NHS PENSION SCHEME REVIEW HIGH EARNERS ISSUES

Transcription

1 NHS PENSION SCHEME REVIEW HIGH EARNERS ISSUES Date: 11 September 2007

2 This paper has been produced by the Government Actuary s Department at the request of the Technical Advisory Group (TAG) to the NHS Pension Scheme Review. The main purpose is to provide additional costings and analysis in relation to some recently proposed options around the benefit/contribution structure for high earners within the NHSPS. The paper is set out as follows: Executive Summary Section 1): Yields from introducing a fourth tier of whole-pay contribution rates Section 2): cost of new entrant GP accrual at 1.87% Section 3): comparisons of value for assessing the opportunity cost of removing the cap Section 4): cost to the Employer from lifting the earnings cap for future service 1

3 Executive summary TAG HIGH EARNERS ISSUES - FINAL i) Assuming the third contribution tier starts at the top of AfC Band 8b, a whole pay fourth-tier contribution rate at 8.5% or 9.0% for earners over 108,600 pa linked to future movements in average earnings yields a rounded estimated 0.05% of whole-scheme pay in both cases; ii) iii) Based on the same assumption, a whole pay fourth-tier contribution rate at 8.5% for earners over the top point of AfC Band 9 yields a rounded estimated 0.15% of whole-scheme pay; Based on the limited available data, similar yields are estimated to be achieved if the third contribution tier starts at the top of Band 8c. However, the data on this range of the NHSPS pay distribution are particularly limited, and the out-turn in practice might differ significantly; iv) The cost (expressed on a scheme-wide basis) of a 1.87% (rather than 1.80%) accrual rate for new GPs is projected to rise slowly over time to slightly above 0.05% of pay in 2020 but then approaching 0.10% of pay by 2050; v) If the Employer and members (where necessary) extend their contributions at the relevant rate to all earnings in excess of the earnings cap, the cost of providing uncapped future service benefits on the valuation basis is by definition met in full; vi) vii) viii) ix) Higher rate fourth-tier contributions from high earners can therefore be regarded as being required both to cover the cost of the extra GP accruals and to give some recognition (and cushion against adverse risks) in relation to the additional value to the high earners of removing the cap on their benefits; Mainly because of the lack of available data and likely materality, previous actuarial valuations of the NHSPS have not taken account of the potential savings from applying the earnings cap. If the earnings cap is retained, it would be reasonable to suppose that from some future point, valuations would take account of these savings. However, it is very difficult to anticipate the timing and the manner in which such savings would be recognized and expressed; If the earnings cap is now removed, therefore, this constitutes a lost opportunity for future savings to the scheme; There are also many difficulties associated with defining and estimating the additional value to high earners if the cap on benefits is removed (or the equivalent lost opportunity savings to the scheme); x) One such difficulty is that, if the earnings cap is assumed to continue to rise in line with price inflation, an increasingly large proportion of NHSPS earnings would be above the cap, and an increasingly large proportion of members would be affected by it. In the long-term, this might arguably be regarded as unsustainable; xi) Currently, the earnings cap is at a level ( 108,600) considerably higher than the top of the Agenda for Change pay structure. Based on the projection 2

4 assumptions, it might lie around the midpoint of Band 8b (currently about 55,000 pa) thirty years hence; xii) xiii) xiv) xv) xvi) It is projected that, within a thirty-year time horizon, over 90% of the current membership will NOT be affected by a (price-linked) earnings cap; Based on a number of simplifying assumptions and a thirty year time horizon, one pragmatic assessment of the additional value to high earners from removing the cap on benefits suggests that (when expressed on a schemewide basis) this would be equivalent to less than ¼% of whole scheme pay initially, increasing to approximately 1% of pay after 30 years. However, this latter figure depends heavily on the assumption that the linkage of the earnings cap to prices would continue to be maintained over that period if instead it was assumed to be linked more closely to movements in average earnings (as the Lifetime Allowance is in its initial years), there would be a considerable reduction in this estimate. An 8.5% contribution rate above an earnings-linked threshold of 108,600 pa would fall slightly short of covering the projected cost of GP accruals at 1.87% instead of 1.80%, whilst a 9.0% rate above this threshold would cover it probably leaving a small margin over the whole projection period; An 8.5% contribution rate above the top of AfC Band 9 would leave a margin of close to 0.1% of pay as a cushion against the risk of unanticipated future increases in costs associated with high earners; Given the uncertainty in the estimates, and the many different factors at work, a regular review specifically focused on the high earners contribution/benefit structure, and relativities with other groups of members, might well be required; 3

