Examination of Witness (Questions 1080-1099)
SIR BRIAN
BENDER, KCB, CB
6 DECEMBER 2006
Q1080 David Taylor: From whom were
you getting that advice?
Sir Brian Bender: Primarily from
the RPA, but we were obviously challenging that advice. If I had
my time again what would I do differently? I think I would go
with Helen. There are two issues. One would be to cross-examine
very hard the reliability of the particular timetable dates, but
the second would be to go very hard on the interim payments, because
once one got significantly beyond December and through the payment
window it would be a greater hardship to farmers and any diversion
of effort in making interim payments would have to be weighed
against that. With hindsight there are other things that I might
have done differently, to which I can turn if the Sub-Committee
wants, but that is my answer to the specific question.
Q1081 Sir Peter Soulsby: Do you think
that behind the decision to go that way was too much focus on
the risk of disallowance and not enough on the other risks in
the project?
Sir Brian Bender: I tried to make
that point earlier. I was trying to give signals on a number of
occasions, that I thought I was getting across clearly, that the
RPA needed to balance appropriately the risk of disallowance against
the risk of failure to deliver. In the end, they got the worst
of both worlds. As Accounting Officer potentially I was prepared
to take a higher disallowance hit, provided it was carefully understood
and managed, in order not to screw up the payment to farmers.
With hindsight, particularly given what Mark Addison achieved
in his first few weeks in terms of relaxing some of the restrictions,
it does not seem to me that that message was received sufficiently
clearly at all levels in the RPA. One of the lessons that I draw
from that is that if you think you are saying something, even
if it is said in writing and in minutes, you follow it up more
effectively. I think that is one of the lessons. I believe that
the culture of the RPA was turning out to be too focused on compliance
and avoidance of disallowance, which is important, and not enough
on the business processing and payments to farmers and getting
that right. I thought I was making it clear; I said it several
times privately to Johnston McNeill and in meetings recorded in
minutes, but the fact that immediately when Mark Addison got there
he lifted some restrictions and released some money in my view
confirms what I have heard anecdotally since March that culturally
the balance was wrong.
Q1082 Sir Peter Soulsby: I have a
similar question on the focus of the RPA at that stage. Do you
think it is true to say that it was overly concerned with the
IT system at the expense of the wider business processes of which
it was a part?
Sir Brian Bender: As I have been
trying to say at various points in this hearing, obviously one
cannot separate the one entirely from the other. But at heart
I do not believe that what went wrong was simply the IT; it was
the capability and productivity of the agency in understanding
the IT and using it productively to get the money out the door.
These things are blurred, and I am not trying to mince words.
What I am saying now is based essentially on conversations that
I have had with people since March, but I believe that the compliance
side of the organisation had too high a profile and, if you like,
the business processing side—that the agency was actually
there to make payments to farmers—did not have a high enough
profile. I say that with the benefit of hindsight and anecdote.
Q1083 Chairman: Was there ever a
time when you as permanent secretary put a note to the Secretary
of State asking for consideration to be given to interim payments
given the worsening situation in 2005 to try to give yourself
a bit more time to sort out the overall process problems?
Sir Brian Bender: It was not done
in those terms, but there was discussion with ministers about
the pros and cons of pressing ahead with interim payments and,
as part of that, the policy side of the department went to Brussels
to try to get legal cover for doing it, because the legal position
was a bit blurred. I would say that in the summer/autumn of 2005
it was under active consideration.
Q1084 Chairman: Did you make any
kind of international comparisons? When the Sub-Committee went
to the German agriculture ministry it learnt that it had three
times as many customers, four separate computer systems and 18
different Länder involved in the process. Because
of the complexity of what they were doing it was clear that they
had to adopt an interim payments strategy in order to get over
the problem and make certain that farmers were not left in the
lurch. Given that you had only 120,000-odd customers and you were
on a par with that complexity, did that ever ring any bells?
Sir Brian Bender: These issues
were discussed in the department—I cannot remember the exact
date—in my time. I concur with Helen that with hindsight
this was something that we simply should have done, as opposed
simply to discussing it and putting a little bit on the back burner
in the belief that the bulk of the payments would be delivered
by the end of March. I agree with Helen and with hindsight had
I been there throughout I would have done it differently, too.
Q1085 David Taylor: Have you had
an opportunity to look at the NAO report?
Sir Brian Bender: Yes, I have.
Q1086 David Taylor: You note what
it said about Accenture falling short of expectations in the early
stages of the new programme once Defra decided to implement the
CAP. You will also note that the OGC expressed concerns in January
2005. At what point did you leave the department?
Sir Brian Bender: I left the department
at the end of September 2005. I am happy to say a bit more about
the OGC.
