THE PENSIONS TRUST SOCIAL HOUSING PENSION SCHEME FRS102 METHODOLOGY Month ends from 31 March 2019 to 29 February 2020
TABLE OF CONTENTS 1 Introduction... 3 2 Proposed Assumptions... 4 3 Discount Rate... 5 4 Price Inflation... 7 5 Pension Increases... 9 6 Earnings Increases... 10 7 Life Expectancy... 11 8 Commutation of Pension... 12 9 GMP Equalisation... 13 10 Other assumptions... 15 JLT FRS102 METHODOLOGY TABLE OF CONTENTS 2
1 INTRODUCTION This report outlines our recommended assumptions methodology which would be appropriate for use when preparing accounting disclosures under Section 28 of FRS 102 for employers participating in the Social Housing Pension Scheme ( SHPS ). This report has been commissioned by TPT Retirement Solutions Limited. This report sets out the proposed approach for setting the assumptions for month ends between 31 March 2019 and 29 February 2020. For illustration we have shown assumptions based on market conditions as at 31 March 2019. The same assumptions derivation methodology will be adopted for subsequent month ends up to and including 29 February 2020. The financial assumptions will be updated to allow for any changes in market conditions between 31 March 2019 and the subsequent month end. It is important to note that assumption setting is a subjective process. Other sets of assumptions may be acceptable. Employers may wish to take independent actuarial advice and discuss the assumptions with their auditors. JLT Benefit Solutions Ltd 5 April 2019 COMPLIANCE This work is subject to and complies with the Technical Actuarial Standard (TAS): Principles for Technical Actuarial Work issued by the Financial Reporting Council. JLT FRS102 METHODOLOGY INTRODUCTION 3
2 PROPOSED ASSUMPTIONS The following table sets out the proposed assumptions to use for month ends 31 March 2019 to 29 February 2020 (using market conditions as at 31 March 2019 for illustration). The financial assumptions will be updated to allow for changes in market conditions using the same derivation methodology as set out in this report. Participating employers in SHPS will have different durations and so we have provided assumptions appropriate to use for those different durations. Financial assumptions 31 March 2019 market conditions Duration Discount rate Break-even inflation (RPI) RPI inflation assumption 10 years 2.10% p.a. 3.50% p.a. 3.30% p.a. 14 years 2.25% p.a. 3.50% p.a. 3.30% p.a. 18 years 2.30% p.a. 3.50% p.a. 3.30% p.a. 22 years 2.35% p.a. 3.45% p.a. 3.25% p.a. 26 years 2.40% p.a. 3.40% p.a. 3.20% p.a. 30 years 2.40% p.a. 3.35% p.a. 3.15% p.a. Inflation (RPI) Inflation (CPI) Break-even inflation (RPI) less 0.2% RPI assumption less 1.0% Earnings growth CPI plus 1.0% Demographic assumptions Mortality: Base table Pre-retirement: Nil Post-retirement: 103% of S2PXA Improvement allowance CMI_2017 [1.25%] for males CMI_2017 [1.00%] for females Life expectancy from age 65 Pensioners (currently aged 65) Non-pensioners (currently aged 45) Commutation Other demographic assumptions 21.8 years for males 23.5 years for females 23.2 years for males 24.7 years for females 75% of members take the maximum cash at retirement As per most recent Technical Provisions assumptions The following sections provide further details behind the derivation of each of the main assumptions. JLT FRS102 METHODOLOGY PROPOSED ASSUMPTIONS 4
3 DISCOUNT RATE Proposed assumption We propose the following discount rate assumptions as at 31 March 2019: Duration Discount rate 10 years 2.10% p.a. 14 years 2.25% p.a. 18 years 2.30% p.a. 22 years 2.35% p.a. 26 years 2.40% p.a. 30 years 2.40% p.a. Sensitivity Typically increasing (decreasing) this assumption by 0.1% p.a. would decrease (increase) the liabilities in the order of 2%, although the precise impact will vary from employer to employer based on their membership profile. How assumptions are set Under Section 28 of FRS102, the discount rate used to value the defined benefit obligation is set by reference to market yields at the reporting date on high-quality corporate bonds. The currency and term of the corporate bonds must be consistent with the currency and term of the benefit obligation. We have proposed single equivalent discount rates that vary with the duration of the employer s liabilities. The discount rates have been chosen so that when they are used to discount the projected benefit cash flows underlying a pension scheme with the equivalent duration, they would give broadly the same result as using a full AA corporate bond yield curve to discount the same cash flows. The cash flows used for this calculation are sample cash flows. The full AA corporate bond yield curve is based on Bank of America Merrill Lynch s Sterling AA corporate bond yield curve which is published up to terms of 30 years. Beyond 30 years the curve is assumed to remain constant. The employer specific duration of the liabilities will be calculated using the sensitivity of the employer s liabilities to changes in the discount rate. The calculated duration will then be compared to the above discount rate table with an appropriate interpolation between the quoted duration markers. The resulting equivalent discount rate will then be used to discount the employer s obligations. JLT FRS102 METHODOLOGY DISCOUNT RATE 5