5 1) Yields from introducing a fourth tier of whole-pay contribution rates 1.1 In Part A of the GAD paper TAG TIERED MEMBER CONTRIBUTION STRUCTURES, various options for introducing tiered member contribution rates were examined. More recently, attention has been focused on a proposed three-tier wholepay member contribution structure of 5%/6.5%/7.5% with thresholds at the top of Agenda for Change ( AfC ) Pay Band 2 and the top of Pay Band 8b. 1.2 The TAG has now requested that we should analyse the effects of introducing a fourth tier of member contributions for the highest earners in the NHSPS. Most recently, we have been asked to focus specifically on whole-pay type rates of 8.5% or 9.0% applicable to the whole earnings of members earning above one of two possible new thresholds: (i) (ii) the top of AfC Pay Band 9 (currently 88,397 pa) or 108,600 pa (the current level of the earnings cap). In both cases, the threshold is assumed to move in future broadly in line with movements in future levels of average earnings. 1.3 In addition, we have been asked to examine the effect if the third-tier threshold was increased from the top of AfC Band 8b to that of Band 8c. 1.4 We have taken the contribution structure described in paragraph 1.1 as a suitable baseline for the current analysis, so that the estimated yields from introducing a fourth tier (and in some cases also adjusting the third tier threshold) are expressed as additions to the yields from introducing the paragraph 1.1 structure. 1.5 Currently, members subject to the earnings cap pay no member contributions on any earnings in excess of that cap. The calculated yields from the contribution structure described in paragraph 1.1 included 1.5% (ie 7.5% less 6%) of these earnings, and we have followed a similar approach in the current analysis. The extension of member contributions at the rate of 6% to earnings over the earnings cap is covered in Sections 3 and 4 of this paper. 1.6 The earnings distribution model used for previous costings of changes in the member contribution structure was based mainly on DH estimates of the likely distribution of AfC grades following the full roll-out of AfC. However, relatively high earners not covered by AfC were incorporated based on assumed average earnings for each of the relevant occupational groups. 1.7 In order to undertake the current costings, we have relied on the more recent but rather limited and incomplete data-set for high earners (ie those realistically within range of the earnings cap). We have constructed an assumed current earnings distribution for this group which fits as closely as possible with the limited data available on the upper part of that distribution (which has been analysed focusing on relativities with the earnings cap). Although less than ideal, we believe this produces results which are for the most part sufficiently robust. 1.8 Data relating to those earning between 50,000 pa and 70,000 pa are particularly limited more specifically, the assessed cost of increasing the third tier threshold from the top of AfC Band 8b is very sensitive to the distribution of earnings of various intermediate groups of earners on which little detailed information is available. 4