Q1087 David Taylor: To finish the
quote, it expressed concerns about significant weaknesses in the
management of the testing team which was crucial. Do you want
to say something about that?
Sir Brian Bender: As to the management
of the testing team, this was something that I personally took
up, as did others, with Accenture and they strengthened it, but
there were problems at the early stages. They corrected it but
that led to some delays and difficulties at that stage which I
believe was in 2003 or 2004. As far as concerns the OGC, I think
that there were lessons, which I had been discussing with John
Oughton, as has Helen, about what this exercise told us about
the Gateway review process. As the Sub-Committee knows, there
were regular Gateway reviews. In each case as I understand and
recall it all the recommendations that the review team made were
acted on and in no case did the OGC say to me or anyone else it
was reporting to, "Stop. This will fail." The question
is the extent to which it was looking at a broad enough or too
narrow a picture. I think there are issues that we are discussing
in government about what lessons this has for the whole Gateway
review process. But as late as September—my last month in
Defra—the last Gateway review contained the sentence: "We
are convinced that the programme has a reasonable chance of delivering
SPS payments within the 2006 window", which for cautious
people was quite encouraging. Obviously, it did not say that it
would deliver in February/March.
Q1088 Lynne Jones: How did it gather
evidence that would lead it to that conclusion?
Sir Brian Bender: The OGC Gateway
process essentially involves some experienced peer reviewers having
a whole series of interviews with different players. But one of
the questions is: what does this episode show for the breadth
of the Gateway review process in what is a complex and high-risk
programme? My sense, again with hindsight, is that it was a bit
narrow each time and not dynamic; it did not look back and say,
"Hang on. Can you do all of this?" which was something
on which we needed advice.
Q1089 David Taylor: You said that
it was part of the decision process. I do not want to go too far
into what was said earlier, but a detailed critical path analysis
was made of the whole project which seemed to suggest to all concerned
that on balance it was feasible and there would have been some
slack and float in it. It would be predicated eventually on the
opening of the payment window on 1 December. In your time the
start window was moved to February, was it not?
Sir Brian Bender: That is correct.
Q1090 David Taylor: At the time that
you effortlessly glided out of the department to your new empire
in Victoria Street, as it were, presumably you would have looked
at the present state of the critical path analysis because all
of the slack and float would have disappeared by then, would it
not? You had a very tight system.
Sir Brian Bender: Yes, it was
very tight. The last Gateway review in September had the sentence
in it which I quoted. I had a meeting in my last couple of weeks
both with Mark Addison, who by that stage was going to be acting
permanent secretary before Helen arrived, and the independent
challenge person Karen Jordan. My recollection is that she did
share some concerns but still expressed the view that it was more
than possible but do-able. I also made it very clear both to Margaret
Beckett and Mark on leaving that this needed to be one of the
continuing high priorities because it was difficult stuff. You
used the phrase "glided out". This was an issue to which
I gave a lot of attention in my time, and my parting words both
to Margaret Beckett and Mark Addison were that this deserved quite
a lot of attention because it was high-risk stuff. But the advice
we were getting was that it was still on track. We then come back
to the position in January 2006 when the public statement was
made that not only was it on track but it would be achieved.
Q1091 David Taylor: Accenture said
to you at some point that all was set fair but it had a cautionary
note that it depended on, to use your words, the productivity
of the RPA staff and their ability to use the system that Accenture
had supplied. To simplify it a little, was it not the case that
because it was a hybrid system which incorporated an averaging
element to pay any claim you needed to have the great bulk of
claims in so that you could average it out? Was that not a serious
flaw in the sense that you could have had a moving average? On
the basis of the first 20% of claims processed it would be very
likely, assuming that one had chosen the cross-section intelligently,
that the average calculated at that time would be very close to
the overall average when 95% of claims were in.
Sir Brian Bender: I understand
the question you ask. Part of my answer goes back to the answer
I gave to Sir Peter a few minutes ago. We come down to the balance
between the disallowance risk and delivery risk. On the basis
of the great gift of hindsight, there is no doubt that the balance
was wrong and in effect that was what happened with the interim
payments.
Q1092 David Taylor: I am not sure
that that is hindsight. I am sure that it would have been possible
for the people in charge of the process at the time to deduce
that a very risky element of the whole process was the need to
have the vast bulk of claims in before any payment was made in
respect of the averaging element. Surely, could not someone have
picked that up and incorporated it into the business processes
being utilised?