Market indicator The Bank of America Merrill Lynch (BoAML) Sterling AA Corporate Bond indices provide a benchmark for the yields available on high-quality corporate bonds. The graph below is a 12-month plot of the index yield and shows that the average long-term AA corporate bond yield on 31 March 2019 was 2.31% p.a. The following graph plots the term structure of yields on AA corporate bonds. JLT FRS102 METHODOLOGY DISCOUNT RATE 6
4 PRICE INFLATION Proposed assumption We propose the following RPI inflation assumptions as at 31 March 2019: Duration Break-even Inflation RPI inflation (RPI) assumption 10 years 3.50% p.a. 3.30% p.a. 14 years 3.50% p.a. 3.30% p.a. 18 years 3.50% p.a. 3.30% p.a. 22 years 3.45% p.a. 3.25% p.a. 26 years 3.40% p.a. 3.20% p.a. 30 years 3.35% p.a. 3.15% p.a. We also propose that the CPI inflation assumption is based on the RPI inflation assumption with a deduction of 1.0% p.a. Sensitivity For a typical scheme, increasing (decreasing) the inflation assumption by 0.1% p.a. would increase (decrease) the inflation-linked liabilities by the order of 2%. How assumptions are set The assumption for price inflation impacts the liability calculation for pensions in payment whose increases are linked to inflation, along with the revaluation of deferred pensions. It is also used as the basis for setting the earnings growth assumption. Under Section 28 of FRS102, the price inflation assumption is set by considering market expectations, for example by taking the difference between yields available on long-dated fixedinterest and index-linked gilts (known as break-even inflation ). Adjustments to this rate are sometimes made to reflect an inflation-risk premium, which includes the limited supply and high demand for the gilts. RPI We have proposed single equivalent break-even RPI inflation rates that vary with the duration of the employer s liabilities. The inflation rates have been chosen so that when they are applied to the projected benefit cash flows underlying a pension scheme with the equivalent duration, they would give broadly the same result as applying the full Bank of England inflation curve to discount the same cash flows. The cash flows used for this calculation are sample cash flows. The Bank of England inflation curve is published up to terms of 40 years. Beyond 40 years the curve is assumed to remain constant. JLT FRS102 METHODOLOGY PRICE INFLATION 7
A deduction of 0.2% p.a. has been made to the break-even RPI inflation rates to allow for an inflation risk premium. The employer specific duration of the liabilities will be calculated using the sensitivity of the employer s liabilities to changes in the discount rate. The calculated duration will then be compared to the above inflation rate (RPI) table with an appropriate interpolation between the quoted duration markers. The resulting equivalent RPI inflation assumption will then be used in the calculations of the employer s liabilities at the month end. CPI At present there is no indicator of market expectations of CPI inflation. However, forecasts suggest that CPI inflation is expected to be around 0.8% p.a. to 1.2% p.a. below RPI over the longer term. We recommend that the assumption for CPI inflation is 1.0% p.a. below the RPI inflation assumption. Market indicator The following graph is a 12-month plot of the RPI inflation rate implied by the market and shows that at 31 March 2019 the rate was 3.59% p.a. The rates shown are 15-year projections published by the Bank of England. So 3.59% p.a. is currently the market s expected average rate of RPI inflation that will be experienced over the next 15 years. The following graph plots inflation spot curves for terms of up to 30 years. JLT FRS102 METHODOLOGY PRICE INFLATION 8
5 PENSION INCREASES How assumptions are set Increases to pensions in payment are typically either at a fixed rate, or in line with inflation subject to certain caps and collars. The most common type of increase in payment is in line with inflation, subject to a maximum increase in any one year of 5% p.a. or 2.5% p.a. and a minimum of 0% p.a. The assumption is set by considering the likelihood of inflation being above the cap or below the collar in future years, and applying an adjustment to the relevant inflation assumption to reflect this. A statistical model for the distribution of future inflation returns has been used to determine the appropriate adjustment. Based on the relevant RPI and CPI inflation assumptions, we will use the Black-Scholes model and assume an expected future volatility of price inflation of 1.75% p.a. to derive the various inflation-linked pensions in payment assumptions. CARE revaluation assumptions will be derived using a similar approach to that adopted to derive the increases to pensions in payment assumptions. Deferred revaluation Revaluation of deferred pensions in line with inflation is generally subject to a maximum over the full period to retirement. In practice, the rate of deferred revaluation is taken as the rate of inflation, as over long periods, it is not expected that the average future inflation is likely to exceed the maximum rate. If the maximum rate is less than the average future inflation rate, then the assumption will be subject to the relevant maximum rate. Sensitivity This is not being shown separately because of the main link to inflation. JLT FRS102 METHODOLOGY PENSION INCREASES 9
