Your browser is out of date, and unable to use many of the features of this website

Please upgrade your browser.




This guidance describes an approach that we consider to be good practice for measuring the presence of member data items which are important in the administration of a pension scheme

Issued: June 2010

Key points

  • Poor record-keeping may lead to significant additional costs in a number of areas such as administration, error correction, claims from members, buy-outs, wind-ups and, potentially, may necessitate the making of more conservative actuarial assumptions. Where a scheme’s poor record keeping becomes known, it can cause reputational damage.
  • To address these issues, this guidance details our approach to the testing and measurement of member record data, particularly in relation to:
    • common data – applicable to all schemes
    • conditional data – dependent on scheme type, structure and system design
    • numerical data – to aid understanding of the results
  • The guidance is aimed at trustees, providers and administrators, who as well as checking that data exists, need to ensure that their data is accurate. It sets out a proposed framework for data checking.
  • Our aim is to focus on educating and enabling those responsible for member record-keeping and those who administer pension arrangements, to improve the standard of record-keeping across the industry.
  • We describe our approach to enforcement where there’s evidence of poor record-keeping and no plans to correct it.
  • The principles of good record-keeping apply to all work-based schemes and we have consulted with the Financial Services Authority (succeeded by the the Prudential Regulation Authority and the Financial Conduct Authority on 1 April 2013) in preparing this guidance as it applies to contract-based schemes.
  • This guidance sets out good practice in helping to assess risks associated with member record-keeping. It describes an approach for measuring the presence of member data items which are important in the administration of a pension scheme.

Measuring member data - an overview

  • Measurement of common and conditional data:
    • common data - items which are applicable to all schemes
    • conditional data - dependent on scheme type, structure and system design

    We expect all schemes to have measured their scheme data and we recommend targets for the presence of common data.

  • Reporting in context - numerical information:
    • numerical data - to aid understanding of the results, reporting should include additional data to put the test results into context
  • Risk assessment - evidence of internal controls:
    • the tests are part of risk assessment processes and provide an indication of the effectiveness in meeting the legal requirements to have adequate internal controls
  • Developing improvement plans:
    • where the tests indicate there may be a problem, further investigation may be required and plans put in place to correct errors or fill gaps in data
    • we expect plans will ensure that problems with record-keeping are corrected by the end of 2012
  • Annual measurement (or more frequently if the improvement plans require it):
    • regular testing will provide evidence of changes in the state of the data
  • Review in 2011 (what we have learned):
    • we'll review progress in 2011 on the take-up of our guidance and its effectiveness in addressing problems identified in our earlier consultation
  • The Pensions Regulator's education material:
    • we’ve recently updated the Trustee toolkit to include examples of administration problems associated with record-keeping
    • we’ve signposted below some of the legal obligations to keep records

Introduction and approach to measuring member data

1. This guidance sets out proposals that are good practice in helping to assess risks associated with member record-keeping. It describes an approach for measuring the presence of member data items which are important in the administration of a pension scheme.

2. The proposals have been developed from industry feedback on our two consultation papers Record-keeping: a consultation document (PDF) and Record-keeping: measuring member data (PDF) which reviewed progress in the first year. The consultations covered the importance of maintaining good records, industry evidence of poor record-keeping, some of the costs which this can cause, the results of surveys showing slow progress, and the regulator's approach to this important administrative function. We thank all those who gave time to helping us and for the strong support we found in this subject.

3. The principles of good record-keeping apply to all work-based schemes, and we have consulted the Financial Services Authority (succeeded by the Prudential Regulation Authority and the Financial Conduct Authority on 1 April 2013) in preparing this guidance as it applies to contract-based schemes.

4. We recommend that providers and administrators of pension schemes measure the presence of some of the most important items of member data and report on the results. The proposal is for the measurements to be based on a limited set of data items – it’s not to measure all data.

5. In trust-based schemes, trustees are accountable for the quality of member data; in contract-based schemes it’s typically the manager who’s legally accountable.

6. We wish to emphasise that simply measuring the presence or absence of data doesn’t provide any evidence that the data is accurate. Trustees and providers need to be conscious of this, and they should reassure themselves that they have controls in place to ensure the accuracy of their data, as well as checking that the data exists.

7. We recommend that schemes take up our guidance as an essential tool within their internal controls processes. We continue to be aware of problems with member records and our review in 2009 highlighted that the progress being made to address these was limited.

8. We have therefore concluded that further action is required. We are introducing targets for the standards of data and a deadline (December 2012) by which we hope schemes will have addressed any problems.

9. The proposed framework for data checking is illustrated in the following tables. Examples of the type of data in each heading are included in The data to be measured.

The proposed framework for data tests

Report item Purpose  What's recommended  Method 
Common data

To uniquely identify a member.

(having regard to the possibility that any individual may have had more than one period of membership of a scheme).

All data fields should be populated.


The data and the tests are specified by the regulator following consultation with those involved with pensions.

