This article presents the first set of official statistics on the total obligations, or gross liabilities, of UK pension providers including the UK government. These total pension obligations are also the total pension entitlements of UK households. The article presents breakdowns for private sector pensions, workplace pensions for which the government is responsible (funded and unfunded) and state pensions. The estimates presented here are for 2010 and have experimental status. They have been developed to meet a new requirement for a ‘supplementary table on pensions’ in the National Accounts contained in the internationally agreed 2008 System of National Accounts (SNA2008) and the 2010 European System of Accounts (ESA2010). It is currently expected that all EU countries will be required to produce the supplementary table on pensions for the year 2012 onwards. The 2010 table is produced on a voluntary basis and the UK is the first EU country to publish it.
At the end of 2010, the total accrued-to-date pension obligations of all UK pension providers were estimated at £7.1 trillion, nearly five times the UK’s Gross Domestic Product (GDP). The total comprised £5.0 trillion of government obligations and £2.1 trillion of private sector obligations.
Of the £5.0 trillion pension obligations for which the UK government was responsible at end-2010, £3.8 trillion were in respect of state pensions (263 per cent of GDP). The latest comparable EU-level estimate (for end-2007) was 278 per cent of GDP.
Obligations relating to unfunded pensions for public sector employees in the UK at end-2010 were estimated at £0.9 trillion (58 per cent of GDP). The latest comparable EU-level estimate (for end-2007) was 52 per cent of GDP.
The UK government’s obligations in respect of funded and unfunded workplace pension schemes at end-2010 were estimated at £1.2 trillion (80 per cent of GDP).
UK private sector obligations in respect of workplace pensions – all of which were funded – were estimated at £1.7 trillion at end-2010 (118 per cent of GDP).
There are two annexes to this article. Annex 1 presents detailed results and sensitivity analysis for the state pensions estimates. Annex 2 gives further detail on the data sources and methods used for compiling the estimates.
This article presents new experimental statistics which have been developed for the first National Accounts ‘supplementary table on pension schemes in social insurance’, for the year 2010. This table shows the total pension obligations, or gross liabilities, of UK pension providers, including the government. These are also the total pension entitlements of UK households. Although the 2010 supplementary table on pensions covers workplace pensions only, in line with National Accounts definitions of pensions in social insurance, this article also provides estimates for individual personal pensions, including stakeholder pensions.
In this article, pensions are defined as entitlements to income payments in retirement resulting from contributions made by individuals and their employers during working life. They include any related benefits that may be provided by pension schemes – such as survivor, disability and early retirement pensions – if they are an integral part of the schemes’ benefits. They comprise both state and private pensions, but they do not include income-related (non-contributory) state benefits for pensioners such as Pension Credit or other forms of social assistance such as health and social care.
Most of the key statistics presented here do not appear in the UK National Accounts (Blue Book) 2010. The UK supplementary table on pensions in social insurance has been developed to meet new requirements for the National Accounts contained in the internationally agreed 2008 System of National Accounts (SNA2008) and the 2010 European System of Accounts (ESA2010), which is the EU-specific interpretation of the SNA2008. It is currently expected that all EU countries will be required to produce the table for the year 2012 onwards. The 2010 table is produced on a voluntary basis and the UK is the first EU country to publish it. It should be noted that, as the methodology is still subject to revision, the 2010 table (published here) is classified as 'experimental statistics’.
This article should be read in conjunction with two previous articles on methodology development for the table (Levy 2011a and Levy 2011b). Readers should also note that another article published concurrently with this one (Hobbs 2012) discusses how the pensions figures fit into wider measures of the public sector balance sheet in the UK.
The statistics presented here are new in several ways. First, the UK National Accounts show only the liabilities of funded pensions (those which build up assets in order to fund the benefits paid in retirement); they do not include the liabilities of unfunded pensions (where obligations are not underpinned by a fund). State pensions and most public sector employee pensions are unfunded. The new supplementary table shows the liabilities of unfunded pensions, including state pensions, as well as those of funded pensions. An official estimate of state pension liabilities is published here for the first time since 2005.
Second, the estimates of the liabilities of funded pensions that currently appear in the UK National Accounts are based on the market value of assets. The new supplementary table uses actuarial calculations to estimate liabilities for all defined benefit (DB) pensions (see Section 2); this is now generally accepted as the best way to present such estimates. For funded DB pensions, the market value of assets may be less than the actuarial value of obligations outstanding; in other words, there may be a funding deficit.
While actuarial methods produce good estimates of pension liabilities for DB schemes, interpreting the results is not straightforward. This is because actuarial methods involve modelling based on assumptions, which, in the case of pensions, involves projections extending over a 100-year time horizon. These projections are extremely sensitive to the assumptions used, in particular the rate used to discount future payments back to their present value (the discount rate). For estimating the government’s pension obligations in the supplementary table, there is an EU-level requirement to use a stable discount rate of 3 per cent (real) or 5 per cent (nominal) (see Annex 1). The sensitivity analysis presented in Section 4 shows the impact of varying this assumption for unfunded public sector employee pension schemes and state pensions in the UK.
A third way in which the statistics presented in this article are new is that they show liabilities in respect of funded pensions separately from those relating to life assurance. At present in the estimates of liabilities shown in the UK National Accounts, funded pensions are not separately identifiable from life assurance.
Fourth, the new supplementary table on pensions shows how changes in pension liabilities each year are explained by various factors, including policy changes, changes in financial assumptions (principally the discount rate) and any other assumptions (principally on life expectancy). Thus it will be easier in future to assess the effect of any changes in the pension system on the government’s total pension obligations.
