REVIEW OF PENSION SCHEME WIND-UP PRIORITIES A REPORT FOR THE DEPARTMENT OF SOCIAL PROTECTION 4 TH JANUARY 2013

Transcription

1 REVIEW OF PENSION SCHEME WIND-UP PRIORITIES A REPORT FOR THE DEPARTMENT OF SOCIAL PROTECTION 4 TH JANUARY 2013

2 CONTENTS 1. Introduction Approach and methodology Current priority order Potential s Summary and conclusions Appendix A: Dataset MERCER i

3 1 Introduction Background and terms of reference 1.1 When a pension scheme (other than a defined contribution scheme) winds up, Section 48 of the Pensions Act, 1990 and the Third Schedule to the Act set out how the assets of the scheme must be applied by the trustees to discharge the scheme s liabilities. Section 48 overrides any contrary provision contained in the scheme s rules. 1.2 The Department of Social Protection has engaged Mercer to provide actuarial consultancy support in its review of Section 48 of the Act. The analysis and conclusions are set out in this report. The terms of reference were as follows: To undertake a review of the wind-up priority provision in section 48 of the Pensions Act. The objective of this review is to determine to what extent, if any, the provisions in section 48 of the Act might be revised to provide for a different approach to the distribution of assets in the wind-up of an underfunded pension scheme. This document looks at the technical aspects of the distribution of assets under Section 48 and not at the policy and regulatory aspects. While it identifies policy issues where they arise, e.g., the potential knock-on impact on the funding standard, it is not within the scope of our terms of reference to address these. 1.3 The context of this review is as follows: At present, defined benefit (DB) pension schemes in Ireland are under significant pressure, with at least 70% not meeting the funding standard under the Pensions Act according to the Pensions Board s most recent annual report. If a scheme winds up, priority is currently given to pensioners (excluding future increases in their pensions) before other members. This means that an underfunded scheme, if it has a high weighting of pensioner liabilities, may deliver little or no benefit to active members (actives) and deferred pensioners (deferreds) on wind-up. This outcome is particularly harsh for actives and deferreds close to retirement, who could lose significant accrued benefits. They have very little working lifetime left in which to build up additional pension benefits to compensate for the loss. Many DB schemes are now closed to new entrants and are maturing quickly. In such cases, wind-up will become inevitable at some point in the future when MERCER 1

4 nearly all the members will have retired or died and the fund has become depleted. More critically, the current level of underfunding and the general financial and regulatory environment may see more schemes winding up in the short term than would normally be the case. The deadlines for underfunded schemes to submit recovery plans, which had been suspended pending a regulatory review of DB schemes, have recently been re-introduced. Some schemes may not be able to formulate a viable recovery plan and their only alternative may be to wind up. This increases the urgency to review the wind-up priorities under Section 48. Options 1.4 In October 2011, the Minister for Social Protection signalled an intention to Section 48, suggesting a reduced priority for pensioners of 75% of their current pension or 30,000, whichever was the lower. Once this level of benefits was secured, any remaining assets would be allocated to actives and deferreds up to the same level before any residual portion of benefits was secured. Given the sensitivity and complexities of the issues involved, it was decided to give further consideration to how the assets of the scheme can be distributed on wind-up before making any. 1.5 In considering possible alternative approaches to the distribution of assets on the wind up of a scheme, there are a number of mechanisms that could be applied. These include: Applying less than 100% priority for pensioners, e.g., allowing 75% of pension as highest priority. Applying monetary limits to priority levels, or a reducing scale depending on the amount of pension, e.g., 100% of the first 6,000, 75% of the next 12,000, etc. Advancing the priority of actives and deferreds so that a portion of their pension ranks equally with pensioners. Relating priority to length of service, or to the term to retirement age, for all scheme members. Taking account of the option to buy sovereign annuities rather than conventional annuities for pensioners, which would release more assets for distribution to others. Calculating part of pensioners entitlements on the basis of an actuarial value rather than on the cost of a conventional annuity, which is the current calculation basis. MERCER 2

5 This report explores these possible alternatives, taking into account the input from stakeholders. Of necessity, this report is quite technical in nature, particularly section 4, which analyses the options in detail. Many readers will find it simpler to read the summary in section 5 and refer back to section 4 for more background information if required. 1.6 It is important to appreciate that any to the current priority order, in order to achieve its objectives, must involve some redistribution from existing pensions in payment to actives and deferreds. In a wind-up of an underfunded pension scheme, the available assets, by definition, are not sufficient to cover all liabilities. The objective of this review is to determine to what extent, if any, the provisions in Section 48 of the Act might be revised to provide for a different approach to the distribution of assets in the wind-up of an underfunded pension scheme. 1.7 The pension promise associated with a DB scheme is regarded as unconditional. If the scheme winds up, the only practical way the promise can be fulfilled is to buy an annuity policy from a life assurance company that guarantees to continue the pension for the lifetime of the pensioner and any contingent dependants. The most appropriate way to value such an obligation is therefore by reference to market annuity rates. An alternative method of valuing pensioner liabilities has been proposed and is considered later in this report. Other considerations Legal and other criteria 1.8 If there is to be any, it must also be robust from a legal perspective. In this regard, general legal principles require that any acceptable must satisfy the following criteria: 1) It must serve a legitimate purpose 2) It must be proportionate to that purpose 3) It must avoid invidious discrimination. These criteria must be taken into account in assessing any of the options in this report. We believe that another criterion could be added, which is that any should be practical and be capable of being administered relatively easily. No entitlement to State pension in certain situations 1.9 Some members of pension schemes are not entitled to the State retirement pension, especially in employments that are or were linked to the public service. These members rely entirely on their scheme pension for their retirement income. Any considered to the priority order needs to have particular regard to their circumstances. MERCER 3