6 EARNINGS INCREASES Proposed assumption We propose that earnings growth is assumed to be 1.0% p.a. above CPI inflation. Sensitivity Increasing (decreasing) this assumption by 0.1% p.a. would typically increase (decrease) earningsrelated liabilities by 1% p.a. How assumptions are set The earnings growth assumption is used to project accrued pensions for current active members. It is widely recognised that earnings increases in the economy are related to price inflation; hence it is common to look at the earnings growth assumption relative to the price inflation assumption (this is known as real earnings growth ). Earnings growth could be considered by reference to RPI or CPI inflation. Earnings growth in excess of inflation should reflect best estimate long term increases for the active members of the specific employer and is expected to be positive over the long term as it must take into account not only inflationary increases, but also promotional increases. Therefore this is an area in which the employer will be able to provide the best view, and this assumption will vary from employer to employer. We propose that the default earnings growth assumption is set at 1.0% p.a. above CPI inflation. This is a central assumption that aims to broadly reflect long-term earnings growth expectations for SHPS as a whole. The accounting disclosure modeller will allow employers to vary this assumption to best suit the specific profile of their own workforce. JLT FRS102 METHODOLOGY EARNINGS INCREASES 10
7 LIFE EXPECTANCY Proposed assumption We propose that base tables for mortality rates are set in line with SAPS tables S2 (All Pensioners Pensions Amounts) with a best estimate scheme specific scaling factor. The best estimate scaling factor is 103% for SHPS. The base tables will then be projected using CMI_2017 projection model with a 1.25% p.a. long term improvement rate for males and a 1% p.a. long-term improvement rate for females. We propose no allowance is made for mortality before retirement. Sensitivity Adding 1 year to the life expectancy typically adds approximately 2% to the liabilities. How market assumptions are set There are various base mortality tables based on different data sets and different time periods, the most recent ones being known as the SAPS (self-administered pension schemes) tables. These standard tables are then typically adjusted to allow for any expectation of higher/lower life expectancy of scheme members due to geographic, socio-economic or demographic factors. The Trustee commissions a regular analysis of the membership profile for SHPS using Club Vita. Club Vita has collected a large amount of mortality experience data on UK defined benefit schemes and has used this to identify a number of characteristics which distinguish whether people live shorter or longer than others using various identifiers such as gender, postcode, salary/pension and occupation. These characteristics were overlaid onto the scheme membership to identify a mortality assumption for each individual member. The results are then used to derive a scheme specific adjustment factor for SHPS based on the standard S2 SAPS table which gives broadly the same result as applying different mortality assumptions for each individual member. Allowance is then made for improvements in life expectancy since the tables were made, and for further improvements in future. The Continuous Mortality Investigation (CMI) produces annual projections for this which are now in common use. We propose using the same mortality improvements adopted by the Trustee for the 2018 valuations but adjusted to remove the margin for prudence i.e. CMI_2017 improvements with a 1.25% p.a. long term improvement rate for males and a 1% p.a. long-term improvement rate for females. A new version of the annual projection model (CMI_2018) has been released but further consideration is required before it is adopted noting that it includes additional parameters that were not included in previous versions. Mortality rates before members retire are not expected to have a material impact on the defined benefit obligation. It has therefore been assumed that all members will survive to their assumed retirement dates. JLT FRS102 METHODOLOGY LIFE EXPECTANCY 11
8 COMMUTATION OF PENSION Proposed assumption We propose that the commutation allowance assumes that 75% of members take the maximum cash available at retirement using notional commutation factors (or current factors if higher). A sample of the notional commutation factors used is shown below: SHPS Commutation factor CPI 5% CPI 2.5% Age 60 16.424 15.580 Age 65 14.531 13.873 Sensitivity The difference between assuming nil commutation and the maximum incidence and level of commutation typically amounts to between 4% and 11% of the relevant liabilities (i.e. non-pensioner liabilities). This depends on the generosity or otherwise of a scheme s own commutation factors. How assumptions are set The majority of pension scheme members typically take the maximum available cash lump sum at retirement. It is possible that schemes can be structured so that cash can be taken from a defined contribution section first, which reduces the amount of pension which is commuted from the defined benefit section. A best estimate assumption would normally be set by examining the take up from recent retirements. The assumption that 75% of members take the maximum cash available to them is based on a study commissioned by the Trustee in October 2015 into the take up rate of cash commutation at retirement. JLT FRS102 METHODOLOGY COMMUTATION OF PENSION 12
