11 Belgrave Road
London

IBM PENSIONS

Dear .....,

This letter is about how far BMNN narrows or expands the field of contention that the Ombudsman's Office may want to investigate.

BMNN is helpful in agreeing that trustee decisions could have affected those interests (whether or not they actually did).

BMNN acknowledges that the trust had some powers that could have been used to balance company and member interests, including the power to not allow M-plan contributions to be allocated from the Defined Benefit funds. This brings to focus the issue of whether those powers were appropriately used.

BMNN acknowledges, as far as I can tell, the trustees' legal duty to be aware of and to take account of the norm - as I quoted it from case law:

"They must also consider the level of benefits under their scheme relative to the benefits under comparable schemes; or in the pensions market generally."

This is made important in this investigation by the unusual nature of the trust's review procedures and the gap in outcomes between this trust and the norm.

Some parts of BMNN just repeat my views and say that the trust does not agree. This is to be expected and presumably helpful for the record.

Some parts of BMNN are what I will call "diversionary topics", where instead of responding to my subject BMNN introduces a different subject and comments on that. This is harmless, except insofar as it leaves my topic undiscussed.

BMNN leaves a number of things undiscussed, in particular with respect to IBM. Correctly, BMNN says that the complaint about IBM is not covered by the IDRP. Incorrectly, BMNN extends that to a belief that it cannot say anything about IBM in its response to my complaints about the trust.

The trust has said:

"The trustee has regard to everything that employees were told about pensions by IBM UK or the trustee."

My complaint against the trust pointed out the failure of the trust to do that properly:

"The previous nature [of IBM] relates to expectations that the Trust should have taken into account - employees who are well treated are anticipating being well treated when they make their retirement decisions."

A second connection between IBM and my complaints against the trust is in the possibility that:

"The decisions about what IBM would approve were made outside the trust meetings and carried into the meetings."

A third connection between IBM and my complaints against the trust relates to statutory surplus. In my penultimate paragraph on this subject in my stage 2, I raised the issue of whether it was right for the trust to assume continued increases, in the light of IBM attitudes. BMNN does not comment on this.

Since there are these connections between IBM and my complaints about the trust, it would have been correct for the trust either to deny the connections or to consider the IBM behaviour in the IDRP. It did neither, except via blanket rejection of my complaint.

My first complaint was:

[You might want to establish whether I made a mistake, somehow sending the trust a different version of this complaint. BMNN misquotes the above, omitting the word "solely" and omitting the second sentence which specifies what I complain about. I am in favour of the trust maximising benefits to the IBM balance sheet, subject to a constraint of balance with member interests. I affirm for you and the trust that my complaint is about misdirection - the use of a wrong approach in making decisions.]

About the misdirection complaint, BMNN asserts:

"This appears to be wholly based upon assumption and is refuted."

I refute this assertion. There is no science that allows direct evidence of a person's thoughts. When one person describes the thoughts of others, as Barrie Morgans personally does in BMNN, and I do in my complaint, then they are using knowledge of circumstances, sayings, writings, and actions. This is inference and deduction, not assumption. It would be better if my deductions had extra roots in the voting records, documents from previous times, and dialogues with individual trustees. The trust knows that complainants do not have access to these, so to emphasise the absence of them in IDRP is a "cheap shot".

For the misdirection complaint, I specified seven sets of decisions to be considered in support of:

" I have found that there is a pattern of the Trust considering, and implementing, changes that would benefit IBM, with no balancing considering and implementing of changes that would benefit the members."

On none of the seven does BMNN say that the decisions taken (in contrast to the alternative decisions specified in my complaint) were other than what IBM wanted. On none of the seven does BMNN say that the decision taken (in comparison with the alternative decision specified in my complaint) was beneficial to the members, although in one case it says it might have been. In one case it says the trustees had no choice. This is partial confirmation of the pattern that I claimed.

Investigation of the components of the complaint BMNN does not address will strengthen the evidence of pattern. It is a matter for any adjudicator to investigate and decide whether the overall pattern is established to the extent of supporting a complaint of misdirection.

There is one case where BMNN invokes what I will call the "Omnipotent Ogre" explanation of its behaviour - that if it had made a decision other than what IBM proposed then IBM would have (although IBM did not say so) reacted in a way that moved the balance further to the detriment of the members and the trust would not have been able to correct that balance, then or later. This does not seem to be essentially different from my view that:

".. the wrong principle the trustees were applying was that what IBM wanted should be agreed to if the trustees considered IBM could conceivably do something, legally defensible, which would damage the members "

Here is detail on the seven sets of decisions. I repeat the headings, with heading numbers added.