6 DC assets in a DB scheme 1.10 Some DB schemes hold assets for individual members on a DC basis, such as additional voluntary contributions (AVCs), transfers-in, the DC element of a hybrid scheme, etc. AVCs already have first priority under Section 48, but we believe it would be consistent and appropriate for all DC liabilities to be afforded similar priority. Sovereign annuities 1.11 With the advent of sovereign annuities, trustees have an additional option other than conventional annuities for discharging pensioner liabilities. We come back to this point later in this report. Annuities held by the scheme 1.12 In the majority of pension schemes, pensioners are paid directly from the fund on an ongoing basis. Some schemes, however, have purchased annuities in respect of some or all of their pensioner liabilities and hold them as an asset of the scheme When it comes to a wind-up, it may not be possible to reduce a pensioner s income if it is backed by an annuity policy, as most annuity policies would not provide for a partial or full surrender. In such circumstances, the only practical solution is to transfer the policy to the pensioner (the annuitant under the policy), in which case the pensioner receives automatic priority for the income under the annuity policy This is not a new situation. If a very underfunded scheme winds up at present, with a mix of insured and non-insured pensions being paid, it is possible that pensioners whose pension is backed by an annuity policy would receive higher priority. Such instances are rare, but they might arise more frequently if the priority order is d. We are not suggesting that any corrective measures should (or indeed could) be made, but just wish to point out the potential for the different treatment of pensioners in such circumstances. Impact on lump sums, transfers and annuity purchases 1.15 If the priority for pensioners is reduced and a scheme winds up, those who did not ex pension for cash at retirement may be more adversely affected than those who did. For example, take Members A and B, who had the same entitlement at retirement to a pension of 16,000 per year. Member A chose not to commute any part of it, while Member B commuted 25% for a lump sum and is now receiving 12,000 per year. The scheme winds up in deficit, and all pensioners receive just 75% of their pension. Member B therefore receives 9,000 per year and has already received a lump sum value in respect of 4,000 per year, giving a total pension value of 13,000 per year. Member A simply receives a 25% reduction, which leaves just 12,000 per year. Member A is therefore worse off than Member B in value terms by 1,000 per year, even though both had the same entitlements at retirement. MERCER 4

7 1.16 This inequity could be overcome by applying priority percentages to the original pension prior to any commutation for all pensioners. However, this would be difficult to administer in practice, and is not recommended. It should also be noted that this situation already exists under the current priority order, i.e., in a wind-up of a very underfunded scheme that does not have sufficient assets to cover pensioner liabilities, all pensioners receive the same percentage reduction in their current (post-commutation) pension. It is not, therefore, a new issue For similar reasons, if the priority for pensioners is reduced, it may cause behavioural s as regards future lump sums. In other words, if members have an option to ex part of their pension at retirement, they may choose to maximise their lump sum and minimise their pension if they believe their pension could be at risk of being reduced at some point in the future. In practice, most members tend to maximise their lump sums in any event, so it is not likely that any material will result Another possible behavioural is that members close to retirement might request a transfer to their own pension arrangement in an attempt to avoid any potential risk of reduction in their pension should the main scheme subsequently wind up. Typically, however, the trustees would use their power under the Pensions Act to reduce transfer values if the scheme was underfunded. Any transfers, therefore, should not have any particular consequences for the scheme Changing the priority order might also affect decisions regarding annuity purchase when a member reaches retirement age. Members are more likely to want the trustees to purchase annuities on their behalf because of the added security provided by annuities. Trustees, on the other hand, may prefer not to purchase annuities for exactly the opposite reason, i.e., in order to achieve a more balanced distribution of assets should the scheme subsequently wind up in an underfunded position. Average pension amounts 1.20 If monetary floors or ceilings are introduced as part of a revised priority rule, the distribution of the scheme s resources on a wind-up will be quite sensitive to the actual amounts of pension being paid by the scheme. In Section 2, we describe the pensioner data we have used for assessing the impact of making various s to the priority order, which is based on actual schemes in existence. It is worth summarising the distribution of annual pension amounts here, which is as follows for the model scheme: MERCER 5