6 1.9 The DH projected distribution of AfC grades included about 1% of all earnings for AfC grades (on a whole-pay basis) for earners between the top of Band 8b and that of Band 8c. After making pragmatic adjustments to allow for the fact that a proportion of Specialist Associates, Dentists and various other medical groups are likely to have earnings in this range, this increases to slightly over 2%. The effect of charging this band of earners 6.5% instead of 7.5% of pay is therefore assessed as de minimis according to the usual rounding convention (although this is subject to a degree of uncertainty) The rounded estimated yields, expressed as percentages of total scheme pay, from an additional 1.0% or 1.5% of the whole pay of members whose pay exceeds (or is projected at some future point to exceed) a fourth threshold are set out in Tables A and B. The 1.0% yields correspond with a whole-pay fourth tier member contribution rate of 8.0%, and the 1.5% yields with an equivalent rate of 9.0% Table A shows results where the third threshold is the top of Agenda for Change Pay Band 8b, and Table B shows results where the third threshold is the top of Agenda for Change Pay Band 8c. Table A Whole pay extra 1.0% or 1.5% yields (3rd tier at top of AfC Band 8b) Addition to Fourth Threshold member contribution rate Top of AfC Band 9 Earnings-linked 108, % 0.15% 0.05% 1.5% 0.20% 0.05% Table B Whole pay extra 1.0% or 1.5% yields (3rd tier at top of AfC Band 8c) Addition to Fourth Threshold member contribution rate Top of AfC Band 9 Earnings-linked 108, % 0.15% 0.00% (de minimis) 1.5% 0.20% 0.05% 1.12 The level of the third-tier threshold has little visible effects on the results, although this aspect of the analysis is particularly prone to uncertainty see paragraph 1.8 above. However, on the assumptions adopted, the lost yield from increasing the third-tier threshold is sufficient to cause the rounded 0.00% result appearing in Table B Given the complexity of the structures under consideration and the number of variables involved, the view might be taken that there should be a regular review of the member contribution structure including particularly the fourth-tier threshold and the fourth-tier rate relative to rates in lower tiers. 2) Cost of new entrant GP accrual at 1.87% 2.1 Previous costings have been based on the assumption that new entrant GPs would accrue career average pensions at an accrual rate of 1.80% with lump sum by commutation. Rounded estimates of the additional projected costs if instead new GP accrual was at the rate of 1.87% are set out in Table C. 5

7 Table C: projected costs of new GP accrual at 1.87% instead of 1.80% Projection Rounded additional cost Year % (de minimis) % % % (nearly exact) % % % 2.2 We have been asked to indicate when these projected additional costs might first trigger an increase in the member contribution rate. Our projections suggest that there might first be an 0.1% increase in the year 2020, when the increased cost of just over 0.05% is just sufficient to lead to a rounded increase of 0.1% in the whole scheme contribution rate which would then probably be maintained up to the end of the projection period. It will be appreciated however that there must be a considerable degree of uncertainty in projecting tipping-points for rounding purposes in relation to an increase of only one-tenth of a percentage point in the scheme contribution rate. 2.3 Projected scheme costs with a new GP accrual rate of 1.87% and incorporating the effects of the four Table A 4-tier contribution structures are tabulated at Tables D to G. Each of the tables is based on a scenario in which compound GP dynamisation factors for 2003/06 amount to 36% and in which future salary scales are assumed to steepen by 0.75% pa ultimately and withdrawal rates to decline by 25% ultimately. Note that the format of these tables has changed since this paper was first produced for TAG so that it now matches the format of the tables appended to the Final Review Agreement. Note also however that no change has been made to the estimates of the costs associated with the benefit/contribution structures under consideration when the paper was produced, and further, that these structures do not reflect the structure finally chosen in every detail. 2.4 In each case, the additional high earner contributions would theoretically allow a reduction in the unrounded employer contribution rate from 2008 to 2015, sometimes to below 14.0%. However, we have been asked to assume that during this period a minimum employer contribution rate of 14.0% would apply. Part of the additional high earner contributions (net of the cost of the higher GP accrual rate) is therefore used to provide limited support to the employer contribution rate during this period. The remainder (plus or minus surpluses/deficiencies arising from rounding effects) has been accumulated at the valuation interest rate to Under each option, this leads to a positive (albeit sometimes very small) accumulated amount at the 2016 valuation which is used to reduce the member contribution rate over a 15 year period commencing in

8 Table D: Cost Projections: Extra 1.5% charged on all earnings of those earning over an earnings linked Band 9 Year Total Existing Member Cont Total New Entrant Cont Total Combined Paid Employers Cont Paid Members' Average Cont % 18.9% 20.7% 14.0% 6.7% % 19.7% 20.8% 14.0% 6.8% % 20.3% 20.8% 14.0% 6.8% % 20.8% 21.2% 14.0% 7.2% % 21.3% 21.6% 14.0% 7.6% % 21.7% 21.8% 14.0% 7.8% % 22.6% 22.6% 14.0% 8.6% Table E: Cost Projections: Extra 1.0% charged on all earnings of those earning over an earnings linked Band 9 Year Total Existing Member Cont Total New Entrant Cont Total Combined Paid Employers Cont Paid Members' Average Cont % 18.9% 20.7% 14.0% 6.7% % 19.7% 20.7% 14.0% 6.7% % 20.3% 20.9% 14.0% 6.9% % 20.8% 21.3% 14.0% 7.3% % 21.3% 21.7% 14.0% 7.7% % 21.7% 21.9% 14.0% 7.9% % 22.6% 22.6% 14.0% 8.6% 7