I wonder if BMNN is using "annual review" in the sense that I do, and I believe the occupational pensions industry does. In the pensions industry it means review at yearly intervals and usually implies announcement of the result of the review. The trade press periodical "Occupational Pensions", in its 1999 survey says:

"IBM reviews the award of discretionary increases at intervals of 15-18 months."

"Occupational Pensions" gets its information by asking the scheme administrators for it.

I said that:

"Review periods longer than a year disadvantage members because they lose out in the period after the elapsed year and before an increase is considered."

This is a matter of mathematics, not an opinion. We all know that a 60% return on capital after five years is not as good as 12% annually compounded which is not as good as 1% monthly compound. Longer periods under the "same" rules disadvantage the recipient. BMNN does not contest that.

The disadvantage applies whether the rule is "minimum £5 per month", "percentage of RPI", "percentage of a year's RPI pro-rated" or any similar rule. To this extent the BMNN discussion of n/12ths is a diversionary topic, although it is relevant to item 6 below.

2. The decision to allow Defined Benefit plan funds to be used to fund Defined Contribution plan liabilities.

The decision I wrote about was (with emphasis added here):

"The decision to allow the mechanism to be used was to detrimental to members because, in comparison with the alternative of allowing the M-plan and not allowing the transfers of funds, it reduced the reserves...."

BMNN discussion comparing with some other alternative, involving not having an M-plan, is a diversionary topic.

BMNN says:

"... one cannot assume that there would have been increased security and an increased likelihood of augmentations to benefits if the Trustee had not allowed M Plan employer contributions to be allocated from the general fund."

Maybe one cannot assume this, but it can be deduced. The M Plan assets have to come from somewhere. If they do not come from the Defined Benefit funds then the assets of the Defined Benefit plan and the M plan taken together are higher than if they do come from the Defined Benefit funds. More assets, same liabilities, equals better prospects for members.

I note that, in replying to Dave Mitchell on the matters relating to the M-plan, the trust adopted the "Omnipotent Ogre" line by suggesting that if the trust had not agreed to IBM's plans then IBM would have reduced its M-plan contributions (although IBM did not say that). It has not made that suggestion in BMNN. Is that significant?

I did not say whether the disclosure was lawful or not, I said:

"The Trust chose to tell the retirees the minimum necessary to make the Members' Report accurate in fact, rather than to appraise the retirees of the change in their prospects."

BMNN says:

"The Trustee is satisfied that it has complied with its legal requirements to disclose information to members."

"Compliance with legal requirements" is a low ambition in terms of keeping members informed but I do not think it has been achieved. I do not know about the case law but I do know that in interpreting regulations the law puts a high emphasis on the purpose of the regulations. The purpose of the disclosure regulations is to allow the members knowledge of what they might reasonably think affected their prospects, and thus allow them to make justified challenges.

It was not the trust's prerogative to pre-empt the question of whether the position of Defined Benefit members was, has been, or could be, detrimentally affected. They had an obligation to give reasonable information on which members could assess that question.

The introduction of Defined Contributions in place of Defined Benefits is a major structural change, the biggest in the history of IBM UK pension plans. The funding of a major change is always going to be a matter of interest. The retirees do have reasonable cause for concern, at least in the view of OPAS, irrespective of the outcomes of the current complaints. Is the trust suggesting that the trust gave no consideration to the possibility that members would want, and deserve, a more conspicuous approach to introducing the change?

Of course, the members can eventually learn at least as much (by complaining and other methods) as they should originally have been told. It is in the meantime that they have been disadvantaged.

It was not the trust's prerogative to pre-empt the question of whether investment via AVCs was made more attractive, less attractive, or unchanged. They had an obligation to give reasonable information on which investors could assess that question.

If a private company taking investments delivered a sparcity of information in the way the trust did, then there would be public outrage.

The extent of any "discussion" is disputed. Barry Morley wanted a discussion/review/debate which he did not get. Barry Morley is the person most concerned with this, and the best source of information.

I have no doubt Barry Morley sought information over a period. BMNN says that the administration did not seek and obtain originator's permission for some of the information to be given, until after Barry had resigned. I suppose this might have been due to overwork, but reluctance is a more credible explanation to me.