8 BAND No. of pensioners % of pensioners 0 to 3, % 3,000 to 6, % 6,000 to 9, % 9,000 to 12, % 12,000 to 18, % 18,000 to 24, % 24,000 to 30, % 30,000 to 45, % 45,000 to 75, % 75,000 and above 22 2% 1, % 1.21 The average and median pension amounts are approximately 16,000 and 9,000 per year respectively. The model is scaleable, i.e., using a scheme with 1,000 pensioners is valid for smaller schemes as long as they have a similar overall profile in terms of age and pension amounts It is important to appreciate that members can have low pensions for different reasons. For example, a low pension could be due to short service, to low salary or pensionable salary, to having retired early with an actuarial reduction factor applying to their accrued pension, or to a low accrual rate in the scheme. If it is decided to protect members with low pensions, which we consider later, it is perhaps preferable to favour those whose pension is low by virtue of low pay rather than the other possible factors. Consistency with preservation 1.23 There are anomalies in certain situations between the benefits that have to be preserved for a member on leaving service and the benefits that have to be provided in a wind-up. For example, a scheme may provide benefits in respect of company service before joining the pension scheme such benefits are fully preserved on leaving service but may attract a lower priority in the event of a wind-up. This distinction between the benefits to be provided on leaving service and on wind-up seems unnecessary. It also causes administrative complexity when a wind-up occurs, as leaving-service benefits may have to be recalculated, adding to wind-up costs. This could be simplified as part of the review of Section 48. We discuss this point later in this report. Impact on Section Finally, it might be argued that a review of Section 48 of the Act would also justify a review of Section 50. Section 50 allows the benefits of actives and deferreds, but not pensioners, to be reduced when a scheme is in financial difficulty. If Section 48 is MERCER 6

9 amended and pensions in payment can be reduced on a wind-up as a result, it could be argued that a similar should be made to Section 50 for consistency. Consideration of this matter is not within the scope of this report MERCER 7

10 2 Approach and methodology Scheme profiles 2.1 The current wind-up priorities, and the impact of making any to these priorities, will affect schemes in different ways, depending on how underfunded they are on windup and on the weighting of pensioner liabilities relative to liabilities for actives and deferreds. 2.2 In order to make our assessment, we constructed a number of scheme profiles, reflecting real schemes in practice, that have the following characteristics: Funding levels: 50%, 75% and 90%. In other words, the value of the assets represents either 50%, 75% or 90% of the wind-up liabilities. Membership profiles: mature, average and young. A mature scheme is deemed to be one in which pensioner liabilities account for 75% of total scheme liabilities, with this figure dropping to 50% for an average scheme and 25% for a young scheme. 2.3 This matrix results in 9 possible scheme types. If we take the liabilities of the scheme as being 100 units, the matrix can be summarised as follows: Mature Average Young Liabilities: - pensioners actives & deferreds Assets (funding level 90%, 75% or 50%) 90, 75 or 50 90, 75 or 50 90, 75 or 50 Deficit 10, 25 or 50 10, 25 or 50 10, 25 or Although this is quite a range of possible situations, it does not cover the full extremes of either under-funding or maturity profile that can arise in practice. 2.5 As we modelled the results, it was evident that the impact of making a to the priority order was most clearly illustrated by looking at the most extreme cases, for example, a mature scheme with only a 50% funding level. By contrast, changing the priority order for a young scheme that is only marginally underfunded has very little impact on the distribution of assets. For this reason, in the remainder of this report, we MERCER 8

11 concentrate on two specific cases that we believe are adequate to illustrate the impact of the s: Scheme A: An average membership profile (50% pensioner liabilities) with 75% funding level Scheme B: A mature membership profile (75% pensioner liabilities) with 50% funding level. 2.6 While Scheme A represents a reasonably typical scheme, Scheme B is more extreme in having a high weighting of pensioners and being very poorly funded. However, the Pensions Board has indicated to us that schemes like Scheme B do exist, and we believe it is important to include such a scheme in our modelling to demonstrate the range of outcomes that is possible. Overall, we are satisfied that examining the results for these two scheme types captures the essential issues for all scheme types. Future pension increases 2.7 We also considered the position of future pension increases. Since 2009, the liability of a scheme to provide future pension increases ranks below the liability to provide base (i.e., non-increasing) pensions for all categories of member. There is a valid rationale for prioritising base pensions over future increases when resources are insufficient. We therefore expect that future pension increases will remain as a lower priority under any. This effectively means that we can ignore whether a scheme provides or does not provide future increases in our modelling, as any in priority will apply to base benefits only. It should be noted, however, that some schemes provide pension increases for some categories of member and not for others, in which case applying a reduction to the base pension as well as removing pension increases is a proportionately greater reduction in entitlement for such members. Dataset 2.8 In order to model priorities that include monetary floors or ceilings, or priorities related to age or service, we need to consider the distribution of ages and pension amounts in a typical scheme. 2.9 For this purpose, we constructed a notional scheme membership based on average membership data of real schemes, based on our experience of schemes that we administer. To ensure that the dataset was not atypical, in respect of pensioners at least, we benchmarked the pensioner data with a database from the Department of Social Protection / Pensions Board covering over 40,000 occupational scheme pensioners. The distributions of ages and amounts were similar to our own database We include a summary of the dataset used in Appendix A. MERCER 9