9 Table F: Cost Projections: Extra 1.5% charged on all earnings of those earning over an earnings-linked 108,600 pa Year Total Existing Member Cont Total New Entrant Cont Total Combined Paid Employers Cont Paid Members' Average Cont % 18.9% 20.6% 14.0% 6.6% % 19.7% 20.7% 14.1% 6.6% % 20.3% 20.9% 14.0% 6.9% % 20.8% 21.3% 14.0% 7.3% % 21.3% 21.7% 14.0% 7.7% % 21.7% 21.9% 14.0% 7.9% % 22.6% 22.6% 14.0% 8.6% Table G: Cost Projections: Extra 1.0% charged on all earnings of those earning over an earnings-linked 108,600 pa Year Total Existing Member Cont Total New Entrant Cont Total Combined Paid Employers Cont Paid Members' Average Cont % 18.9% 20.6% 14.0% 6.6% % 19.7% 20.8% 14.2% 6.6% % 20.3% 20.9% 14.0% 6.9% % 20.8% 21.3% 14.0% 7.3% % 21.3% 21.7% 14.0% 7.7% % 21.7% 21.9% 14.0% 7.9% % 22.6% 22.6% 14.0% 8.6% 8

10 3) Comparisons of value for assessing the opportunity cost of removing the cap 3.1 We have also been asked to produce a comparison of the potential value of uncapped benefits versus that of capped benefits. There are a number of reasons, which, when compounded together, mean that any such comparison is likely to be fraught with difficulties and inevitably based on both subjective judgements and arbitrary assumptions. At one level there is the basic difficulty that we do not have the full and detailed individual data necessary to carry out a full valuation of capped benefits. Although we can take an approximate approach, important details (for instance, the different extent to which the cap might affect part-time members) will inevitably be missed. 3.2 More fundamentally, the comparisons are being undertaken in the context of a situation in which the earnings-cap (broadly price-linked from 1990 and 108,600 pa in 2006/07) is being replaced by a Lifetime Allowance ( 1.5 million in 2006/07 increasing to 1.8 million in 2010/11). The Regulations currently in-force effectively allow members to opt-out after they have reached the Lifetime Allowance, and, given that significant tax-charges apply on benefits in excess of the Allowance, it might be expected that many would do so. Any comparisons would ideally take account of the effects of both of these limitations on benefits, but do so it is necessary to make assumptions on how the earnings cap and the Lifetime Allowance will move (or would have moved) in future. 3.3 A further complication arises if it is assumed (as it has been in work completed to date) that the level of the earnings cap would continue to be linked to price inflation. This means that the results of any comparison of benefit values will depend critically on the effective date at which it is made. As real increases in average salaries lead to the projected proportion of total payroll which exceeds the cap to increase with time, the difference in value between capped and uncapped benefits will itself tend to increase over time (although the relationship may differ if the comparison is with benefits subject to the Lifetime Allowance). In deciding on the point in time (or period of time) to which benefit comparisons should relate, there is also the additional complication of incorporating an appropriate treatment for the uncapped pre entrants. 3.4 In terms of timing, it would be expected over a thirty-year time horizon (that is up to the mid 2030s), that the vast majority of the existing scheme membership would have career earnings which never exceeded the earnings cap. The current level of the earnings cap is 108,600 (well beyond the upper end of the Agenda for Change pay grades), and a very small proportion of scheme members currently earn more than this. However, after allowing for steepening average career pay progression of 0.75% pa in addition to assumed average real earnings growth of 1.5% pa over the next 30 years, the equivalent level of the earnings cap in the mid-2030s is projected to be broadly in the middle of Agenda for Change Band 8b (about 55,000 pa currently). This would suggest that, within such a time horizon, over 90% of current members would not expect to be affected by the earnings cap. 3.5 A 30 year time horizon also coincides broadly with the period during which the existing group of high earners (on which projections of contribution yields above the cap have been based) is projected to run-off. 9