Some of the information included in the 2010 supplementary table on pensions has already been published elsewhere in a different format. For instance, recent estimates for the liabilities of unfunded public sector employee pensions have been published by Treasury in the Whole of Government Accounts (HM Treasury 2011) and in the Office for National Statistics (ONS) series on wider measures of public sector debt (Hobbs 2010a and 2010b). However, the estimates of the unfunded liabilities of the UK government have not included state pensions, and they are based on variable (market) discount rates. Variable discount rates make it hard to compare UK results with those of other EU countries.
Pension liabilities do not represent debt in that they do not constitute previous borrowing which has to be serviced or ultimately repaid. However, they are liabilities in the sense that they represent the obligations outstanding implied by current pension rules and legislation. In the case of unfunded pensions which are the responsibility of government, these obligations need to be met out of future revenues and receipts. Of course pension rules and legislation can be changed and, for this reason, pension liabilities are properly defined as ‘contingent pension obligations’ rather than debt1.
It is also important to note that the supplementary table is prepared on an ‘accrued-to-date’ basis, in line with similar information contained in the National Accounts. This means that pension entitlements (or obligations) represent the amounts due for service to date. They do not include rights which will be built up in the future. As the European Central Bank has pointed out (ECB 2010), in order to assess the fiscal sustainability of unfunded pension schemes, “the concept of pension entitlements needs to be extended to include entitlements that will be accrued in future, while at the same time comparing these ‘claims’ with future social contributions and tax payments”.
This article is divided into eight sections. The remainder of the article is organised as follows. Section 2 presents the 2010 experimental supplementary table on pensions and provides an overview of the estimates. Section 3 looks at the estimates of the total obligations, or gross liabilities, of UK pension providers and various breakdowns of the total. Section 4 focuses on unfunded pensions and discusses the sensitivity of the estimates to the assumptions in the actuarial models which are used to produce them. Section 5 considers changes over time: what can be learnt from the flows which take place during the year, for instance in relation to the impact of policy changes. Section 6 looks at the methodology used to estimate private sector pension obligations, because this was not covered by the previous two articles in this series (Levy 2011a and Levy 2011b). Section 7 looks at where further work is required in order to complete the development of the table and change its status from experimental statistics to National Statistics. Section 8 concludes.
It should also be noted that contingent pension obligations are not the same as ‘contingent liabilities’ as defined by HM Treasury for the purposes of the Whole of Government Accounts (HM Treasury 2011) and by the Office for Budget Responsibility (OBR) in its fiscal sustainability reports (for example OBR 2011). These reports define ‘contingent liabilities’ as liabilities which are associated with events which may arise in future but are improbable (less than 50 per cent probability); some of these events are also unquantifiable.
The National Accounts supplementary table on pensions is designed to provide a complete picture of pensions, with the exception of individual personal pensions. In the UK, the government is responsible for part of total pension provision – (i) the state pension and (ii) private pensions for public sector employees. The private sector (employers and insurance companies) is also a major provider of private pensions. The supplementary table covers state pensions (see Annex 1) and all workplace private pensions, which are those shown in the first three columns of Figure 1.
The supplementary table on pensions (Table 2) brings together information shown in the standard or ‘core’ National Accounts (Columns A to F of the table) and information on unfunded pensions, which – under the SNA2008 and ESA2010 – only appears in the supplementary table (Columns G and H). The columns of the supplementary table also break down the information according to:
Who is the ‘pension manager’: government or the private sector (see Box 1).
The type of pension scheme: defined benefit, hybrid and defined contribution (see Box 2).
Where the scheme appears in the National Accounts classification.
The supplementary table requires estimates of entitlements, or obligations, to be calculated on an actuarial basis for DB (including other ‘non-defined contribution’ or hybrid) pension schemes. In the UK these are the schemes in Columns B, E, G and H. Column F is not applicable to the UK because all funded pension schemes for which government is responsible are classified in the Financial Corporations sector. For DB schemes where a private sector organisation is the pension manager (Column B), variable, market-based discount rates are used in the actuarial estimates. However, for DB schemes where government is the pension manager (Columns E, G and H) the methodology incorporates the EU-level requirement to use a stable discount rate of 3 per cent (real) or 5 per cent (nominal) (see Section 1).
In the case of pure DC schemes (Column A), liabilities are equal to assets at market value. Column D is not applicable in the UK because in the public sector all DC pension provision is in the form of personal pensions with (private sector) insurance companies.
It should be noted that in the 2010 table (Table 2), Column J (pension entitlements of nonresident households) and Row 11 (output) have not been completed. ONS is still working on these parts of the table (see Section 7).
The remainder of this section provides an overview of the estimates of pension obligations/entitlements in the 2010 table, focusing first on pension schemes for which government is responsible and then on workplace pension schemes provided or facilitated by private sector employers.
The supplementary table on pensions (Table 2) shows the obligations of all unfunded DB schemes which are the responsibility of government. These can be broken down into:
Column G: centrally-administered unfunded pension schemes for public sector employees, of which the main schemes are those for: civil servants, teachers, National Health Service employees, members of the Armed Forces, police officers, firefighters, the judiciary, members of the security services (MI5 and MI6), UK Atomic Energy Authority employees, DFID overseas employees and people working for the Research Councils. The value of liabilities for Column G at the end of 2010 was estimated at £852 billion; this estimate is based on information published in the annual resource accounts of the schemes (see Levy 2011a) except in the case of the police and fire schemes, where the figures are estimated, as no accounts are currently available (see Section 7).
Column H: the state pension system, which covers most people in the UK and some pensioners resident overseas. The estimated value of total liabilities for Column H at the end of 2010 was £3,843 billion (or £3.8 trillion); further details are presented in Annex 1.