12 2.11 While we are satisfied that the dataset used is reasonably representative, there will be schemes in practice with higher or lower average pensions. These schemes may be affected to a greater or lesser degree than the standard scheme by s to the priority order, especially if the priority order includes monetary limits. In the interests of keeping the output manageable, we have not sought to model every possible scheme profile here. Nevertheless, it is a factor to be borne in mind when considering the analysis. Output 2.12 The results for each scenario before and after the are shown at a scheme level, i.e., the average level of cover for pensioners and for actives and deferreds. We then show the impact for some individual members in each category. For this purpose, we have taken sample illustrative members as follows: Pensioners aged 52 and 72, with three different levels of pension of 1,500 per year, 15,000 per year and 115,000 per year. Actives / deferreds aged 42 and 62, with three different levels of accrued pension of 1,750 per year, 12,500 per year and 45,000 per year. The 42-year-old has average completed service of 10 years while the 62-year-old has average completed service of 22 years We selected these data points for the following reasons: Firstly, for pensioners, choosing an age of 72 is representative of those who have retired normally. We have also chosen an age of 52 to highlight the impact on those who may have taken early retirement or who may be spouses of deceased employees. Although fewer in number, it is important to consider the impact of on such smaller categories. The average pension in our data is approximately 16,000 per year. The amounts are quite skewed, with 84% of pensions being below 30,000 and with a median pension of approximately 9,000 per year. Further information is contained in Appendix A. We wanted to illustrate the impact of on a low, medium and high pension to gauge the range of impacts, particularly where monetary floors or ceilings were considered. We therefore selected sample pension amounts of 1,500, 15,000 and 115,000 per year. Although the latter may seem very high, such pensions do exist and were evident in our data. The range of accrued pensions for active and deferred members tends to be narrower, and the average accrued pension lower, than for pensioners because they have not completed their full service. In this case, we found that the data could reasonably be represented by low, average and high accrued pensions of 1,750 per year, 12,500 per year and 45,000 per year respectively. We chose MERCER 10

13 sample ages of 42 and 62 to illustrate the impact of s on a person of average age and on someone close to retirement While the above sample data points were taken to illustrate the impact on individual members, it should be appreciated that the overall scheme data in each case was weighted to reflect the numbers of scheme members at different ages and with different past service periods and pension amounts (see Appendix A for further information on this distribution). Using the above examples, therefore, is a fair and accurate representation of the results. Assumptions 2.15 We based our modelling results on actuarial bases that represent (conventional) annuity rates and standard transfer value rates (as per the Society of Actuaries in Ireland s guidance in Actuarial Standard of Practice Pen-2) prevailing in September When modelling the impact of valuing pensioner liabilities by reference to actuarial values rather than by reference to the cost of annuities, we also used the standard transfer value basis for this purpose. However, if an actuarial value option were to be introduced, it is possible that a different actuarial basis would be prescribed. We have not considered this further in our report, but would simply point out that the use of a different actuarial basis might alter the outcomes shown in this report to some degree. Consultations 2.17 The Department of Social Protection arranged a round-table consultation meeting of various interest groups to discuss different options and obtain general reactions. Written submissions were subsequently received from some of these groups, and indeed from other groups and individuals. All of the submissions received were taken into account in our review One proposal received and considered in this report is that the method of valuing pensioner liabilities could be d to a mixed valuation approach involving the application of an actuarial value to some portion of the liabilities, rather than an annuity cost. 1 The proposal envisaged a in the nature of the benefit from an income to a lump sum calculated using a higher yield than used in market annuity rates, thus resulting in a lump sum smaller than the amount needed to purchase an annuity to cover the pension. The use of an actuarial value to calculate a portion of the pensioner liability would be consistent with the calculation of active and deferred member liabilities, which are also on the basis of actuarial values. 1 An annuity cost can also be regarded as an actuarial value, but for the purposes of this report we will refer to an actuarial value as it is defined in the Pensions Act, which is broadly the equivalent cash value of a benefit allowing for certain financial assumptions and survival probabilities. MERCER 11

14 2.19 The proposal referred to in the previous paragraph further suggested that the actuarial value could be transferred as a lump sum to an appropriate investment vehicle such as an Approved Retirement Fund (ARF) and drawn down as the member wished, subject to the normal ARF rules. It should be noted that the ARF option is currently not generally available to DB scheme members (apart from proprietary directors, or members of DB schemes who have additional voluntary contributions). As an alternative to transferring the lump sum to an ARF, it could be used to buy a further annuity, although probably of a lower amount than the pension it replaces (see 2.21 below) Moving away from an annuity is not, however, a straightforward matter. While offering an actuarial value instead of an annuity may provide more flexibility to pensioners to decide how to allocate their retirement savings, the member s need in retirement may be for a lifetime income rather than a capital sum. Only an annuity can satisfy this need fully It is important to recognise that valuing and discharging a pensioner s liability on an actuarial basis is likely to result in an effective net loss to the pensioner if the actuarial value is used to purchase an annuity. The extent of the loss will depend on the actuarial basis used. In current market conditions, the difference between an annuity cost and an actuarial value (for this purpose, we have used the standard transfer value basis set by the Society of Actuaries in Ireland) ranges from approximately 20% for a person in their early 60s down to nil for a person in their mid-80s. This means that the pensioner s ability to purchase an annuity to guarantee an income for life is reduced proportionately This proposal also suggested that, if the mixed valuation approach was adopted, the Funding Standard should be adjusted to reflect this ; however, this is outside of the scope of our terms of reference and is a matter for consideration in the regulatory impact assessment prepared by the Department. MERCER 12

