enable their system to work properly?
Sir Brian Bender: The one area
that it did register both in writing in meetings was that successful
implementation would depend on the RPA's productivity in using
the system.
Q1058 Chairman: When did it say that?
Sir Brian Bender: More than once.
Again, we can try to find out and let the Sub-Committee know.
Q1059 Chairman: What was the reaction
to the warning that if the rest of the system did not work hard
enough there would be a problem?
Sir Brian Bender: A lot of the
reaction was about what the RPA's own staffing productivity was
and what the issues might be to improve it.
| 