The supplementary table on pensions also shows the obligations of funded DB pension schemes for which government is responsible (Column E). The largest scheme in Column E is the Local Government Pension Scheme for local authority employees. This column also includes pension schemes for BBC and Transport for London employees and schemes with a Crown guarantee such as those for members of formerly nationalised industries (railways, coal and telecommunications).
The estimated value of total liabilities for Column E at the end of 2010 was £313 billion. It should be noted that the estimates for Column E are less robust at present than those for the other columns in the supplementary table because many schemes in this column do not publish information on a regular basis (see Section 7).
The value of total liabilities of workplace pension schemes provided or facilitated by private sector employers was estimated at £1,726 billion at the end of 2010. This figure comprised £386 billion in DC pension obligations and £1,340 billion DB pension obligations. Thus, despite the widely reported shift in active pension scheme membership from DB to DC pensions1, the bulk of the outstanding obligations of private sector employer-sponsored pension schemes at the end of 2010 remained in the form of DB pensions.
Readers should note that the value of entitlements/liabilities shown in the supplementary table for the private sector is for workplace pensions only and does not include individual personal pensions (see Figure 1 above). To produce estimates for workplace pensions only involves making a number of assumptions, as the source data does not always split personal pensions into workplace and individual pensions (see Section 6). The next section provides a complete estimate of UK pension liabilities including individual personal pensions.
ONS (2011a) presents statistics on the shift in workplace pension scheme membership from DB to DC.
The concept of ‘pension manager’ refers to the organisation which is ultimately responsible for the pension obligations. For schemes in Columns D-H, this is the government. For schemes in Columns A-C, this is a private sector employer or an insurance company (also private sector).
This article refers to Columns A-C as ‘private sector’. However, readers should note that if National Accounts terminology were to be used, Columns A-C would be described as ‘non-general government’ rather than ‘private sector’. This is the heading used in the supplementary table.
National accountants divide the economy into different sectors - the principal ones being General Government, Financial Corporations, Non-Financial Corporations and Households. Although most financial and non-financial corporations are in the private sector, there are a number of government controlled or ‘public’ corporations. The Royal Mail is an example of a public non-financial corporation and the Royal Bank of Scotland is an example of a public financial corporation.
Defined benefit (DB) pension schemes are those in which the rules specify the rate of benefits to be paid. The most common DB scheme is one in which the benefits are based on the number of years of pensionable service, the accrual rate and final salary. However, Career Average Revalued Earnings (CARE) schemes are becoming increasingly common in the UK. These base the pension on earnings over the whole career, adjusted by prices or earnings.
Defined contribution (DC) pension schemes are those in which the benefits are determined by the contributions paid into the scheme, the investment return on those contributions (less charges), and any annuity purchased on retirement. They are also known as money purchase schemes.
Hybrid pension schemes are those offering a choice, or mixture, of DB and DC rights at retirement. In the supplementary table on pensions, hybrid schemes are referred to as ‘other non-DC schemes’ and are included with DB schemes.
This section presents information on total obligations, or gross liabilities, of all UK pension providers for all types of pension. It goes beyond the figures presented in the National Accounts supplementary table on pensions because it includes data for individual personal pensions. This provides a complete picture of UK pension liabilities, which are also the total pension entitlements of UK households.
Table 3 shows that total obligations of all UK pension providers at the end of 2010 were estimated at £7,060 billion (or £7.1 trillion), nearly five times the UK’s Gross Domestic Product (GDP) and over seven times its household Gross Disposable Income (GDI). This total comprised £5,009 billion of government obligations and £2,052 billion of private sector obligations outstanding at end-2010.
|Suppelmentary table column||Private sector||Sub-total private sector||Government||Sub-total Government||TOTAL|
|DC workplace funded||DB workplace funded||DB workplace funded||DB workplace unfunded||State pensions|
|% of GDP||22||26||92||140||21||58||263||342||484|
|% of household GDI||33||39||137||210||32||87||392||511||721|
Although private sector liabilities are included in Table 3, caution should be used in comparing private sector and government liabilities because the estimates are calculated differently: private sector DC liabilities are based on (variable) market values of assets and private sector DB liabilities are estimated using (variable) market discount rates, while estimates of the government’s liabilities are based on a stable discount rate (see Section 2). It should also be noted that while the UK government’s pension liabilities can be compared with those of other countries, direct comparisons between UK private sector pension liabilities and those of other countries are inadvisable because the DB/DC composition of private pension systems varies from country to country and the discount rates used also vary between countries.
Total obligations for which the government was responsible were estimated at £5,009 billion at the end of 2010, equivalent to 342 per cent of GDP and 511 per cent of GDI. Over three-quarters of this total was in respect of Column H state pension obligations (see Figure 4), which were estimated at 263 per cent of GDP and 392 per cent of household GDI at end-2010 (Table 3). Unfunded workplace pensions for public sector employees, corresponding to Column G of the table, accounted for 17 per cent of the total government liability (Figure 4) and were worth 58 per cent of GDP and 87 per cent of household GDI (Table 3).
The latest comparable EU-level estimates are for end-2007 (ECB, 2010)1. These put Column H liabilities at 278 per cent of GDP (415 per cent of household GDI) and Column G liabilities at 52 per cent of GDP (79 per cent of household GDI). These EU-level averages are slightly above the UK figures for Column H and below the UK figures for Column G. However, readers should note that the EU-level estimates will have changed since end-2007, both because of changes in outstanding pension liabilities and because of the performance of GDP and household GDI, which will have been affected by recession in much of the EU.