15 3 Current priority order 3.1 Section 48 of the Pensions Act, 1990 currently requires trustees of pension schemes (other than defined contribution schemes) that wind up after 29 April 2009 to discharge the liabilities in the following order: 1. Benefits payable in respect of additional voluntary contributions (AVCs) made by employees, or AVCs transferred in; 2. Pensions currently being paid to pensioners, or due to be paid to members who have reached normal pensionable age but have not yet drawn down their benefits, excluding any future increases in such pensions; 3. Accrued benefits 2 for current and former employees who have not yet reached normal pensionable age, excluding any future increases in pensions after retirement. Accrued benefits in this regard include an allowance for revaluation of benefits between the date of wind-up and normal pensionable age in respect of reckonable service after 1 January Revaluation of pre-1991 benefits is not included in this priority unless the scheme rules provide for such revaluation. 4. Any pension increases to pensions provided under the rules on benefits in the second and third priorities above. These rank equally between pensioners and non-pensioners. 5. Revaluation of pre-1991 benefits for active members and deferred pensioners, plus any residual benefits, that are not already allowed for under the third priority above. The expenses of winding up can be discharged before any benefits are paid. 3.2 Section 48 overrides any contrary provision which may be contained in a scheme s trust deed and rules. 3.3 Where pension increases under no. 4 above are linked to CPI, the trustees can secure annuities for pensioners that incorporate a fixed rate of increase rather than being fully index-linked. This is known as fixed rate substitution, and is usually a cheaper option for the scheme. However, where there are assets left over after discharging the other 2 Specified in paragraphs 2, 3 and 4 of the Third Schedule of the Pensions Act MERCER 13

16 priority liabilities, the trustees must use the residual assets to increase benefits for pensioners up to the cost of index-linked annuities, to the extent possible. 3.4 The priority order above has d on a number of occasions. In 2002, AVCs were included as a priority ahead of all other benefits. In 2009, future pension increases (no. 4 above) were designated a lower priority than base pensions. 3.5 If we apply the current priority order to Scheme A (average/75%) and Scheme B (mature/50%) as described in paragraph 2.5, the following picture emerges: Scheme funding position Scheme A (Average/75%) Scheme B (Mature/50%) Liabilities: Pensioners Actives & deferreds Assets: Coverage: Pensioners 100% 66% Actives & deferreds 50% 0% Individual member coverage levels (Accrued) Pension Scheme A (Average/75%) Scheme B (Mature/50%) Pensioner (Any age) 1, % 66% 15, % 66% 115, % 66% Active/ deferred (Any age or service) 1,750 50% 0% 12,500 50% 0% 45,000 50% 0% 3.6 In Scheme A, the assets of 75 cover all the pensioner liabilities of 50, so there is 100% coverage for pensioners. This feeds down into the individual pensioners benefits. The remaining assets of 25 only cover half the actives and deferreds liabilities of 50, so these members only receive 50% of their accrued benefits. For example, an active member aged 62 with an accrued pension of 45,000 would only receive value for an accrued pension of 22, The weaker position of actives and deferreds under the current priority order is more pronounced in Scheme B. In this case, all of the assets of 50 are applied to secure twothirds of pensioner liabilities of 75. This leaves no coverage for actives and deferreds, so that a person in this category aged 62 receives no benefit. 3.8 An existing option for trustees on a wind-up is to discharge pensioner liabilities by purchasing sovereign annuities rather than conventional annuities. If we assume that a MERCER 14

17 discount of 15% were available in respect of sovereign annuities and such annuities were purchased, the table would look like this: Scheme funding position Scheme A (Average/75%) Scheme B (Mature/50%) Liabilities: Pensioners Actives & deferreds Assets: Coverage: Pensioners 100% 78% Actives & deferreds 65% 0% Individual member coverage levels (Accrued) Pension Scheme A (Average/75%) Scheme B (Mature/50%) Pensioner (Any age) 1, % 78% 15, % 78% 115, % 78% Active/ deferred (Any age or service) 1,750 65% 0% 12,500 65% 0% 45,000 65% 0% 3.9 In this case, purchasing sovereign annuities rather than conventional annuities releases more resources for other purposes. In Scheme A, the cost of discharging pensioner liabilities falls from 50 to 42, resulting in a jump in coverage for actives and deferreds from 50% to 65%. In Scheme B, the cost of discharging pensioner liabilities falls from 75 to 64; there is still no coverage for actives and deferreds, but pensioners see the nominal level of their pension improve from 66% of their current pension to 78% The availability of sovereign annuities provides trustees with an additional option on the wind-up of a scheme. The impact of this option will vary from time to time depending on prevailing bond yields. While it remains an option in its own right, it is also a consideration in the context of each of the options considered in the modelling exercise In the next section, we look at the impact of making various s to the priority order. MERCER 15

18 4 Potential s 4.1 There are various types of s to the priority order that can be considered, e.g.: Prioritising pensioners up to a certain amount or percentage of benefit and then allocating monies to actives and deferreds before distributing any remaining monies. Prioritising according to term to normal retirement age rather than retirement status. This option would provide some protection for older actives and deferreds compared to the present system. Prioritising according to service. This would allocate more monies to longerserving members. The method of calculating and discharging liabilities could be d, which might alter the distribution. For example, an annuity might have to be bought for the first 12,000 per year of pension, with an actuarial value of the pension above this amount being paid as a lump sum to the member or to another pension arrangement. Changing the method of valuation could be combined with some of the priority s outlined above. Also, sovereign annuities could be used in place of conventional annuities in respect of some or all of the pensions in payment. Paying lump sums or buying sovereign annuities would be cheaper than buying conventional annuities at times when the latter are particularly expensive, thereby increasing the remaining assets available for distribution. 4.2 As mentioned in Section 1, it is important to appreciate that any to the current priority order, in order to achieve the objective of a fairer share of the available resources on wind-up than at present, must involve some redistribution from pensioners to actives and deferreds. This means that some members must benefit at the expense of others. 4.3 We have modelled a wide range of possible s to the current priority order and we can now examine the results. Option 1 Percentage and monetary cap 4.4 As mentioned in paragraph 1.4, an option originally proposed in October 2011 was a reduced priority for pensioners of 75% of their current pension or 30,000, whichever was the lower. Once this level of benefits was secured, any remaining assets would be MERCER 16

