1 Discussion note on insurance companies' recording of pensions transactions Introduction This note sets out a number of issues about the way insurance companies record pensions transactions, on which ONS would welcome views. They have arisen as part of the review of the MQ5 surveys of insurance companies and self-administered pension funds. Some of the issues relate to how the questionnaires should be completed to complement the other sources of information used to compile pension contribution statistics and also how the questionnaires were in practice being completed by respondents. This note concentrates on these issues although those relevant to the National Accounts are also discussed. The investigations to date have led to improvements to the detailed questions and the notes in the quarterly and annual surveys of insurance companies and pension funds and are being introduced from Q These improvements will lead to better estimates of pension contributions later this year. Clarification of issues set out in this note will also lead to improvements in pensions statistics in due course. Importance of these issues for National accounts treatment of pension contributions In the national accounts, employers' contributions to pension schemes form part of compensation of employees and contribute to Gross Value Added (GDP). Pension contributions and payments are also used in estimates in the sector accounts. The attached note describes the treatment of non-state pensions in the national accounts (please follow link to access 'Treatment of pensions in the national accounts'). National accounts use different sources for occupational and personal pensions in their estimates of pension contributions. Estimates for contributions to occupational pensions are taken from the MQ5 surveys of insurance companies and selfadministered pension funds. Inland Revenue data, rather than MQ5 survey data, are used for personal pension contributions. Thus if respondents to MQ5 report personal pension contributions as occupational pension contributions, double counting will occur in national accounts estimates. As part of the pension statistics review, a table was published in Annex B showing reconciliation of pension contributions data reported in the MQ5 surveys with Inland Revenue data for pension contributions for tax relief. This Annex B table has been updated subsequently (link to Annex B table).
2 Potential double counting areas 1. Group Personal Pensions Group personal pensions (GPPs) are personal pension plans that an employer has organised for a group of employees, usually with an insurance company or other financial institution. As part of the testing of new questionnaires we asked respondents where they reported GPPs. Discussion with the Association of British Insurers (ABI) had led us to believe that these would be reported as personal pensions. This treatment was also confirmed with the Inland Revenue who also include GPPs as personal pensions in their data. However, in our investigations, most of the respondents said they reported them under group sponsored (ie occupational) pensions. In reconciling pension contributions data, national accountants assume GPPs are part of personal pensions. Using the table in Annex B of the pensions review report, personal pension contributions (from MQ5) are subtracted in line G and Inland Revenue data for personal pension contributions are added back in line M. To the extent that GPPs are being reported as occupational pensions in MQ5, then total pension contributions and contributions to funded occupational schemes will be overstated. From 2004, we are explicitly asking respondents to include GPPs under personal pensions on the new quarterly questionnaires and we have asked for data on GPPs separately on the annual questionnaire for This change on the MQ5 questionnaires will be reflected in the results from 2004, and it is likely that it will cause revisions to the MQ5 data. What we would like views on is: Are we right to count group personal pensions as personal pensions rather than as group sponsored (occupational) schemes? 2. Insurance managed funds There is potential double counting of pension contributions which flow from the self-administered pensions sector into insurance managed funds. In the selfadministered pension funds survey, contributions are reported on the income and expenditure questionnaire, and investment in insurance managed funds (and insurance policies and annuities) is shown on the transactions questionnaire and annual balance sheet. The extent to which this flow of funds is regarded by insurance companies as part of its pensions business and is reported by them as pension premiums will lead to double counting in the published estimates of pension contributions As part of our investigations we confirmed with insurance companies that they had included premiums from insurance managed funds in their group sponsored (occupational) pension returns and did not report them as "other business". Where companies wrote this business, most said they had included these premiums under group sponsored (occupational) returns. From 2004, we are asking respondents for data on premium income from insurance managed funds separately on the new quarterly questionnaire and on the annual questionnaire for 2004.