This section looks at the total value of obligations/entitlements in private pensions in the UK. Private (or non-state) pensions are provided by employers in both the private and public sectors and by insurance companies. They may be provided or facilitated via the workplace, in which case they appear in the supplementary table on pensions, or taken out as individual pensions, which are not included in the table.
The total value of private pension obligations at the end of 2010 was estimated at £3,217 billion (220 per cent of GDP), of which £1,165 billion (80 per cent of GDP) were public sector employee entitlements and £2,052 billon (140 per cent of GDP) corresponded to private sector entitlements (some of which may be personal pensions for public sector employees provided by insurance companies, which are not separately identifiable). Figure 5 shows that:
Private sector DB pension entitlements were worth 92 per cent of GDP at end-2010, compared with 49 per cent of GDP for private sector DC pensions.
Public sector funded and unfunded DB pension entitlements for public sector employees were worth 21 per cent and 58 per cent of GDP respectively at end-2010.
Although the previous section shows that DB type pensions are still dominant in terms of entitlements accrued to date in 2010, active pension scheme membership in the private sector is currently moving away from DB provision as such schemes close to new members and DC options are offered instead. DC pensions are expected to become the dominant type of pension provision in the private sector in coming years. Therefore it is of interest to look in more detail at DC pension obligations.
Figure 6 shows that most DC pension entitlements in the UK relate to pensions provided by insurance companies; these amounted to 40 per cent of GDP at end-2010. The split between individual and workplace pension obligations of insurance companies shown in Figure 6 (22 and 17 per cent of GDP respectively) is estimated (see Section 6).
Finally, this section looks at how the outstanding liabilities of funded pension schemes compared with their assets at end-2010. In Table 3, funded pensions comprise all private sector pensions and also DB workplace funded pensions which are the responsibility of the government. The total outstanding value of funded pension liabilities at the end of 2010 was estimated at £2,365 billion (162 per cent of GDP). This compares with £1,868 billion of assets (128 per cent of GDP) in funded pensions at end-20102.
The end-2007 estimates published by the European Central Bank (ECB) were based on calculations by the ECB and the Research Center for Generational Contracts at Freiburg University (Germany) using a sample of government-managed pension schemes in the euro area.
As part of work on the supplementary table in 2011, ONS developed a method for separating out pension assets from life assurance. This enabled us to publish the first official estimate of the value of ‘money in funded pensions’ (for 2009) in Pension Trends (ONS 2011b). This article presents an estimate for 2010.
It is important, when working with actuarial estimates of liabilities for DB pension schemes, to be aware that small changes in assumptions can lead to large changes in outcomes. This is because estimates are produced by projecting future flows over a 100-year time horizon and discounting them to generate a ‘present value’ at a particular date. This section looks at the sensitivity of the estimates of the unfunded pension liabilities in Columns G and H to changes in key modelling assumptions.
The first article in this series presented detailed sensitivity analysis for the liabilities of the largest four public sector employee pension schemes in Column G – the schemes for the National Health Service, teachers, civil service and Armed Forces. The calculations were made by the Government Actuary’s Department for ONS in respect of a central estimate of the liabilities of these schemes as at 31 March 2008, calculated using the same assumptions as the 2010 supplementary table on pensions. This analysis showed that the estimate of pension liabilities (or entitlements) was most sensitive to changes in the discount rate, which is also the most variable financial assumption. A 1 per cent fall in the discount rate increased the liabilities by 16 per cent, while a 1 per cent increase in the discount rate reduced the liabilities by 14 per cent. The article also explored the impact of changes in other financial assumptions, such as nominal wage growth and pensions indexation, as well as the impact of changing demographic assumptions using the ‘low’ and ‘high’ life expectancy scenarios of ONS’s National Population Projections (see Levy 2011a).
As part of the development of the state pensions column of the supplementary table (Column H), which is described in detail in Levy (2011b), sensitivity analysis was carried out for the end-2010 estimate of liabilities. The full details are presented in Annex 1 of this article. This section summarises the findings.
The analysis looked at what happened to the central estimate of state pension liabilities at end-2010 (£3,843 billion), produced by the Forecasting Division of the Department for Work and Pensions (DWP), when the assumptions used to produce the estimate were changed. Each assumption in turn was varied while holding the other assumptions constant.
Table 7 shows the impact of varying the discount rate away from the 5 per cent (nominal) level used to create the estimate for the supplementary table on pensions. Increasing the discount rate to 6 per cent leads to a 21 per cent decrease in total state pension liabilities/entitlements (by £815 billion). Reducing the discount rate has an even greater impact on the estimates. Reducing the discount rate to 4 per cent leads to a 31 per cent increase in total state pension liabilities/entitlements (by £1,174 billion).
|Nominal discount rate (%)||4.00||4.25||4.50||4.75||5.00||5.25||5.50||5.75||6.00|
|Change in pension entitlements (£ billion)||1,174||837||532||253||0||-231||-443||-637||-815|
|Change in pension entitlements (%)||31%||22%||14%||7%||0||-6%||-12%||-17%||-21%|
This shows that changing the discount rate used in the model produces important changes in the results. The analysis also explored the impact of varying the uprating (or indexation) assumptions, earnings growth assumptions and life expectancy assumptions. It should be noted that the state pension model has two main components: Basic State Pension (BSP) and additional state pension (AP). BSP liabilities are much larger than AP liabilities – the central estimate of liabilities at end-2010 was £3,238 billion for BSP and £542 billion for AP – and because the system for building up and paying out BSP entitlements is different from that for AP, the assumptions may be specific to each component. For instance, although life expectancy assumptions are relevant in both cases, for BSP uprating is based on the ‘triple lock’ policy1, while for the AP it is in line with the Consumer Prices Index (CPI) measure of inflation, assumed to be 2 per cent in the long term. Earnings growth assumptions are used in the AP model to revalue entitlements up to State Pension Age, but not in the BSP model (Levy 2011b).