19 allocated to actives and deferreds up to the same level before any residual portion of benefits was secured. A summary of the priority order, therefore, is: 1) Pensioners 75% of pension, or 30,000 if lower 2) Actives & deferreds 75% of accrued pension, or 30,000 if lower 3) Pensioners balance of pension 4) Actives & deferreds balance of accrued pension. 4.5 The impact of making this on Scheme A and Scheme B would be as follows: Option 1 Scheme A (Average / 75%) Scheme B (Mature / 50%) Scheme funding position Coverage: Pensioners Actives & deferreds 100% 82% 66% 63% 50% 68% 0% 8% Individual member coverage levels (Accrued) Pension Pensioner (Any age) 1, % 88% 66% 75% 15, % 88% 66% 75% 115, % 64% 66% 26% Active/ deferred (Any age or service) 1,750 50% 75% 0% 9% 12,500 50% 75% 0% 9% 45,000 50% 67% 0% 8% Comments: 4.6 In Scheme A, the coverage for pensioners reduces, as would be expected, as some of the assets are redistributed to actives and deferreds. Those on the highest pension amounts suffer the biggest reductions in pensions from 100% to 64%. The actives and deferreds, on the other hand, see their coverage increase from 50% to 67%-75%. 4.7 In Scheme B, the redistribution to actives and deferreds is minimal: they only see an increase in coverage from nil to 8%-9%. The main impact here is an allocation from higher-paid pensioners to lower-paid pensioners. At the highest end, this is quite extreme: a pensioner on 115,000 per year would now only receive 26% of this amount ( 30,000), compared to 66% of the pension under current rules. This may be regarded as a disproportionate for such members relative to the current situation. MERCER 17

20 4.8 This option is not particularly effective in addressing the situation where members who are just short of their retirement can end up with little or no benefit while members who have reached retirement age are reasonably well protected (the pension cliff effect). 4.9 One of the problems of pitching the monetary ceiling at 30,000 is that not many pensioners receive this level of pension: in our data, 84% of pensioners have a pension lower than this amount. This means that taking money away from pension amounts above this level and sharing it among those with smaller benefits does not have a great impact overall, while having a materially negative impact on those concerned. Option 1A - Percentage and monetary cap with equal priority 4.10 A variant on this proposal that achieves a greater re-balancing between pensioners and other members is to advance the priority of actives and deferreds to equal that of pensioners for the same priority amount, viz., 75% of (accrued) pension or 30,000 if lower. Under this option, then, the priority order would be as follows: 1) All members 75% pension or accrued pension, or 30,000 if lower 2) All members balance of pension or accrued pension The impact of making this on Scheme A and Scheme B would be as follows: Option 1A Scheme funding position Scheme A (Average / 75%) Scheme B (Mature / 50%) Coverage: Pensioners Actives & deferreds 100% 73% 66% 49% 50% 77% 0% 52% Individual member coverage levels (Accrued) Pension Pensioner (Any age) 1, % 82% 66% 58% 15, % 82% 66% 58% 115, % 46% 66% 20% Active/ deferred (Any age or service) 1,750 50% 82% 0% 58% 12,500 50% 82% 0% 58% 45,000 50% 76% 0% 51% MERCER 18

21 Comments: 4.12 Placing all members in the first priority distribution overcomes the cliff effect. In the table above, an active or deferred member who is close to retirement and whose accrued benefit is less than 40,000 will see an increase in coverage from 50% to 82% in Scheme A and from 0% to 58% in Scheme B. However, the pensioner in receipt of a large pension will see the value of their pension reduce from 100% to 46% in Scheme A and from 66% to 20% in Scheme B Essentially, under this option, monies are redistributed from pensioners, especially those on pensions above 40,000, to other members. An active or deferred member with an accrued pension below the 40,000 threshold receives exactly the same priority as a pensioner in this category. This could be viewed either as a positive (greater equity between categories) or a negative (an over-correction in favour of active and deferreds) We are still left with the issue that those on high pensions suffer a very material reduction. In some instances, coverage for actives and deferreds is higher than for such pensioners. Option 1B - Percentage and monetary cap with equal priority and actuarial value 4.15 We have also looked at a further variant on this option. This involves equal priority as per Option 1A above except that scheme liabilities for pensioners would be calculated on an actuarial basis for pension amounts above 6,000 per year In other words, we have assumed that the trustees would secure the first 6,000 per year of pension by means of a conventional annuity as normal, and, as mentioned in paragraph 2.19, the balance could be transferred to an alternative investment vehicle or used to buy a further annuity The figure of 6,000 has been chosen as a reasonable level of income to protect, based on the data (40% of pensions in payment are less than this amount). To ensure that it remains relevant over time, it could be linked to the State pension, e.g., 50% of the State pension For the purpose of calculating actuarial values, we have used the standard transfer value basis. An actuarial value in current conditions ranges from approximately 80% of an annuity cost at age 60 to approximately 100% at age Making this further does not affect the priority order, which remains as follows (but with pensions calculated on an actuarial value basis above 6,000) : MERCER 19