Conditional data

To provide a view of the data required for the effective administration of the scheme, and to allow providers or trustees to measure changes in the data over time.

The existence of this data will depend on the type of scheme, scheme design, system design, and an individual’s membership status.

That the presence of the conditional data required to administer a scheme is known.

That, where necessary, action plans are put in place to improve conditional data.

Such plans should involve improvement within a specified timeframe.

Scheme-specific measurement.

The methodology is recommended by the regulator, but the actual data and the tests are specified by the trustees/provider.

The regulator will collate information for possible future benchmarking.

Numerical information To enable consideration by the trustees, providers and others of whether the results reported reflect their knowledge of the scheme. Commentary to provide context and rationale for development of improvement plans. Report.

Common data

10. There’s a set of data that’s necessary and applicable to all members of all schemes. In some circumstances, one or two of these data items may be derived rather than held explicitly.

11. The absence of one or more of these items, or an error in any of them, is highly likely to mean that the member cannot be identified or traced, or the member's benefits cannot be correctly established with any degree of certainty. We recommend that every item of this data should be present in all schemes.
12. Details of the items that we consider to be common data, and the validation checks that should be applied to these data items, are included in The data to be measured.

13. The proposed measure of common data is concerned with the presence of basic information about the member. It doesn’t enable the provider to calculate and settle benefits.

14. The targets for the standard of common data we recommend is achieved by December 2012 are:

  • For new data created after June 2010 – 100%
  • For legacy data (created before June 2010) – 95%; this lower level is in recognition that many schemes will have greater difficulties with older data and may inherit problems from previous trustees, managers or administrators.

The consultation response Record-keeping: measuring member data (PDF), published May 2010, sets out the rationale for these targets.

Conditional data

15. There is additional detailed data required for the administration of a pension scheme. The nature of this data will vary from scheme to scheme, and will depend on many factors. The presence of these data items is dependent on the type of scheme, scheme design, a member's status in the scheme, events that have occurred during an individual's membership of the scheme, and system design. This data is therefore conditional upon all of these factors.

16. Scheme providers, administrators and trustees are best placed to decide the exact constituents of this conditional data, based on an understanding of the requirements of their own scheme and administration system.

17. We don’t think it would be desirable to try to produce a data specification covering every possible circumstance. Therefore, we’ve produced tables showing the items that could provide an outline of what can be regarded as conditional data (PDF). These tables should only be regarded as illustrative.

18. The tables show, in broad terms, the circumstances in which a particular conditional data item might be required.

19. Providers, administrators and trustees will need to make adjustments to the list of items to take account of the circumstances of their own scheme and systems. We’re not necessarily suggesting that all data should be checked.

20. Targets for the standards of conditional data should be set by the trustees in conjunction with the administrators, or by the provider. This scheme-specific approach recognises that schemes may have unique characteristics, but we emphasise that this data is still essential, particularly for the calculation of entitlements, and should be treated no less seriously in plans.

Numerical data and context

21. As well as the common and conditional data, there is numerical information regarding membership records that we believe will help put the results of other measures into perspective. It will enable consideration of whether the number of such records is broadly consistent with knowledge of the scheme.

22. Some of the numerical information that we believe could be helpfully provided is set out in The data to be measured. This would be reported at the same time as the data is validated. Again, the list should only be regarded as illustrative.


23. There should be a commentary accompanying the results reported, giving the context and rationale for the information provided. This will help with the development of any further plans.

Risk assessment, controls and action plans

24. The proposals above should form part of a risk assessment process and an adequate internal controls framework. The measurements are designed to be indicators of whether member record-keeping needs further consideration in the context of risks to good scheme administration– measurement is not an end in itself.

25. We’re exploring with the auditing community whether the approach outlined above can be supported by the statutory audit process. For example, auditors communicating the existence of poor member data, together with other risks identified, to their clients as part of the audit process. We will hold trustees accountable for poor member data; they in turn will want to consider what role their auditors should play in meeting this accountability.

26. Where tests indicate that there may be inconsistencies or errors, the regulator recommends that providers and administrators work with trustees and employers to develop action plans (including timetables) for further investigation and correction as appropriate.

27. The action plans should include all reasonable endeavours to meet the targets set out above, and a timetable for corrective action to be completed by December 2012.

28. We believe it’s likely that records identified as having invalid common data (other than out-of-date addresses) are also liable to have other defects. When investigating these records we therefore believe it appropriate that the whole of the member's record be reviewed.
29. Our recommended approach for good practice is one of continuous improvement with reports of progress against plans being made on a regular basis. After completion of the plans, the limited measurement tests should be conducted on an annual basis as evidence that controls are continuing to operate effectively.

Events that cause changes to a scheme

30. During the life cycle of a pension scheme, events occur that cause significant change to the scheme itself, or to members' benefits. These events may present additional risk of data errors occurring, or increase the need for data accuracy. A list of the events that we have identified is in Appendix 1.