11 3.6 Beyond such a horizon of course, an increasing proportion (mainly of entrants after 2005) might be affected by the cap, but it is questionable whether successive Governments would have allowed the cap to be eroded in real terms to the extent that it was affecting the benefits of a large proportion of members of occupational pension schemes (and this point may also be of relevance to the Lifetime Allowance). Indeed, given the level of increases in the amount of the Lifetime Allowance from 2006 to 2010, it might not be unreasonable to suppose that, had the Lifetime Allowance not been introduced, adjustments might have been made to the formula underlying movements in the earnings cap so that it was more closely linked henceforth to movements in average earnings than to price inflation. 3.7 In the circumstances, it may not be unreasonable to compare benefit values subject to a price-linked earnings cap currently at the level of 108,600 pa to those subject to a Lifetime Allowance of 1.8 million in 2010/11 and linked to prices thereafter. 3.8 It is thought that about 45% of the high earners group for which contribution projections have been undertaken are hospital Consultants. NHS Employers has provided us with a typical career pay path under the Consultant Contract based on the pay structure in 2005/06. In addition, there are the effects of general earnings escalation assumed to be 1.5% pa. 3.9 Taking account of both of these elements, final uncapped earnings at retirement for a 23-year old trainee Consultant joining in 2005 on a salary of 19,700 pa would be projected to be about 10 times earnings on entry (measured in constant-price terms). This multiple decreases to just over 5 times earnings on entry if a (price-linked) earnings cap is applied or nearly 7 times if a (price-linked) Lifetime Allowance is applied (when re-expressed in the form of an equivalent reduction in assumed final earnings) This analysis suggests that, for a trainee Consultant joining in 2005 at age 23, the value of benefits subject to a price-linked earnings-capped regime might be about 75% of equivalent benefits subject to a price-linked Lifetime Allowance regime. The value of benefits subject to a price-linked earningscapped regime might be close to 50% of that of benefits which were unlimited in any way However, these probably represent over-estimates of the average differences in value, as the pay progression scales of General Practitioners and other high earners will probably not be as steep as that of hospital Consultants and no account has been taken above of part-time working. Against that, it might be contended that the impact of the Lifetime Allowance will on average be less than it is for a long-serving hospital Consultant Against the above background, we have proceeded on the assumption that for a new entrant in 2005 in any group expected to form part of a future cohort of high earners, the average value of benefits subject to the earnings cap might be about three-quarters of that of uncapped benefits. In valuing the latter, we have included some allowance for the effects of the Lifetime Allowance, and for this purpose, the Allowance is assumed to move in future in a manner consistent with the way the cap might have been expected to move had it not been abolished. 10

12 3.13 It should be noted that the above analysis (based on the post-pay modernisation pay-scales) is entirely independent of the approach taken to actuarial valuations of the scheme to date. These are based on historical data on earnings patterns, and the full effect of the new Consultant Contracts on observed salary-scales are not expected to be reflected in Scheme experience for a considerable time In theory, therefore, this analysis should not be used in connection with existing members, as a current valuation would not anticipate the full effects of the steeper salary scales expected to be eventually observed as a result of pay modernisation. However, as Consultant data are currently grouped together with other health professionals, there is no separate historical Consultant salary scale on which to rely. We have therefore for simplicity also applied the new entrant type analysis to the existing post-1989 high earners: as a result, the estimate in relation to existing members has been expressed as an upper bound The difference in net value to the member of capped and uncapped benefits also needs to take account of the extra contributions members would pay on earnings over the cap in an uncapped regime. These have been assumed to be at the existing rate of 6% of pay (any excess from higher-tier contribution rates has been incorporated elsewhere in the overall costings model) The value of 25% of benefits for the post-1989 joiners amongst the existing high earners less the value of their extra contributions amounts to less than ¼% of whole-scheme pay The equivalent amount for new entrants calculated on the assumptions described above is approximately 1% of pay. This latter can be thought of as a rate based on observed experience once the full effect of the new payscales are reflected in the data, in (say) 30 years time, but also reflecting some assumed savings from the application of the Lifetime Allowance It is almost impossible to anticipate with any degree of certainty how, if the scheme earnings cap was maintained, the resulting savings in amounts of benefits (relative to the 1999 valuation baseline) would emerge and be recognised and expressed in the results of future actuarial valuations. However, based on the analysis in this section, it might not be unreasonable to think in terms of a short-term saving of less than ¼% of pay increasing over the next three decades to about 1% of pay. 4) Cost to the Employer from lifting the earnings cap for future service 4.1 Historically, no allowance has been made in the NHSPS valuation calculations for any future savings arising as the earnings cap progressively impacts on the level of benefits of post-1989 entrants. The last formal valuation of the scheme had an effective date of 31 March 1999, well before the recent series of pay modernisation reforms were introduced. Not only did it then appear that relatively few members would be affected by the earnings cap, but there were also no individual pay data available which would have been required to cost such effects at that valuation. 11