22 1) All members 75% pension or accrued pension, or 30,000 if lower 2) All members balance of pension or accrued pension. The effect on Schemes A and B is as follows: Option 1B Scheme A (Average / 75%) Scheme B (Mature / 50%) Scheme funding position Coverage: Pensioners Actives & deferreds 100% 79% 66% 54% 50% 81% 0% 56% Individual member coverage levels (Accrued) Pension Pensioner (Any age) 1, % 85% 66% 62% 15, % 85%* 66% 62%* 115, % 55%* 66% 22%* Active/ deferred (Any age or service) 1,750 50% 85% 0% 62% 12,500 50% 85% 0% 62% 45,000 50% 80% 0% 55% Comments: * Percentage cover for pension, but on the basis of an actuarial value in respect of pension in excess of 6,000 per year. As outlined in paragraph 2.21, an actuarial value can represent a loss of 20% or more compared to the cost of replacing the income being received in the insurance market Applying the actuarial value option in respect of pensions over 6,000 per year does not have a very material impact on the results of Option 1A it increases the percentage cover by a few percentage points in most cases. For example, coverage for all members on pensions below 40,000 increases from 82% to 85% in Scheme A and from 58% to 62% in Scheme B. Note that the values that are asterisked in the table above are calculated on a mixed valuation basis and do not reflect the cost of buying an annuity for the full value of expected benefits The drawback of this option in respect of larger pensions remains: a person in receipt of 115,000 per year will see their entitlement reduced from 100% to 55% in Scheme A and from 66% to 22% in Scheme B. (continued) MERCER 20

23 Option 2 - Floor plus percentage priority 4.22 A second option is to apply a much lower monetary amount than 30,000 as a floor, and then a percentage of benefit above this floor without any further monetary limit applying. The example we took was as follows: 1) Pensioners 100% of first 6,000 of pension plus 75% of the balance 2) Actives and deferreds similar priority 100% of first 6,000 of accrued pension plus 75% of the balance 3) All members balance of pension or accrued pension This results in the following impact on Schemes A and B: Option 2 Scheme A (Average / 75%) Scheme B (Mature / 50%) Scheme funding position Coverage: Pensioners Actives & deferreds 100% 82% 66% 66% 50% 68% 0% 0% Individual member coverage levels (Accrued) Pension Pensioner (Any age) 1, % 100% 66% 81% 15, % 85% 66% 69% 115, % 76% 66% 62% Active/ deferred (Any age or service) 1,750 50% 78% 0% 0% 12,500 50% 68% 0% 0% 45,000 50% 61% 0% 0% Comment: 4.24 This is an improvement on Option 1. There is better protection for those on low pensions, as first priority is given to the first 6,000 of pension. For example, a pensioner receiving 1,500 per year sees their coverage remain at 100% in Scheme A and actually increase from 66% to 81% in Scheme B. At the same time, those on higher pensions are not overly disadvantaged: those on 115,000 per year receive 76% of their pension in Scheme A and 62% in Scheme B, which compares with 64% and 26% respectively under Option 1. However, where the scheme is particularly poorly funded (Scheme B), actives and deferreds close to retirement still end up with no benefit. MERCER 21

24 Option 2A - Floor plus percentage, with equal priority 4.25 Again, we can amend this option to advance the priority for actives and deferreds into the first priority distribution. This would give us the following priority order: 1) All members 100% of first 6,000 of pension or accrued pension plus 75% of the balance 2) All members balance of pension or accrued pension This results in the following impact on Schemes A and B: Option 2A Scheme funding position Scheme A (Average / 75%) Scheme B (Mature / 50%) Coverage: Pensioners Actives & deferreds 100% 72% 66% 49% 50% 78% 0% 53% Individual member coverage levels (Accrued) Pension Pensioner (Any age) 1, % 88% 66% 60% 15, % 75% 66% 51% 115, % 67% 66% 45% Active/ deferred (Any age or service) 1,750 50% 88% 0% 60% 12,500 50% 77% 0% 52% 45,000 50% 69% 0% 47% Comments: 4.27 This option eliminates the cliff effect and sees a greater rebalancing of benefits without unduly affecting those on low pensions. Even in Scheme B, actives and deferreds now receive a reasonable level of coverage, ranging from 47% to 60% in our examples above compared to zero under the current system This improvement for actives and deferreds comes at the expense of pensioners, whose coverage reduces in Scheme B from 66% under the current priority to a range between 60% and 45%, with the lowest coverage applying to those on the highest pensions. MERCER 22