3 What we would like views on are: Are we right in our assumption about how funds flow between pension schemes and the insurance sector? 3. Group additional voluntary contributions Group additional voluntary contributions (that is avcs paid by members of an occupational scheme to buy additional benefits) data are collected separately in the self-administered pension funds survey. These contributions might also be included on the long-term insurance income and expenditure questionnaire from those insurers who write this business. The methodology used to reconcile pension contribution statistics in the table in Annex B of the review report assumes that group avcs are not included in group/company sponsored pension premiums reported by insurance companies. As part of our consultations with the industry we asked where insurance companies had been reporting group avcs. Most of the companies said they had included them in group (occupational) sponsored pension premiums. This would lead to some double counting of contributions to funded occupational schemes and an overstatement of total pension contributions. What we would like views on is: Are all group avcs passed on to insurance companies and reported as group sponsored pension premiums? 4. Bulk buyouts Bulk buyouts arise when a self-administered fund winds up and purchases policies from insurance companies to provide benefits for their members. Premiums from bulk buyouts would be reported by insurance companies on the income and expenditure form. Previously these premiums would have been reported by the self-administered funds as contributions. While we have been unable to find a self-administered fund which has continued to return questionnaires through to wind-up and we do not know how bulk buyouts would be reported by them, discussion with OPRA suggest that they might be reported as transfer payments. As part of our consultations, we asked insurance companies identified as writing this business where they had been reporting bulk buyouts. Most of the companies said they had included them in group (occupational) sponsored pension premiums. We have included a separate box for bulk buyouts on the revised questionnaire. What we would like views on is: Are we right not to count bulk buyouts as pension premiums? Other issues investigated 1. Group life premiums Group life premiums are for life assurance, provided as an additional benefit for pensions contributors, not direct pension contributions. In discussions with the
4 ABI, it was suggested that insurers might be including group life cover premiums, where this was a benefit purchased by the pension scheme, under pension premiums rather than under life. We investigated this with respondents as part of the various consultation exercises. Seven of the companies contacted said they included these premiums under pensions; other companies writing this business included these under life assurance or permanent health. Some companies include life cover as an integral part of the pension scheme paid for from contributions. As companies review their pension provision, group life cover might become a separate contract paid for by the company, and this is likely to be the case where provision is shifted to money purchase or group personal pensions. While these are not pension contributions, it has been argued that some of the cover provided may be used to fund pension benefits such as widows' or dependants' pensions. For National Accounts, premiums paid for group life cover as part of the pension package would be regarded as compensation of employees. We have specified that these premiums should be included under group (occupational) pension premiums on the new form. What we would like views on is this: How do insurance companies view group life cover premiums - as part of pension provision or as life assurance? 2. Single premiums It has been suggested that group (occupational) single premiums reflect mainly movement of assets from the self-administered sector to insurance companies and movement of assets between insurance companies. Movement of assets between insurance companies will be reflected in adjustments for transfers made by National Accounts. There are a number of single premium occupational pension products which will be reported here, and any additional one-off payments by a pension scheme would also be reported here. However, it is likely that premiums for bulk buyouts have been recorded here which would overstate these premiums. We have included a separate box for bulk buyouts on the revised questionnaire. What we would like views on is: Are we right not to treat all single premiums from group/company sponsored schemes as double counting? 3. Insurance companies universe The register used for the insurance companies survey is based on regulatory data. This list has been checked to ensure that companies included in the universe have insurance provision as their main business rather than fund management. Other issues not yet investigated In discussing the findings of our investigations, a number of other issues have been raised which have not yet been fully followed through:
5 1. Additional voluntary contributions The issue of how pension schemes report employee additional voluntary contributions where these buy added years rather than money purchase benefits. It was suggested that these contributions might be retained as part of the overall fund rather than be passed on to insurance companies. 2. "Key man policies" The issue of where insurance companies report contributions to a personal pension scheme where the employer also contributes. While this should be reported as personal/individual contributions, there is a possibility that this may have been reported under group/company sponsored. How do insurance companies report these premiums - as personal pensions or group/company sponsored? 3. Pension annuities The issue of where insurance companies have been reporting pension annuities and income drawdown has been raised. We believe that these are being reported as personal/individual pension premiums. This would overstate the personal pension premiums reported in our MQ5 survey, although it would not affect any other estimates as Inland Revenue administrative data is used for personal pension contributions. How will insurance companies have been reporting pension annuities - as personal pensions or group/company sponsored? Impact on GDP Any change to contributions to funded occupational pension schemes in MQ5 will feed into estimates for compensation of employees, a component of GDP. Many of the issues featured in this note are relatively new and have developed over the last five to ten years, so the impact in the early 1990s would have been minimal. The full extent of these changes will not become apparent until results from the new questionnaires have been analysed. The results of our investigations indicate that there is considerable uncertainty about the pension contributions data reported in the MQ5 surveys, and the results from the new questionnaires may cause other changes, not yet known. It should never be assumed that a change to a component of one measure of GDP would affect the measure of GDP in a precise way, due to the effects of balancing. The effect of these changes on GDP will need to be assessed when they are taken into the National Accounts dataset, bearing in mind the possible impact of other data revisions.
6 Plans for publication The new questionnaires will be used from the first calendar quarter 2004 and for the 2003 annual survey. Quarterly results published will be moved on to the new basis during 2004 excluding the double counting and missing flows. It is possible that this may cause discontinuities in the MQ5 dataset which feed through into National Accounts. The extent of these will not be known precisely until the annual survey for 2003 (using the new questionnaire) can be compared with the original quarterly results, although steps will be taken to preserve the continuity of aggregate series in the National Accounts. Revisions to earlier periods would then be available for Blue Book Please send your comments, by Friday 23rd April 2004 to: Christina Forrest National Statistics Room D245 Government Buildings Cardiff Road Newport NP10 8XG January 2004