The results of the sensitivity analysis for the uprating, earnings growth (AP only) and life expectancy assumptions showed that:
For BSP, increasing uprating by 1 per cent leads to a 31 per cent increase in liabilities, while reducing it by 1 per cent leads to a 22 per cent reduction in liabilities.
For AP, increasing uprating by 1 per cent leads to a 16 per cent increase in liabilities, while reducing it by 1 per cent leads to a 13 per cent reduction in liabilities.
For AP, a change of +1 per cent in the earnings growth assumption leads to a 13 per cent increase in liabilities, while a change of -1 per cent leads to a 10 per cent decrease in liabilities.
The effect of using mortality rates based on the ‘high life expectancy’ and ‘low life expectancy variants’ of ONS’s National Population Projections2 produces changes in pension liabilities of +/-9 per cent for BSP and +/-6 per cent for AP.
Full details of the sensitivity analysis for state pensions as at end-2010 are in Annex 1.
The triple lock policy for BSP uprating, announced in June 2010, states that BSP will increase each year by average earnings growth, inflation (using the CPI) or 2.5 per cent, whichever is higher.
The DWP model used to produce BSP and AP estimates for the supplementary table used ONS’s 2008-based principal National Population Projection. ONS also produces variants around the principal (or central) projection, including one based on the assumption that life expectancy increases more rapidly than expected in future (the ‘high life expectancy’ variant) and another based on the assumption that life expectancy increases less rapidly than expected (the ‘low life expectancy’ variant).
Many of the estimates of unfunded pension liabilities that have been published in the UK in recent years have been difficult to interpret because they do not show why the figures have changed from one year to the next. Most previously published estimates use variable (market-based) discount rates, so changes in the estimates due to discount rate changes tend to obscure the impact of other changes. The supplementary table on pensions overcomes this problem in two ways:
The change between the opening and closing balance of entitlements/liabilities each year (Rows 1 and 10) is broken down into its component parts, which are shown separately in Rows 2 to 9 (see below).
For schemes where the government is the pension manager, including all unfunded pensions, the table uses a stable discount rate which does not change from year to year (see Section 2). This means that other changes are not obscured by changes in discount rates. Even if the discount rate is revised (which may occur from time to time), the impact should be shown separately from the impact of other changes.
The remainder of this section shows how the information in Rows 2 to 9 of the supplementary table helps to explain changes in entitlements/liabilities over time. Section 5.1 provides an overview of the information in these rows. Section 5.2 focuses on Row 7, which is particularly useful for understanding the impact of changes brought about through legislation, such as increases in State Pension Age, and through official announcements,such as the change from the Retail Prices Index (RPI) to the CPI for indexation of public sector employee and state pensions.
The distinction between DB and DC pension schemes is important for the information presented in the supplementary table.
In the case of DB schemes, there is information for the present value’ of entitlements accrued up to the start and end of each year – the opening and closing balances in Rows 1 and 10 respectively, calculated on an actuarial basis. There is also information for most of the rows in between (Rows 2 to 9). These include actual cash flows, such as the value of actual contributions (Rows 2.1 and 2.3) and pension benefits (Row 4); the impact of policy changes (Row 7); and the value of changes in the actuarial assumptions used to calculate DB pension entitlements (Rows 8 and 9). Specifically:
Rows 2.1 and 2.3 contain actual contributions paid by employers and employees.
Row 2.4 is the ‘unwinding of the discount rate’, or re-valuation of liabilities between the start and end of each year, calculated by multiplying Row 1 by the nominal discount rate.
Row 4 shows the payment of pension benefits during the year.
Row 6 shows transfers of pension entitlements between schemes (if any).
Row 7 contains an estimate of changes in ‘past service cost’ (i.e. benefits accrued in previous years) due to legislation enacted during the year in question and official announcements made during the year.
Row 8 shows any changes in entitlements due to changes in financial assumptions including discount rates, wage growth, indexation of benefits and inflation.
Row 9 shows any changes due to other assumptions, mainly changes in demographic assumptions such as life expectancy; for schemes which use ONS population projections, the figures in this row are likely to change every two years when new projections are published.
Row 2.2 in the case of employee pension schemes and Row 3 in the case of state pensions are calculated as balancing items1.
For employee pension schemes, Row 2.2 captures ‘experience effects’, often referred to as ‘experience gains and losses’ (differences in the outcome because some actual changes turn out to be different from those assumed in the model). It also reflects any differences between increases in liabilities during the year as a result of employee service (‘current service cost’) and actual contributions received during the year. If the values in Row 2.2 are high over several years, it means that the assumptions may need to be reviewed and contribution rates adjusted.
For state pensions, Row 3 captures the increase in entitlements for people who do not make National Insurance Contributions (NICs) through work (for instance people of working age who are unemployed, looking after children or caring for elderly relatives); these people are credited with qualifying years which allow them to build up state pension entitlements. It also captures, for working people, the difference between the actual value of NICs made by them or their employers and what they will receive in state pension payments as a result of these contributions. In addition, it contains a small ‘residual’ (differences in the outcome because some actual changes turn out to be different from those assumed in the model).
