FOI reference - FOI-306
Date - 2 September 2024
Request
Table 1
Please can you provide the information as set out in the attached example table 1 in respect of the year to 31 March 2024. Please can the data be provided in a separate table for each of the 4 covenant groups (Strong, Tending to Strong, Tending to Weak and Weak).
Table 2
Please can you provide the information as set out in the attached example table 2 in respect of asset data with an effective date on or after 31 March 2023 (recognising that schemes don't always provide updated asset splits in their annual returns). Please can the data be provided in a separate table for each of the 4 covenant groups (Strong, Tending to Strong, Tending to Weak and Weak).
In addition, please can you confirm the number of schemes who have notified The Pensions Regulator of their intention to refund surplus over the year to 31 March 2024. If there are any, please can you split those between ongoing schemes, and those schemes in wind up.
Response
I confirm that we hold some of the information you have requested. The information we have been able to supply is in the spreadsheet (Excel, 80KB).
Due to restrictions regarding the sharing of individual scheme data that we hold, we have not provided the full set of ranges requested, since from these it may have been possible to identify individual schemes. We have therefore expanded the ranges requested, where needed, to ensure that individual schemes cannot be identified.
I do need to make you aware of the following caveats and limitations which apply to the information we have provided.
Modelling methodology
The data for funding level distributions on a Technical Provisions basis and Buyout basis have been estimated by the actuarial team at TPR by adjusting the position from a schemes latest triennial Part 3 valuation (some of which relate to scheme’s positions as far back as 2019) up to 31 March 2024. These estimates allow for the following calculation methodology:
In estimating the technical provisions, we assume that trustees do not make any changes to the funding methodology due to changing market conditions. We assume financial assumptions were set with reference to gilt markets and adjust for changes in gilt markets only.
In estimating buyout liabilities, we make allowance for changes in gilt markets and some broad adjustments to allow for changes in insurance premia over time.
In estimating assets 31 March 2024, we use relevant index returns based on the asset allocations submitted to TPR in the 31 March 2024 Scheme Return. (These asset allocations are as at the date of each Scheme’s latest audited accounts, rather than all being a consistent date).
The key assumptions underlying our model include.
- Trustees do not take any management action in regards of buying/selling assets or changes to the investment allocation over time;
- Assets are adjusted in line with market indices, plus estimated benefit outgo, future accrual contributions and deficit repair contributions;
- Assets are rebalanced daily in order to retain the same investment allocation.
There are many more simplifications and approximations in the methods we use to estimate aggregate and individual funding positions, compared with the more robust calculations carried out for formal valuation and recovery plan reporting by scheme actuaries for trustees. Additionally, the greater the magnitude of change in market conditions, the less reliable the simplified method and data will be in illustrating the impact. It should be noted that this is not a TPR-specific issue, but a global actuarial issue when using the approximate ‘roll-forward’ methodology to estimating assets and liabilities at alternative dates.
Where asset data indicates that schemes have leveraged LDI (Liability Driven Investment), we have made broad adjustments to the asset values for any collateral calls that may have applied.
Given the high level data that we hold, our calculations use broad assumptions and approximations combined with general actuarial methods and techniques. We cannot take account of all scheme-specific characteristics and the actual position of individual schemes will vary, depending on a number of individual factors not covered in our data or methodology.
Furthermore, it should be noted that results are always approximate in nature and whilst we do not expect a pronounced systemic bias in the model, results may be materially inaccurate at an individual scheme level where experience differs to those of our key assumptions. This could be because of any of the following.
- Trustees taking positive management action to change investment strategies, which are not reflected in our current data set due to time lag issues; and/or
- Actual asset returns are materially different to index returns or scheme experience materially differs from that assumed.
Raw data
The scheme population was derived from TPR’s Register of Pension Schemes as at 31 March 2024 and includes all private sector occupational Defined Benefit (DB) and Hybrid schemes (and sections of schemes) subject to Part 3 funding requirements. This data is based on the information provided to us from the 2024 annual scheme return and recovery plan submissions received up to that date.
The data covering Covenant Grades (CG) is drawn from a number of sources, the principal source being the dataset underpinning the regulator’s annual Scheme Funding Statistics Publication. In certain instances, we have supplemented this with more recent CG assessments, or assessments estimated using in house models.
The raw data and model outputs have been subject to actuarial data checks which look to highlight key outliers and are adjusted as necessary.
We have also used external data sources, these include.
- Market indices data obtained from London Stock Exchange Group (LSEG) in order to estimate assets as at 31 March 2024; and
- Office for National Statistics (ONS) to provide estimates of benefit outgo and contributions in respect of future accrual.
We rely on the accuracy of the third-party information that is provided to us. In some instances, the information is incomplete, and we take steps to estimate information via other means.
It is important to note that:
- the data is historical. Schemes have up to 15 months to complete a valuation and need to only undertake a valuation every three years. As such, the asset value held is generally between one year and four years out of date
- the data is a high-level summary of the results of the valuation i.e. we do not have individual member data or detailed benefit structure information that schemes have access to
- the asset breakdown is based on the last audited accounts over the scheme return year. Schemes have up to seven months from the year end to complete their audited accounts
As such, the asset breakdown is generally around two years out of date.
When referring to 'schemes' in this report this is in fact referring to individual segregated sections of assets and liabilities: for example, the data for this modelling includes 4,868 sections as at March 2024. This report describes these as 'schemes' for simplicity, in line with common usage. It should be noted that some of these 'schemes' are subsections of a single scheme, so the number of schemes (if defined as separate trusts) is less than this.
Caveats and limitations of advice
We are content that the data used in the model is appropriate to provide high-level estimates of the distribution of the DB scheme universe as at 31 March 2024.
As the model has been calculated at an aggregate level, the results should not be used to draw conclusions for individual schemes, nor provide any analysis for scheme funding levels.
Compliance
This work complies with Technical Actuarial Standard 100 (TAS 100) v2.0, as published by the Financial Reporting Council (FRC). TAS 100 v2.0 applies to technical actuarial work (as defined in Section 4 of TAS 100 v2.0) that is completed on or after 1 July 2023.