25 Option 2B - Floor plus percentage, with equal priority, and actuarial value 4.29 We can again vary this option by adding in the actuarial value option for pension amounts above 6,000 per year. This gives the following priority order: 1) All members 100% of first 6,000 of (accrued) pension plus 75% of the balance. For pensioners, the first 6,000 is valued on a conventional annuity basis and the balance is valued as an actuarial value. 2) All members balance of (accrued) pension on actuarial value This results in the following impact on Schemes A and B: Option 2B Scheme funding position Scheme A (Average / 75%) Scheme B (Mature / 50%) Coverage: Pensioners Actives & deferreds 100% 77% 66% 47% 50% 82% 0% 57% Individual member coverage levels (Accrued) Pension Pensioner (Any age) 1, % 93% 66% 64% 15, % 79%* 66% 55%* 115, % 71%* 66% 49%* Active/ deferred (Any age or service) 1,750 50% 93% 0% 64% 12,500 50% 81% 0% 56% 45,000 50% 73% 0% 50% Comments: * Percentage cover for pension, but on the basis of an actuarial value in respect of pension in excess of 6,000 per year. As outlined in paragraph 2.21, an actuarial value can represent a loss of 20% or more compared to the cost of replacing the income being received in the insurance market Again, this option adds a few percentage points to coverage in most categories and otherwise has the same impact as outlined in paragraphs 4.27 and 4.28 above for Option 2A. Note that the values that are asterisked in the table above are calculated on a mixed valuation basis and do not reflect the cost of buying an annuity for the full value of expected benefits. MERCER 23

26 Option 2C - Floor only, with equal priority, and actuarial value 4.32 One of the submissions received proposed an interesting variant on the other versions of Option 2, which was as follows: 1) Pensioners 100% of first 6,000 of pension on conventional annuity basis 2) Balance of pension for pensioners (actuarial value) plus all of accrued pension for active and deferreds We modelled this scenario and it results in the following impact on Schemes A and B: Option 2C Scheme funding position Scheme A (Average / 75%) Scheme B (Mature / 50%) Coverage: Pensioners Actives & deferreds 100% 84% 66% 59% 50% 76% 0% 41% Individual member coverage levels (Accrued) Pension Pensioner (Any age) 1, % 100% 66% 100% 15, % 86%* 66% 65%* 115, % 78%* 66% 44%* Active/ deferred (Any age or service) 1,750 50% 76% 0% 41% 12,500 50% 76% 0% 41% 45,000 50% 76% 0% 41% Comments: * Percentage cover for pension, but on the basis of an actuarial value in respect of pension in excess of 6,000 per year. As outlined in paragraph 2.21, an actuarial value can represent a loss of 20% or more compared to the cost of replacing the income being received in the insurance market This example has a similar outcome to Option 2B except that it favours those on lower pensions slightly more. The pensioner on 1,500 per annum in either scheme still receives 100% of his or her pension, whereas there would be some reduction in Option 2B. At an overall level, there is a redistribution from pensioners to actives and deferreds, with the redistribution being more pronounced for those receiving higher pensions Note that, if the scheme was one with generally low benefits, with most pensions under 6,000 per year, this option would not achieve any material redistribution from pensioners to actives and deferreds. MERCER 24

27 Option 2D - Floor only, with equal priority 4.36 We also examine the outcome if annuity values were required for all pensions in payment. In other words, the priority order could be as follows: 1) Pensioners 100% of first 6,000 of pension on annuity basis 2) Balance of pension for pensioners (annuity basis) plus all of accrued pension for active and deferreds This results in the following impact on Schemes A and B: Option 2D Scheme funding position Scheme A (Average / 75%) Scheme B (Mature / 50%) Coverage: Pensioners Actives & deferreds 100% 79% 66% 55% 50% 71% 0% 37% Individual member coverage levels (Accrued) Pension Pensioner (Any age) 1, % 100% 66% 100% 15, % 83% 66% 62% 115, % 73% 66% 40% Active/ deferred (Any age or service) 1,750 50% 71% 0% 37% 12,500 50% 71% 0% 37% 45,000 50% 71% 0% 37% Comments: 4.38 This example has a similar outcome to Option 2C except that coverage levels are slightly lower all round. Again, the pensioner on 1,500 per annum in either scheme receives 100% of his or her pension, so those on lower incomes are protected. While at an overall level there is a redistribution from pensioners to actives and deferreds, this is more pronounced for those receiving higher pensions. For example, a person receiving 115,000 per annum sees his or her value reduce from 100% to 73% in Scheme A and from 66% to 40% in Scheme B. Varying the floor 4.39 It would be possible to improve the outcomes in 2D for pensioners on mid-range pensions by increasing the floor from 6,000 to a higher amount. We have looked at the MERCER 25

28 impact of alternative floors of 9,000 and 12,000. A summary of the results is as follows: Scheme A Scheme B Pensioners (any age or service) Actives & deferreds (any age or service) No Floor Floor Floor No Floor Floor Floor 6,000 9,000 12,000 6,000 9,000 12,000 % % % % % % % % , , , , , The higher floors improve the fortunes of the pensioner on 15,000, but at the expense of the pensioner on 115,000 and all actives/deferreds. The effect is fairly modest in Scheme A, but is more pronounced in Scheme B - coverage for the high pension amount if the floor is 12,000 is just 32% compared with 40% if the floor is 6,000. The coverage for this pensioner is currently 66% under the current priority rules, and would be 50% if the underfunding was shared proportionately across all members. A further reduction to 32% would represent a significant drop in benefit. (continued) MERCER 26