9 GMP EQUALISATION Introduction The English High Court ruling in Lloyds Banking Group Pension Trustees Limited v Lloyds Bank plc and others was published on 26 October 2018, and held that UK pension schemes with Guaranteed Minimum Pensions (GMPs) accrued from 17 May 1990 must equalise for the different effects of these GMPs between men and women. The case also gave some guidance on related matters, including the methods for equalisation and decided that method C2 was lawful in principle and met the minimum requirements to achieve equality. Method C2 provides the better of male and female comparator pensions each year, subject to accumulated offsetting and an allowance for interest. The court decided that other methods are available, but would require employer agreement. The Trustee of The Pensions Trust Social Housing Pension Scheme (the Scheme) will need to obtain legal advice covering the impact of the ruling on the Scheme, before deciding with the employers on the method to adopt. The legal advice will need to consider (amongst other things) the options for GMP equalisation solutions, whether there should be a time limit on the obligation to make back-payments to members (the look-back period) and the treatment of former members (members who have died without a spouse and members who have transferred out for example). An allowance will need to be made for the expected impact of equalisation, which will require assumptions about how equalisation will be achieved. Assumptions method C2 applies to past and future benefit payments there is no limit on the look-back period for rectification only consider members who currently have GMP liabilities within the Scheme (not members who have died without a spouse or members who have transferred out for example). The calculation result represents the ultimate cost of GMP equalisation for the Scheme Calculation approach JLT have centrally developed a financial model to estimate the percentage impacts of applying a C2 equalisation methodology on the values of GMP liabilities accrued between 17 May 1990 and 5 April 1997. This model allows for the different features of active, deferred and pensioner members, and as such caters for a wide range of scenarios enabling an assessment to be made regarding the impact of GMP equalisation. An allowance is then made for the historic underpayment for members currently in receipt of a pension. JLT FRS102 METHODOLOGY GMP EQUALISATION 13
This model has been used to derive model points applicable to SHPS which vary by: - GMP and excess revaluation rates (pension increases before retirement) - Sex Statutory GMP revaluation. For the excess RPI 5% revaluation. Male or Female - Date of leaving Based on 5 year age bands - Current age Based on 5 year age bands - Retirement age Members are able to retire from age 60 unreduced - Pension increase rates for GMP and excess Statutory GMP pension increases. For the excess CPI 5% pension increases. - Normal Retirement Age (NRA) Normal Retirement Age of 65 This produces 42 model points that are relevant for SHPS. Each member in SHPS is allocated to the most appropriate model point which provides a percentage impact of applying a C2 equalisation methodology to that member s relevant GMP. This percentage is then applied to each member s relevant GMP liabilities and combined to produce an employer by employer impact. The data used for the calculations was provided by TPT Retirement Solutions Limited and has an effective date of 31 March 2018. The resulting employer by employer percentage impacts will be applied to the calculated liabilities at the relevant balance sheet date. JLT FRS102 METHODOLOGY GMP EQUALISATION 14
10 OTHER ASSUMPTIONS Other assumptions We propose that no assumption is made for the award of any discretionary benefits. All other assumptions are proposed to be consistent with those used for the most recently completed triennial actuarial valuation. JLT FRS102 METHODOLOGY OTHER ASSUMPTIONS 15
CONTACTS Murray Wright FFA Principal murray_wright@jltgroup.com Eric Kwon FIA Senior Consultant eric_kwon@jltgroup.com JLT Employee Benefits St James s Tower 7 Charlotte Street Manchester M1 4DZ Whilst all reasonable care has been taken in the preparation of this presentation no liability is accepted under any circumstances by Jardine Lloyd Thompson for any loss or damage occurring as a result of reliance on any statement, opinion, or any error or omission contained herein. Any statement or opinion unless otherwise states should not be construed as independent research and reflects our understanding of current or proposed legislation and regulation which may change without notice. The content of this document should not be regarded as specific advice in relation to the matters addressed. JLT Benefit Solutions Limited. Authorised and regulated by the Financial Conduct Authority. A member of the Jardine Lloyd Thompson Group. Registered office: The St Botolph Building, 138 Houndsditch, London EC3A 7AW. Registered in England No. 02240496. VAT No. 244 2321 96.