When this item 4 is investigated, the knowledge of B K Morley and B I Petch is the best source of information.

This is the one case of the seven where BMNN claims its action might have benefited the members:

"At the time the Trustee was mindful that proposing a higher increase might have been counter-productive."

This is the Omnipotent Ogre argument. IBM might not have agreed to the proposal, but what possible grounds could IBM have for acting against the members simply because the trust proposed that members of the IBM scheme should have their interests given the same weight as was the norm? The trust and IBM were aware that the pensions-in-payment history showed IBM members significantly behind the benefits enjoyed by the vast majority of comparable members. The trust was aware that IBM had no financial difficulties. (The IBM share price, with stock splits considered, has increased eight-fold in the decade. The amount from IBM enjoyed by just one particular US employee, in one year, would pay for roughly half of the cost of bringing all the members of the IBM UK DB scheme up to the norm, for the full lifetime of the scheme, according to the model in the 1997 actuarial report.)

If the trust contemplated such a move being made by IBM against the member interests, why did the trust believe it could not be countered by its own powers, or balanced by its actions in some other area? The trust has said, with respect to the situation in year 2000:

"In view of the expectations that have been raised amongst retirees, in the event that increases were not at the level of 70% the Trustee would wish to understand the company's reasons for this and would, therefore, need to consider what options the Trustee, itself, might take."

If the trustee can consider, in year 2000, options to counter IBM taking members further from the norm, why would it not do so in 1997? Was legal advice on this solicited?

The "Omnipotent Ogre" motivation, if present, was ill informed or irrational in this instance.

I wrote:

"...when 514 people from Data Sciences transferred into the Defined Benefit plan, the "surplus" in the plan, per member, was reduced. Reduced reserves per member are detrimental to members' interests in the same way as reduced reserves are."

BMNN does not contest that this decision was detrimental to members. It says that the trust had no option but to accept this IBM decision. I agree that, in view of the wording of the deeds, the trust could probably not have successfully used its powers to resist this decision.

However, the trust could have balanced this detriment by other actions that improved the position of members. BMNN does not suggest that it did.

BMNN says I was not clear about some topic. This is correct because I did not address that diversionary topic. The reserves per member were reduced, irrespective of that topic.

BMNN says:

"The Trustee considers that the practice of reviews to pensions in payment and the granting of discretionary increases falls within the paragraph 5(2)(c)(ii) which you quote."

The relevant paragraph and the preceding one read:

(i) the valuation rate of interest shall be reduced to take account of pension increases at the rate provided for under the scheme, provided the net investment yield is not so reduced to an amount less than 3 per cent. per annum;

Two alternatives need to be considered, depending on whether "provided under the scheme" is interpreted as meaning provided under the scheme in a formal way, or provided in practice. I suggest the former interpretation is correct (and matches "provided for under the scheme" in paragraph (i)). However, I discuss both alternatives below.

Prior to stage 2, I had asked the trust (by email) how 5 (2) (c) (ii) could apply and was directed to a section of the deeds. The deeds refer to events happening "from time to time". In my stage 2, I asserted that "from time to time" did not imply "regular", which the regulations are using in the sense "calculable in time". (It may even imply "irregular").

BMNN does not reference the deeds. I infer that the trust no longer wants to claim that the deeds formally provide for regularity.

IBM pension increases are, in practice, irregular. There is a table (which has been agreed with the trust) in the appendix to this letter which demonstrates this.

I do not have access to the trust's minutes, to find out if they record reviews of potential increases at constant intervals.

I would expect the regulations to be using the term "reviews of pensions" in the same sense as the occupational pensions industry uses it. The pensions industry does not regard IBM reviews as regular; see item 1 above.

I have received two descriptions of the trusts review process and the calculations involved. They are not the same; see the appendix.

Examination of the description of the review process given in BMNN, together with any further details that can be obtained from the trust, together with the appendix to this letter, together with consideration of normal pensions industry terminology, will lead to the conclusion that the description in BMNN is wrong, or at least confusing.

BMNN says explicitly that it will not address these decisions. I specified these decisions by reference to a document that the trust already has. I do not see why this makes a logical difference from stating them explicitly in full in my phase 2 submission.

Item 7 is just as much part of my complaint as the other six, just as much part of the IDRP, and just as much part of the grist for the Ombudsman's Office mill.

[I assume the Ombudsman's Office has Barry Morley's resignation letter as part of its investigation into the Mike Cawley complaint. If it does not, it is able to get it.]