In the case of DC pensions, there is also information for Rows 1 and 10 but many of the rows in between are not applicable or not populated because liabilities are not calculated using actuarial assumptions – they are simply equal to the value of the assets. Thus Rows 2.2 and 3 are not applicable and there are no values in Row 7 or Row 9. Also, Row 2.4 is not calculated using a discount rate; it simply records ‘property income’ (interest and dividends on pension assets and any sales of self-administered funds’ assets).
Thus, to compile Column A for private sector DC schemes:
Actual information is entered for Rows 1 and 10 (pension assets).
Actual information is entered for Rows 2.1 and 2.3 (pension contributions), Row 2.4 (property income), Row 4 (pension benefits) and Row 6 (transfers).
Row 8 is the residual row (balancing item); it shows the increase/decrease in the capital value of the assets during the year2.
In the context of concerns at EU level about the ageing population and its implications for public finances, an explicit objective of the supplementary table on pensions is that it should be able to track the impact of pension reforms on governments’ pension obligations. According to ECB (2010):
“If recorded systematically over a long time horizon, data on accrued-to-date pension obligations may contribute usefully to gauging, inter alia, to what extent the size of pension obligations changes in response to reforms of government pension schemes, e.g. increases in the statutory retirement age”.
The supplementary table on pensions shows each policy change in the year that the legislation is passed (or announced if legislation is not required). For instance, in 2010 the government announced that the indexation of pensions for public sector employees and state pensions would in future be based on the CPI rather than the RPI.
Therefore the indexation assumptions used in the opening balance for Columns E and G are based on the RPI, while the closing balance is based on the CPI assumption. It is possible to detect the impact of the move from the RPI to the CPI in Row 7: the change reduced the value of Column E liabilities by an estimated £32 billion and the value of Column G liabilities by an estimated £95 billion. Thus the total saving to government in respect of public sector employee pensions was £127 billion (10 per cent of total liabilities at the start of 2010).
For state pensions (Column H), Row 7 showed an increase of £33 billion in 2010, equivalent to 1 per cent of the value of total state pension entitlements at the start of the year. The value of AP entitlements fell by £124 billion due to the change from the RPI to the CPI for uprating AP payments during retirement (see Annex 1). However, this was outweighed by a £162 billion increase in the value of BSP entitlements as a result of the change to the ‘triple lock’ uprating policy, announced in June 2010, which replaced uprating by earnings under the Pensions Act 20073.
It should be noted that many of the changes to state pension obligations due to reforms in recent years are captured in the opening and closing balances of the 2010 supplementary table but are not shown separately in the 2010 table because they date from before 2010. For instance, although increases in women’s State Pension Age began in April 2010, they became law in 1995. Therefore they are not shown separately in the 2010 table. They would have been shown in Row 7 of a 1995 supplementary table, had such a table existed that year.
Similarly, the set of state pension reforms contained in the Pensions Act 20074 would have been recorded in Row 7 of a 2007 supplementary table, had it existed. As part of the work on Column H of the 2010 supplementary table, changes which took place as a result of the Pensions Act 2007 were examined by DWP analysts. They estimated that the measures introduced by the Act increased state pension entitlements/liabilities by £540 billion, equivalent to 15 per cent of the opening balance in the 2010 supplementary table5.
The 2011 supplementary table on pensions will show the impact of the State Pension Age changes in the Pensions Act 2011 (in Row 7 of Column H) and the 2012 table will pick up the move of Royal Mail pension liabilities from the private sector (Column B) into an unfunded employee pension scheme for which government is responsible (Column G). Although it is difficult to predict when other policy changes will be recorded until they have been approved by Parliament, supplementary tables in coming years are likely to show (in Row 7) changes due to the measures announced in the March 2012 Budget to replace the BSP and AP with a single-tier, flat-rate state pension and to increase State Pension Age more rapidly than under current legislation to take into account increases in longevity.
These rows will be affected by a methodology change planned for future supplementary tables (see Section 7).
This row will be affected by a methodology change planned for future supplementary tables (see Section 7).
For details of the uprating assumptions used in the 2010 opening and closing balances of Column H, see Levy (2011b).
For details of the Pensions Act 2007, see Annex 1 and ONS (2011c).
Although the Pensions Act 2007 had the effect of increasing state pension liabilities, it also reduced expected future payments of Pension Credit. The changes introduced by the Act were designed to move people off Pension Credit and onto contributory state pensions. Pension Credit is an income-related benefit rather than a pension, so it is not included in the supplementary table (see Section 1).
The previous two articles in this series (Levy 2011a and Levy 2011b) looked at the methodology for producing figures for pension schemes where government is the pension manager. This section looks at the methodology used to estimate private sector pension obligations. A summary of data sources and methods for all columns of the supplementary table is provided in Annex 2.
Annex 2 explains in detail the methods used to produce figures for Columns A and B. As outlined in Section 5, the conceptual difference between these two columns is that while Column B estimates are calculated using actuarial models – so that entitlements/liabilities are on an actuarial basis and the flows during the year reflect policy changes and changes in actuarial assumptions as well as actual flows and the unwinding of the discount rate – the Column A estimates (both stocks and flows) are actual values recorded in data sources, except in the case of Row 8 (the residual).
In theory, this should mean that Column A is simpler to compile than Column B. However, this is not in fact the case. One reason for this is that until the turn of this century most active members of workplace pension schemes were in DB schemes; even in 2010 two-thirds of employee members of workplace schemes were in DB schemes, according to the Annual Survey of Hours and Earnings1. DC pensions have been in existence for less time than DB pensions. Therefore official data collection for workplace pension schemes is more developed for DB schemes than for DC schemes.
