Skip to content

Annex 2: Data Sources and Methods for the UK Supplementary Table on Pensions, 2010

Released: 27 April 2012 Download PDF (98.7 Kb)

Abstract

This document is an annex to the article ‘Pensions in the National Accounts – A fuller picture of the UK’s funded and unfunded pension obligations’ published on 27 April 2012. Owing to complexity of the UK pension system, there is no ‘national model’ or single data source for the supplementary table on pensions. Each column has been compiled using the most appropriate source available.

Introduction and contents

This document is an annex to the article ‘Pensions in the National Accounts – A fuller picture of the UK’s funded and unfunded pension obligations’ published on 27 April 2012. Owing to complexity of the UK pension system, there is no ‘national model’ or single data source for the supplementary table on pensions. Each column has been compiled using the most appropriate source available.

This document is in three parts. Part 1 lists the sources used in each column, and Part 2 provides a brief description of the sources used. Part 3 provides details of any modelling or assumptions used (with references to published articles for further details).

Part 1: Sources used in each column

Sources used in each column

Column Source data used
A – group personal pensions (defined contribution) Financial Services Authority (FSA) data for Rows 1 and 10. HM Revenue and Customs (HMRC) data for actual contributions (Rows 2.1 and 2.3). Office for National Statistics (ONS) insurance companies survey data for other rows.
A – occupational pensions (defined contribution) For Rows 1 and 10 (insured pensions): FSA data. For Rows 1 and 10 (self-administered pension funds): ONS self-administered pension funds survey.  For other rows: ONS self-administered pension funds survey or ONS insurance companies survey, as appropriate.
B – occupational pensions (defined benefit) 2011 Purple Book estimates for Row 1 and Rows 7 to 10. A small number of Pension Protection Fund (PPF)-eligible schemes that are classified in Column E are excluded. ONS self-administered pension funds survey for Rows 2 to 6.
E Pension scheme valuations and annual resource accounts. Data for the Local Government Pension Scheme (LGPS) used in Rows 2 to 6 is provided by the Department for Communities and Local Government (for LGPS England and Wales) and by the Northern Ireland Local Government Officers’ Superannuation Committee; figures for Scotland are estimated (see Part 3). For other schemes, information comes from scheme valuations and annual accounts; in the 2010 table, much of it is estimated (see Part 3).
G Pension schemes’ annual resource accounts. This column brings together 13 sets of accounts for unfunded schemes, plus estimates for the police and fire schemes (see Part 3).
H Department for Work and Pensions (DWP) Forecasting Division’s model (see Part 3). Rows 2.1 and 2.3 are adjusted for consistency with actual contributions paid into the National Insurance Fund, as published in the core National Accounts.

Download table

Part 2: Relevant data sources

Financial Services Authority (FSA)

The FSA has administrative data for pension reserves held by insurance companies in respect of private pensions. The dataset covers reserves for both the accumulation and decumulation phases (the latter comprises reserves associated with annuities in payment), and for both occupational and personal pensions. ONS has access to this data for use in the National Accounts. Of personal pensions, only group personal pensions are included in the supplementary table; individual personal pensions are excluded because they are not part of pensions in social insurance.

Three key adjustments are needed in order to make the FSA’s insurance company pension reserves data suitable for use in the supplementary table. These are:

  1. Reinsurance is subtracted from gross reserves to produce reserves net of reinsurance.

  2. Entries for ‘insurance‐managed’ funds are removed to avoid double counting, as insurance-managed funds are included in ONS’s survey of self‐administered pension funds. This is done using a list provided to ONS by the FSA, which identifies companies that principally do insurance‐managed pensions business.

  3. Removal of some entries for intra‐group annuity reinsurance on advice from FSA.

HM Revenue and Customs (HMRC)

HMRC publishes tables based on administrative data showing contributions to personal pensions (including stakeholder pensions) on the HMRC pensions statistics webpages. The supplementary table only uses the data for employer‐sponsored pensions (Table PEN 2).

ONS surveys of pension funds and insurance companies (MQ5)

ONS runs a series of annual and quarterly surveys of insurance companies, self‐administered pension funds, investment trusts, unit trusts and property unit trusts. These are known as the MQ5 series. Income and expenditure data are provided for insurance companies (quarterly and annual) and self-administered pension funds (quarterly). In addition, annual balance sheet data are reported for all sectors.

The 2011 Purple Book

The Pension Protection Fund (PPF) has a unique dataset that covers funded defined benefit schemes in the private sector. The PPF and the Pensions Regulator (TPR) use this dataset to produce an annual publication called the Purple Book. The sample taken for the 2011 Purple Book covered 6,432 schemes in 2010/11, representing 98.2 per cent of PPF‐eligible schemes and 99.6 per cent of their total liabilities. Most private sector defined benefit schemes are PPF‐eligible.