13 4.2 As costings for the NHSPS Review have followed the valuation methodology, these have also ignored the effects of the cap. The standard contribution rates so calculated have therefore theoretically represented the costs arising in an uncapped scheme. However, as general levels of real pay increase, and particularly given the background of recent pay increases for high earners in the NHSPS under the modernisation programme, the cap is now becoming of increasing potential financial significance to the scheme as a whole. 4.3 In practice, we understand that employers have been paying standard contributions only on earnings up to the cap (in the case of members subject to and earning in excess of the cap). As standard contribution rates have been calculated as if in an uncapped scheme, the value of the extra contributions in excess of the cap which would be required in an uncapped scheme has been taken as a reasonable proxy for the future service cost to the Employer if the cap was removed. 4.4 Based on this approach, we have previously reported to the TAG that the cost of removing the earnings cap (for the future service of those existing members who are subject to it) was a rounded 0.10% pa of whole scheme pay, payable over the projected future working lifetime of the existing membership. We have maintained the same approach for the current costings, but now extend the model to also encompass new entrants. 4.5 This estimate of 0.10% of pay was based on an assumed overall population of existing members with a realistic prospect of reaching the cap of about 65,000, of whom the data available suggest that over three-quarters of this group (measured in earnings terms) are pre-1989 joiners whose benefits and contributions are not subject to the earnings cap. The earlier cost assessment in effect assumed that, where one of the group subject to the earnings cap (about 20,000) had future earnings in excess of the cap, member contributions on these earnings would become payable, and this position is maintained in the current costings. 4.6 It is also necessary to consider the potential cost of removing the earnings cap for new entrants, although there is inevitably a certain degree of subjectivity in any such assessment see paragraph 3.3 above. 4.7 We have therefore again taken a pragmatic approach under which we first assume that the earnings pattern of existing members relative to the current level of the cap (including those pre-1989 joiners not presently subject to the cap) represents an average distribution for future new entrants. The resulting costs have then been weighted by the respective projected payrolls of existing members and new entrants over the period during which existing capped members are expected to run-off (about 30 years). 4.8 Under this approach, the total cost of removing the cap for future service (including both those existing members who are subject to the cap and future entrants) is assessed as a rounded 0.30% pa of whole scheme pay (if allowance is made for member contributions in the relevant categories being extended to earnings above the cap at the rate of 6%), this being equivalent to the cost for existing members only of 0.10% of pay in paragraph

14 4.9 Our analysis is based on the assumption that the earnings cap would be applied to Practitioners benefits in a manner analogous to and consistent with that in which it is applied to Officer benefits Of the Employer cost of 0.30%, a rounded 0.15% is estimated to relate to General Medical Practitioners and the approximately equal balance to Officers GAD s original 2005 costings of removal of the earnings cap for existing members were based on 2004 data and on the valuation assumption for real increases in average earnings of 1.5% pa, plus an additional amount of 0.5% pa to allow for career pay progression. Given the limited time available, this assumption has been retained for these latest costings The additional allowance of 0.5% pa took into account the fact that a large proportion of the affected members are GPs on career average benefits, the value of which (on the valuation assumptions) is not particularly sensitive to career pay effects. However, a sensitivity test was also undertaken to assess the potential effects of pay modernisation on calculations based on the earlier data. As the latest data-set now largely reflects the effects of modernisation, the specific test applied in the 2005 work is probably no longer relevant Nevertheless, the TAG may wish us to undertake new sensitivity tests on the overall 2% pa real earnings growth assumption underlying the work in this section of the paper. Government Actuary s Department 11 September