DC pensions take two main forms: occupational schemes and personal pensions, which include stakeholder pensions. All occupational DC schemes are workplace schemes, as their name suggests, but personal pensions may be organised through employers (which means that they are within the scope of the supplementary table) or taken out individually (outside the scope of the table). In both cases, personal pensions are provided by insurance companies.
Another complication is that although funds associated with DC occupational schemes may be self-administered or insured in the accumulation phase, when people are building up pensions, on retirement members are likely to buy an annuity from an insurance company. The reserves held by insurers to pay such annuities in the ‘decumulation phase’ cannot be separately identified from those held to pay annuities for individual or workplace personal pensions. The Financial Services Authority (FSA) collects information on pension reserves of insurance companies, which is used for compiling Rows 1 and 10 of Column A. This can be divided into individual and workplace pensions for the accumulation phase. However, in the decumulation phase, when pensions are paid out as annuities, there is no such breakdown available. Therefore the breakdown has to be estimated (see Annex 2).
While FSA data is used for Rows 1 and 10 of Column A, much of the data for the intervening rows comes from ONS’s surveys of self-administered pension funds and insurance companies, known as the MQ5 surveys. When developing the 2010 supplementary table on pensions, ONS found that there was a lack of information for some of the breakdowns required. The survey of self-administered pension funds asked questions about what proportion of contributions related to DB and DC pensions, but such questions were not asked for other items such as payments of benefits, transfers out and ‘property income’. The surveys of insurance companies collected some data at the level of insurance and pensions business combined (which meant that pensions could not be separately identified), while other data was collected for pensions business separately but without a split between individual and workplace pensions. The required splits have now been introduced into the ONS MQ5 surveys and this information will be used in future supplementary tables.
For the 2010 supplementary table, a number of assumptions had to be made to estimate the missing breakdowns for Column A. These are documented in Annex 2. In particular, users of the table should note that, because the division between workplace and individual pensions is estimated, the scope of Column A of the table could be overestimated or underestimated with respect to the part of the estimates that remain outside the table (pensions taken out on an individual basis).
The compilation of Column B is more straightforward. There are two data sources which between them provide all the information required: the Purple Book (The Pensions Regulator, Pension Protection Fund 2012) for actuarial estimates and ONS’s MQ5 survey of self-administered pension funds for income and expenditure data.
For the 2010 supplementary table, Rows 1 and 10 of Column B were compiled using estimates of pension liabilities on a full buy-out (risk-free discount rate) basis from the 2011 Purple Book. As Purple Book liability figures are published for the end of each financial year, these were adjusted by ONS to end-calendar year figures (see Annex 2).
The 2011 Purple Book used a dataset of 6,432 Pension Protection Fund (PPF)-eligible DB pension schemes – most of those in the ‘PPF universe’ at end-March 2011. In order to avoid double counting with Column E, any pension liabilities in the PPF universe for which the ‘pension manager’ is the government were removed from the Column B estimates. Box 3 explains how ONS decides which schemes should be in Column B and which schemes should be in Column E.
To compile Rows 2 to 9 of Column B, the main sources used are the MQ5 survey of self-administered pension funds and the Purple Book. The self-administered pension funds survey provides information on contributions received, payment of benefits and transfers between schemes. The proportion attributable to DB schemes has to be estimated for payment of benefits and transfers out, although this should not be necessary in future as the required splits have now been introduced into the survey (see above). The rows which are based on the actuarial model (Rows 2.4 and 7, 8 and 9) are estimated by ONS on the basis of advice received from the PPF, including advice on the appropriate nominal discount rate to use for Row 2.4 (the unwinding of the discount rate, or re-valuation of liabilities) and the impact of any change in the risk-free discount rate, which is recorded in Row 8.
The ASHE figures are presented in ONS (2011d).
In the core National Accounts all funded pension schemes are classified in the Financial Corporations sector, and there is no distinction according to whether DB schemes are the responsibility of government or private employers. However, in the supplementary table on pensions, schemes which are the responsibility of private sector organisations are to be shown in Column B, while those which are the responsibility of government are to be shown in Column E (see Box 1). The following classifications decisions have been taken for the 2010 supplementary table on pensions:
Column E contains those schemes which appear in the Financial Corporations sector in the core accounts where either the parent organisation is classified in the General Government sector or there is an explicit Crown guarantee.
Column B contains mainly private sector schemes, but also some schemes that are associated with ‘public’ corporations (see Box 1) where the employees or former employees are not government employees and the government has no legal obligation for the liabilities. In 2010 this was the case of the Royal Mail Pension Plan (RMPP). It is also the case for the nationalised banks, except where government has accepted responsibility for the pension scheme liabilities.
The 2012 Budget announced that RMPP liabilities accrued up to 31 March 2012 would be transferred to a new unfunded public service pension scheme (the Royal Mail Statutory Pension Scheme) from 1 April 2012. This will result in a transfer of RMPP pension liabilities from Column B to Column G (in the 2012 supplementary table).
The 2010 supplementary table on pensions presented in this article is designated as ‘experimental statistics’, which are defined in the Code of Practice for Official Statistics as:
"new official statistics undergoing evaluation. They are published in order to involve users and stakeholders in their development and as a means to build in quality at an early stage".
ONS expects that when the development of the table is complete – in time for the 2012 supplementary table – it will be designated as National Statistics in line with the rest of the UK National Accounts. This section provides a brief description of the work that still needs to be done before the development of the table is complete. This falls under two headings: (i) methodology issues and (ii) improved data sources.