The Purple Book dataset contains estimates of liabilities which are reported by private sector defined benefit schemes on the basis of a standard set of assumptions. This allows the PPF to produce estimates of total liabilities on a ‘full buy‐out’ basis, which uses a Gilts‐based discount rate (a risk‐free market rate). These estimates are used in the supplementary table.

However, the full buy‐out estimates are only published for the financial year. In order to convert to a calendar year basis, ONS uses the PPF’s estimate of liabilities on a s179 basis, which is published on a monthly basis in the PPF’s 7800 index. To produce the closing balance for Column B (Row 10), the ratio of full buy‐out liabilities to s179 liabilities at the end of the financial year is used to rate up the s179 estimate at the end of the previous calendar year (see Part 3 for details).

Figures for Rows 7 to 9 are estimated by ONS after consultation with the PPF, which advises ONS of any changes to key assumptions and the approximate size of their impact on these rows.

Pension scheme valuations

Special requests to Department for Communities and Local Government (responsible for LGPS and firefighters’ scheme) and Home Office (responsible for police officers’ scheme) and the Scottish Public Pensions Agency (for Scottish LGPS, firefighter and police schemes); and to Government Actuary’s Department (the actuary for all these schemes).

Pension scheme resource accounts

Internet searches and special requests to individual scheme administrators.

Northern Ireland Local Government Officers’ Superannuation Committee (NILGOSC)

Annual reports and valuation reports supplied on request.

Department for Communities and Local Government

For income and expenditure of England and Wales LGPS: SF3 Statistical Release “LOCAL GOVERNMENT PENSION SCHEME FUNDS ENGLAND” and “LOCAL GOVERNMENT PENSION SCHEME FUNDS WALES

Scottish Government

For income and expenditure of Scottish LGPS: Scottish Local Government Financial Statistics publications pages.

Part 3: Methods, modelling and assumptions

1. Column A (defined contribution pensions, non‐general government)

This column is compiled using three spreadsheets: (i) self‐administered pension funds, (ii) occupational insured pensions, and (iii) Group Personal Pensions (including group stakeholder pensions). Originally spreadsheet (ii) was split into two parts for defined contribution (DC) and defined benefit (DB) pensions with the latter designed to feed into Column B, but this part was found to be so small (and estimates not available on an actuarial basis) that it was not worth treating separately.

There is no complex modelling used in Column A, but the following ratios are applied:

(i) For self‐administered funds, where data is from the MQ5 series:

a. The appropriate MQ5 uplift factor is used to correct for incompleteness in the sampling frame (as with core National Accounts estimates that use MQ5 data for self‐administered pension funds);

b. The 2010 supplementary table uses a ratio to split out the DC element from the total (DB+DC) – this ratio assumes that the DB/DC split for employer and employee actual contributions (for which MQ5 collects data separately) can be applied to other lines for which no split is currently available. Improvements to the MQ5 questionnaire from January 2012 should reduce the need to make such assumptions in future.

(ii) For occupational insured pensions, data is from FSA and MQ5. In order to produce estimates of contributions and benefits paid from the MQ5 data, it is necessary to first adjust the relevant series by (a) removing information for insurance‐managed funds and (b) removing transfers (net of transfers for any companies identified as insurance‐managed funds). In addition, for occupational insured pensions, assumptions are used to estimate:

a. The proportion of decumulation phase assets (comprising reserves associated with annuities in payment, from FSA) associated with DC pensions. This ratio is based on the proportion of DC assets in the accumulation phase using FSA data.

b. The proportion of actual social contributions (from MQ5) paid by employers and employees (as is currently the case in the core National Accounts). This ratio is based on information about employer and employee contributions from ONS’s Occupational Pension Schemes Survey.

c. For interest and dividends in respect of insurance and pensions business (from MQ5), the proportion attributable to DC pensions. This ratio is based on the ratio of DC pension assets to total insurance and pension reserves using FSA data.

For occupational insured pensions, one further adjustment is made: part of the benefits paid under the heading 'individual personal pensions’ by insurance companies are re‐allocated to the line for benefits paid by occupational insured pensions (thus bringing them within the scope of the supplementary table). This is because although such pensions are recorded as occupational in the accumulation phase, they disappear from the occupational pensions category when the pension fund is converted into an annuity on retirement, and insurance companies then record them under the heading of ‘individual personal pensions’. The methodology here is not robust, but at present, given the lack of information, the only other option would be to make no adjustment. In our view, not making such an adjustment would distort the results and is therefore a worse option than making a crude adjustment.