I repeat that:

" The pattern of behaviour will not have been properly studied until these items are taken into account."

With respect to when decisions were made, BMNN says:

The decision about funding the M-plan could not precede a decision to have an M-plan - this is an assertion I have made in my stage two complaint and it has not been contested. So the statement above implies the decision to have an M-plan was also made in 1996.

The statement that I found unclear was the one I quoted:

"The introduction in 1997 of the M Plan within the existing trust was discussed and agreed at its meeting of 12 December 1996 and at its meeting of 24th April 1997, when the new definitive trust deed and rules was (sic) adopted."

The information, new to me, that both decisions were made at the first meeting, requires me to alter the scenario I gave of the two meetings to:

Parliament had made provision for elected directors. The purpose of this regulation was to reduce the risk that trustee meetings would be unduly dominated by company interests. The first meeting occurred before the provisions came into play, the second after. The second meeting had a majority of company appointed trustees (some new), and the statutory minimum of newly-elected trustees.

The first meeting decided to have a Defined Contribution Plan, to endorse IBM's position of delivering pension erosion that was significantly worse than the norm prevailing for the vast majority of comparable members outside IBM, and to fund the Defined Contribution Plan at the expense of Defined Benefit scheme members' interests in the reserves/surplus then associated with their scheme.

It is still to be decided if this was all legal, but it was certainly an important set of decisions with potential for far-reaching effects. It was known that the effects would have to be managed by a markedly different set of trustees, starting in 1997. The trustees who had to manage the ramifications were not the people who made the decisions. The important question is "Why"? What was the urgency that required the decisions in 1996 rather than four months later?

I note that the trust cannot maintain that the 1996 trustees were better placed to make the decisions because it has said:

"However, all directors had access to all relevant information and advice which had been obtained by the Trustee and your implication that the Trustee in April 1997 was unlikely to have absorbed the factual background and to make an informed decision to proceed with the execution of the documentation is refuted."

Although the new information has forced me to change my description of the circumstances, there is no change to my complaint. What happened is still an instance of the misdirected principle of always doing what IBM wanted.

BMNN agrees there was a flawed document. Since BMNN says it cannot speak for IBM, does that mean the trust is taking all responsibility for the flaw? Both IBM and trust logos are on the document.

BMMN says:

"Your interpretation that discretionary increases might be higher or lower than 70% RPI is correct."

BMNN does not address the circumstances when the increases will be higher or lower but the Trust has described that before in answers to questions. I do not give the full material, but here is the indicative beginning:

'In "IBM Pension Matters number 2", which carries a foreword by the IBM UK Pensions Trust Director, the question is asked "Why can't pension payments be index linked?" The answer given begins "The answer is simple. Affordability."'

The full material reflects the way in which increases reflect circumstances in usual pension schemes. The history for the IBM scheme does not correspond. Why did BMNN choose not to comment on this anomaly? Is it suggested that what the trust said in 1991 was not relevant to at least items 2 and 4 above?

I used this material in my stage 2 as an example of:

"...the expectations that members were given, by IBM and the Trust, that erosion in the value of their pensions would be made less if economic conditions were favourable."

"The trust administration contributed by their unwillingness to work with the elected members of the board and I complain about that."

BMNN says that it believes that this allegation is answered in its responses. It is hard for me to see why it believes that. It did not respond on item 7 above (with the 22 subcomponents I mentioned) at all. Elsewhere I gave a reference to subjects on which trustees received inadequate support and BMNN addresses only the one subject which I used as a "for example".

BMNN purports to address the unfairness I described, but does so without reference to differing pension erosion and the differing structure of the M-plan. These were clearly described as underlying the unfairness I referred to.

The following are important matters which the IDRP did not process effectively:

(a) Whether the trust has the wording of my complaint that I intended.

(b) Whether the trust can unilaterally treat a component of my complaint as if it was not there, as it has done explicitly for item 7. My misdirection complaint is based on a pattern of behaviour. If any component is removed, or taken only in isolation, the pattern is not properly considered.

(c) Whether the trust can unilaterally treat a component of my complaint as if it was not there, as it has implicitly done, to a large degree, for item 2, by addressing a diversionary topic. The decision I was specifying was clearly stated by me.

(d) Whether the trust was right to assume that it could not refer to IBM behaviour in response to complaints against the trust. If this was not assumed, why were the three connections I note above not addressed?