The most important methodology change is the inclusion of a new row in the supplementary table, following a revised layout proposed by Eurostat, the EU statistical agency. At present, pension entitlements/liabilities are recorded gross without deductions for taxation or service charges. In technical guidance published in January 2012, Eurostat proposed to include a Row 2.5 entitled ‘Less: pension scheme service charges’ (Eurostat, European Central Bank 2011). This means that in future the table’s closing balance each year would show pensions gross of taxation but net of service charges. This would have an impact on the values shown in the 'residual'rows (Row 2.2 in the case of DB employee pension schemes, Row 3 in the case of state pensions and Row 8 for DC employee pensions). In the case of DB pension schemes, it would have the advantage of making Row 2.2 (employer imputed social contributions) easier to interpret than at present, because large negative figures which can be seen in the 2010 table would not appear in future tables1.
Other methodology areas that ONS plans to work on are:
Population of Column J (pension entitlements of non-resident households), if such entitlements are found to be significant.
Population of Row 11 (output), which will equal Row 2.5 (service charges).
Incorporating data on the pension liabilities of the PPF and the Financial Assistance Scheme, if it is decided that these should be classified in the National Accounts as pensions rather than insurance.
In terms of improvement of data sources:
The estimates shown in Columns A and B should benefit from the new data that is now being collected by the MQ5 surveys, which should remove much of the need for estimation using ratios (see Section 6).
Estimates in Column E could be improved by regular publication of scheme valuations for all parts of the Local Government Pension Scheme (LGPS), including the Scottish LGPS for which no figures are currently available. It is also hoped that the smaller funded employee pension schemes for which government is responsible will publish regular valuations or resource accounts in future.
Similarly, it is to be hoped that valuations for the police and fire schemes, which are needed to improve the Column G estimates, will be published soon – and that regular valuations will be published for these schemes in future.
Finally, the accounts of some schemes in Columns E and G do not show separately the impact of changes in demographic assumptions, but instead include them with changes in financial assumptions. The estimates for Rows 8 and 9 could be improved if all scheme valuations and resource accounts showed these changes separately.
It is expected that the figures in Row 2.2 should generally be positive after Row 2.5 is subtracted, although there might be very small negative figures remaining in some cases due to experience effects (see Section 5.1).
At present, estimates of obligations of pension schemes which are the responsibility of the UK government are not available on consistent, complete and regular bases from any official source. The National Accounts supplementary table on pensions fills this gap by providing a full picture of the government’s funded and unfunded pension obligations.
Estimates of government pension obligations published elsewhere change from year to year reflecting discount rate fluctuations. By contrast, the supplementary table on pensions takes a long-term view: the use of the stable discount means that year-to-year changes shown in the table reflect changes in ‘fundamentals’ rather than changes in discounting.
The supplementary table on pensions also tracks the impact of policy changes on government pension obligations. The 2010 table shows the impact of the changes in uprating policy announced in June 2010, while the 2011 table will reflect the changes to State Pension Age contained in the Pensions Act 2011. Future tables will show the changes to pension arrangements for Royal Mail employees (in 2012) and, subject to legislation, the replacement of the two-tier state pension system with a single-tier, flat-rate state pension.
The supplementary table on pensions and the accompanying analysis of individual personal pension entitlements presented in this article shows how important private sector pensions are in the UK pensions landscape. Over time, the table will show not only the scale of the accumulated entitlements but the impact of the shift from defined benefit to defined contribution pensions and the increasing role of insurance companies in pension provision.
While the accrued-to-date estimates presented in the supplementary table are a valuable contribution to official pension statistics, readers should bear in mind that they have some limitations. In particular, they are not an indicator of the financial sustainability of pension systems and they exclude income-related benefits and other forms of social assistance.
A list of the job titles of those given pre-publication access to the contents of this release is available on the website (23.4 Kb Pdf) .
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: firstname.lastname@example.org
European Central Bank (2010): ‘Entitlements of Households under Government Pension Schemes in the Euro Area – Results on the Basis of the New System of National Accounts’, European Central Bank Monthly Bulletin, January 2010.
European Union (2012, forthcoming): 2012 Ageing Report: Economic and budgetary projections for the EU Member States (2010-2060). Due to be published in May 2012.
Eurostat, European Central Bank (2011): Technical Compilation Guide for Pension Data in National Accounts. Eurostat Methodologies & Working papers.
HM Treasury (2011): Whole of Government Accounts: Unaudited Summary Report for the year ended 31 March 2010, published July 2011.
Hobbs, D. (2010a), ‘Wider measures of public sector debt: A broader approach to the public sector balance sheet’, Office for National Statistics.
Hobbs, D. (2010b), ‘Wider measures of public sector debt: an update’, Office for National Statistics.
Levy, S. (2011a): ‘Pensions in the National Accounts: Compiling a complete picture of UK pensions including unfunded pensions for public sector employees’ Office for National Statistics (ONS), 3 August 2011.
Levy, S. (2011b): ‘Pensions in the National Accounts: Compiling estimates of state pension obligations for the National Accounts (a methodology article)’, Office for National Statistics (ONS), 2 December 2011.
Office for Budget Responsibility (2011): Fiscal sustainability report, July 2011.
Office for National Statistics (2011a): Pension Trends Chapter 6: Private Pensions, 2011 edition.
Office for National Statistics (2011b): Pension Trends Chapter 9: Pension scheme funding and investment, 2011 edition.
Office for National Statistics (2011c): Pension Trends Chapter 5: State pensions, 2011 edition.
Office for National Statistics (2011d): Pension Trends Chapter 7: Pension scheme membership, 2011 edition.
Pensions Commission (2004) Pensions: Challenges and Choices, The First Report of the Pensions Commission. The Stationery Office: London.
The Pensions Regulator, Pension Protection Fund (2012): Purple Book: DB Pensions Universe Risk Profile 2011.