(iii) For Group Personal Pensions (GPPs), the data is from FSA, HMRC and MQ5. The assumptions made are:

a. For the opening and closing balances in Rows 1 and 10, the FSA data shows the proportion of accumulation phase reserves that are in GPPs, but does not show the proportion of decumulation phase reserves in GPPs. The proportion in the decumulation phase is assumed to be the same as in the accumulation phase.

b. For contributions data (from HMRC), a conversion from financial year to calendar year is necessary; this is done on a pro rata (¼ + ¾) basis.

c. For interest and dividends in respect of insurance and pensions business (from MQ5), it is necessary to estimate the proportion attributable to GPPs. To do this, a ratio is used which is based on the ratio of GPP pension reserves to total insurance and pension reserves (using FSA data).

d. For benefits paid and transfers (from MQ5), it is necessary estimate the proportion that are related to GPPs as opposed to individual personal pensions. This ratio is based on MQ5 contributions data, which shows the split between GPPs and individual personal pensions (this split is not available for benefits paid/transfers).

2. Column B (defined benefit pensions, non‐general government)

This column is compiled using one spreadsheet: for self‐administered pension funds. The data used is a combination of MQ5 and PPF data. For the lines that use MQ5 data, the same approach is followed as for the self‐administered funds spreadsheet in Column A. For the lines that use PPF data: 

  • Rows 1 and 10 are calculated using the method in Box 1. 

  • Rows 2.4 and 7, 8 and 9 are compiled by ONS after consultation with the PPF (including seeking advice on what is the appropriate discount rate to use for the year in question).

Note that a proportion of PPF liabilities must be subtracted from Column B and re‐allocated to Column E in respect of schemes which the PPF includes in its private sector universe but which are classified by the National Accounts as the responsibility of general government. Advice is sought from PPF on what proportion to subtract (see Box 1).

3. Columns E, G and H (defined benefit pensions, general government)

The methodology for producing estimates for these columns is complicated, as it is based either on the results of actuarial modelling that has already been carried out for schemes’ valuations and resource accounts (Columns E and G) or on modelling by DWP’s Forecasting Division at ONS’s request (Column H). In both cases, adjustments have to be made to meet the specifications of the supplementary table on pensions. The two approaches are explained in detail in Levy (2011a) and Levy (2011b) respectively – see Further reading section.

At present, some of the data sources needed to produce accurate estimates for Columns E and G is not published. For the 2010 table, the following approach has been used to fill gaps where data is missing:

a. In Column E (funded employee schemes for which Government is responsible), the largest scheme is the LGPS. ONS has obtained information for England, Wales and Northern Ireland but figures for Scotland are estimated using a rating up factor as no scheme valuations or resource accounts are available for Scotland. The rating up factor for Scotland is based on figures for scheme assets at market value, which is available for the UK as a whole and its constituent countries.

b. For other schemes in Column E, an approximate approach is used for the 2010 table because although there is some information from resource accounts and valuations that have been published in the past, these have been published at different dates and some contain only very limited information. In some cases no information at all is available, but this tends to be for very small schemes.

c. In Column G (unfunded public sector employee schemes), information is available for all schemes except those for the police and firefighters. For the police, there are no published scheme valuations. A scheme valuation has been published for English firefighters, showing figures as at 31 March 2007, and this could be used to produce figures for England using the methodology described in Levy (2011a). However, there are no published valuations for Wales, Scotland or Northern Ireland. Therefore in the 2010 table, a ratio has been used to rate up the estimates of the schemes for which information is available (in the form of annual resource accounts) to produce estimates for the whole universe. This ratio is based on the relationship between the Whole of Government Accounts (WGA) estimate of liabilities of all unfunded public sector employee schemes and the WGA estimate of liabilities of all schemes except police and fire. It is used to produce estimates for all rows in Column G, not just the opening and closing balances. The Home Office and the Department for Communities and Local Government, which are responsible for producing valuations for the police and fire schemes respectively, have plans to publish them soon. Therefore it is expected that in future years the Column G estimates will be able to use this information instead of rating up the resource accounts estimates to include police and fire.

Further reading

Columns E and G sources and methods:

Levy (2011a): Pensions in the National Accounts: compiling a complete picture of UK pensions including unfunded pensions for public sector employees (methodology article 2011), Office for National Statistics

Column H sources and methods:

Levy (2011b): Pensions in the National Accounts: compiling estimates of state pension obligations for the National Accounts (methodology article 2011), Office for National Statistics

General reading:

ONS (2011): Pension Trends Chapter 14: Pensions in the National Accounts, 2011 edition.

Background notes

  1. Details of the policy governing the release of new data are available by visiting www.statisticsauthority.gov.uk/assessment/code-of-practice/index.html or from the Media Relations Office email: media.relations@ons.gsi.gov.uk

Content from the Office for National Statistics.
© Crown Copyright applies unless otherwise stated.