(e) The need to clarify the process that incorporates "n/12" and "annual review", in IBM trust terminology. Does it correspond to approach B or approach C in my appendix or neither? Does the terminology correspond to common usage in the pensions industry?

(f) The uncertainty about the trust's concern for the norm. If the trust was fully informed and considerate of the norm in 1996, and again in 1999, why was this one of the considerations that led Barry Morgan to resign because he felt prevented from doing his duty?

(g) The uncertainty about the reasons for the urgency to make changes in 1996, when those changes would have to be managed by different trustees under a different arrangement.

(h) The uncertainty about why the advice given by the trust in "IBM Pension Matters 2", about how discretionary increases would reflect economic conditions, turned out not to match reality. Trustee efforts to make it match are relevant, as well as IBM policy.

(i) Whether there were mass influxes of members (in addition to the Data Science influx) as I postulated, that further reduced the "reserves per member".

(j) Whether the trust has taken responsibility for the flaw in the document discussed under "My personal involvement".

(k) The possible misunderstanding referred to above as "More assets, same liabilities, equals better prospects for members.".

(l) Which of 6.1 and 6.2 above corresponds to the correct interpretation of the statutory surplus regulations.

Yours faithfully,

B L Marks ....

cc The IBM Trust

IBM UK

Dave Mitchell

Alan Murphy

Here are three ways in which a review might calculate a postulated pension increase. (I use the term increase even if the change represents a drop in the value of the pension.)

Approach A. At each interval of one year, the change in the Retail Prices Index (RPI) over the year is calculated. Some calculation, for example limiting the change to 5% or multiplying it by 0.7, may be applied. The resulting percentage change is the postulated change to the pension in payment. Off course, the review may choose not to apply the change, for whatever reason.

It is a characteristic of this method that the reference month for the RPI change is always the same. The vast majority of pension schemes in the UK use this approach A.

Approach B. From time to time, the interval from the previous time is calculated. The RPI change over that period (or a period offset to be a few months earlier) is calculated. Some calculation, for example multiplying by 0.7, may be applied. The resulting percentage change is the postulated change to the pension in payment.

Approach C. From time to time "increases" (which may be zero) are made. After an elapsed year from one increase, the RPI change over that year is calculated. Some calculation, for example multiplying by 0.7 may be applied. A decision is made on the date of the increase; it will be later than the elapsed year and allow time for consideration and organisation of the increase. The number of months between the increases is calculated, call it n. The RPI change over the year is multiplied by n/12. The resulting percentage change is the postulated change to the pension in payment.

The IBM Trust has provided me with two explanations of its approach. One corresponds to Approach B with a multiplying factor of 0.7, (the 70% RPI rule), and with a three month offset to allow time for consideration and organisation of the increase. This explanation was given to me by Kevin Waller. Kevin has been with the IBM Pensions organisation since before I retired and I am confident that what he told me was a knowledgeable and genuine account of the process as he saw it.

The other explanation that the IBM Trust has given me is in BMNN. It corresponds to Approach C with a multiplying factor of 0.7.

I deduce that the trust believes these two approaches are the same, and either explanation is as good as the other.

They are not the same. One existence proof of difference is that Approach B can lead to increases at an interval of less than a year, and Approach C cannot.

The question then arises, which approach does the Trust use?

The evidence is that it uses Approach B. I know of two instances where the trust has documented historical increases; there are probably more places. One of the places is in the document that I reproduced an increases table from in the "My personal Involvement" section of my stage 2; BMNN acknowledges that (apart from the single flaw) that document was accurate. The other is given below. In both cases the column for RPI increase gives values related to varied intervals. Such values are relevant to Approach B and not relevant to Approach C.

Approach C is not good in theory. Each interval between increases involves taking the RPI change over an elapsed year. The intervals are longer than a year. So the changes of RPI in some periods never get involved in increase calculations.

The IBM Trust process and history was recorded (by Kevin Waller and me, with Kevin representing the trust) as:

The RPI increases are calculated from 3 months prior to the beginning of the interval to 3 months prior to the end of the interval.

Summary: In practice such increases have been awarded at the level of approximately 70% of the rise in the Retail Price Index, with such increases being granted between 12 and 18 months apart.

(The Trust has given the summary in the paragraph above but would agree now that 10 months to 21 months is correct.)"

Note that the RPI increases are related to the intervals, not to elapsed years. The shortest interval is ten months, which is not possible with Approach C.