Sir Brian Bender: Yes. What I
do not recall are discussions about that in 2004. I do recall
in 2005 discussions about it and the options for dealing with
it, including interim payments and other things that I described
earlier.
Q1093 David Taylor: That gave Accenture
a get-out-of-jail card, did it not? Accenture has repeatedly said
to us that it delivered what it was asked to do. That suggests
that the RPA and your overall leadership were taking on the whole
design of the business process and IT specification and Accenture's
approach was merely to say that it did what it was told.
Sir Brian Bender: I think Helen
Ghosh discussed some of this at the hearing that I attended with
her in May. One of the fundamental questions must be: did the
RPA understand at a sufficiently senior level the systems that
it was designing or asking Accenture to design and what the implications
were? I suspect that part of the answer to that comes back to
the balance between compliance and the business process.
Q1094 Lynne Jones: To return to the
quotation that I read out earlier relating to the discussion with
Accenture, I referred to the wrong page. There was a definite
"no" to the question about Accenture having any input
on whether the system chosen for CAP reform would have any significant
impact on the computer system being developed. In the evidence
it is definite that it was not invited to comment on the impact
on the IT systems which, as I said earlier, I find extraordinary.
Sir Brian Bender: I find that
odd to say the least. If after this hearing there is any light
that I can shed on it I will do so.
Q1095 David Taylor: To come to the
point at which de-scoping was decided upon, both Lord Bach and
the NAO referred to the optimism of the RPA's self-assessment
of the chances of meeting the target for making payments by the
revised date of end of March. Your successor told the PAC on 30
October that advice to ministers was based on "over-optimistic
interpretations of perhaps inadequate management information .
. .The `de-scoping' of the management information systems meant
that the RPA did not have good information on the progress they
were making." Do you agree that the fact it had been de-scoped
immediately weakened the reliability of such management information
that there was to measure progress?
Sir Brian Bender: I understand
that to be the case. Looking at my own checklist in terms of what
I think the lessons are, one of them must be not only whether
the agency understood the tasks that it had to take on but whether
it was able to have the necessary management information. It is
an obvious point, but it was creating, not deliberately, a misleadingly
rosy picture. The question is why. Was it simply because of the
de-scoping? I am sure that it was more complicated than that.
I do think that some of it came back to whether or not it understood
enough about its own business processes and staff productivity,
because that is also linked to the fact that, despite what I said
to the RPA on several occasions about SPS delivery taking priority
over efficiency savings if necessary, we were advised that it
was not necessary to retain the extra staff. There is no doubt
that de-scoping played a part in this, but I think there is a
wider issue about whether or not it had enough understanding of
what was going on within the organisation and its own staff productivity
that allowed it to keep saying there was no problem about releasing
all the staff.
Q1096 David Taylor: You are talking
about them in the third person. Is that a subtle distancing of
the RPA from your overall supervisory role?
Sir Brian Bender: I was overall
responsible. One comes back to the principle of executive agencies.
This was an arm's length body that was accountable for the implementation
of the policy under the 1980s Next Steps Agency model. Therefore,
one gets to the question: how do I and ministers get the right
assurances without crossing the line and blurring accountability
by second-guessing and backseat driving?
Q1097 David Taylor: I fully understand
that it is a tricky line to walk, but this was a crucial project
which was hitting very choppy water, which you had picked up.
You told us that you de-scoped some of the IT so that the tasks
set would be less difficult. I am sure that it was de-scoped with
good intent, but it has proved to be a pretty short-sighted decision,
has it not?
Sir Brian Bender: I am sorry to
repeat myself, but what I do not have a feel for is the extent
to which the inadequate management information—plainly,
historically it was—resulted from that de-scoping or wider
issues. I do not think that it was that simple. It is certainly
the fact that there was not the right management information coming
through. I am not trying to shuffle that off, but there is a question
about how much someone at central department level can appropriately
become involved in what is going on in the interstices of an arm's
length organisation.
Q1098 David Taylor: The de-scoping
involved the stripping out of the management information software
from the IT system, did it not?
Sir Brian Bender: Yes, but I come
back to the point that there was more to what was done in the
RPA than simply that decision.
Q1099 David Taylor: You signed off,
endorsed and approved the de-scoping recommendation without necessarily
being fully aware that one of its implications was that the information
available to monitor progress and test performance would no longer
be available in any reliable form.
Sir Brian Bender: I would not
put it exactly like that, but essentially I did not have an appreciation
of the quality of the management information that would be coming
forward. I had assumed from everything I was told that we would
be getting enough management information to know what was going
on. One then gets to the question: how much intrusive management
does one do in these issues? With hindsight, the answer must be
more.
| 