31. For some of these events, the regulator would strongly recommend that a data-cleansing exercise is undertaken, for example entry into a Pension Protection Fund (PPF) assessment period. Other events that are listed, whilst not requiring immediate action, present the providers, administrators and/or trustees with a good opportunity to review record-keeping. We have differentiated between these types of event.

Obtaining and reporting on data

32. We acknowledge that there are numerous systems in use for recording data. These include:

  • systems provided by specialist pensions software providers
  • systems designed and run by insurance companies
  • systems designed by schemes administered by an in-house pensions department
  • manual systems.  

Many computer-based systems are supplemented by paper and microfiche records.

33. The amount of checking that we propose is defined as a small set of common data fields and a limited number of conditional data fields, some of which may be scheme-specific or system-specific. It’s not a proposal for a check of all data. In this context, our proposal shouldn’t be unduly onerous and moreover is part of the good governance of a pension scheme. Most computer system providers will be able to provide programs to perform data checks at this level and produce reports. Some products already do something similar, if not more.

34. Where data isn’t held on the computer system, providers, administrators, trustees and employers should consider whether this is the best method of holding records and whether their internal controls over this aspect are appropriate, taking account of the risks involved.

The data to be measured

35. The tables below set out general descriptions of each test and its purpose. They do not attempt to be technical specifications which would be difficult to prescribe for all circumstances.

36. There are very significant differences in the data requirements and legal structure of contract-based schemes. They are predominantly defined contribution (DC) schemes, and trust-based schemes which include final salary, DC, flat rate, career average revalued earnings (CARE) and hybrid schemes. The data tables show where tests are appropriate to trust-based and contract-based schemes, and also show which items of conditional data apply to different scheme types.

Considerations for trustees and providers

37. Dialogue between trustees/client and provider is essential in designing the tests and the action plans. If there’s evidence of problems, or data can’t be tested, consideration should be given to other forms of assurance. Appendix 2 describes some types of data which we haven’t included in the tests, but which should be subject to other forms of controls and risk assessment.

Common data

38. The most basic data items required to uniquely identify a scheme member are common to all schemes. It’s therefore possible to test the overall quality of recording of these data items at scheme level. As all these items are completely fundamental, these fields should be fully populated.

39. A poor result in any tests should lead to plans to identify the causes. As well as the scheme-level summary, a member-level breakdown of the data item failures should therefore be available.

40. Below is a list of the common data items, together with an indication of the tests that we propose should be carried out. The test is checking for the presence of data which is valid and appropriate to the type of field described.

41. The targets for the standards we expect to be achieved by December 2012 are:

  • For new data created after June 2010 – 100%.
  • For legacy data (created before June 2010) – 95% (this lower level to recognise that many schemes will have some difficulties with older data).

The consultation response Record-keeping: measuring member data (PDF) published May 2010 sets out the rationale for these targets.

Common data items required for all members of all schemes

Data field Comment 
National Insurance Number 'TN' formats should be regarded as missing data. The final character of NI numbers is not essential.
Surname Check that surname is present.
Forename(s) or initials Forenames are preferable but initials are an acceptable alternative.
Sex Check that sex is present
Date of birth Check that date of birth is present and consistent (earlier than date joined scheme, retirement, date of leaving). False dates should be classed as missing data.
Date pensionable service started/policy start date/first contribution date
For trust-based schemes this will be date pensionable service started. For contract-based schemes this will effective start date of the policy or the first contribution date, depending on the provider's requirements.
Expected retirement/maturity date (target retirement age)
This field may be derived or explicit; for most DB schemes it will probably be derived as the scheme's normal retirement date. Need to check that it is populated if that is a scheme/system requirement, that it is consistent with scheme rules and statutory requirements, and is later than date of birth and pensionable service date/first contribution date.
Membership status
Check that a current valid status is recorded for each member. This may be a dual status, eg active or deferred member with partial retirement. For contract-based schemes this may be 'active' or 'inactive'.
Last status event Check that benefits taken are consistent with status, and, if status history is recorded, that the latest status is the same as the explicitly recorded current status.
An address should be present for all members of all schemes. Because of DPA requirements an exception is permissible for active members of those trust-based schemes in which communication with members is normally sent via the employer. 'Gone away', 'unknown' or similar should be treated as missing data.
Check that a postcode is present if address is not identifiable as being overseas. Will assist with valuations for actives, for whom storing full address may breach DPA principles.

Illustrative conditional data tests

42. Providers, administrators and trustees are best placed to decide the constituents of the conditional data required for effective administration, based on their understanding of the requirements of their own scheme and administration system.

43. We recognise that there are scheme types other than those shown in the tables. Providers, administrators and trustees will need to use their knowledge of the scheme to decide which data items are relevant.

44. We recommend that the presence of the conditional data required for effective administration is measured and that, where necessary, a plan is developed for improving it.

45. Reporting of the results should be supplemented by a commentary on detailed data records. The results should be available for comparisons to be made at intervals.

46. We've put together an illustrative list of conditional data. This list is not exhaustive and it should be tailored to the requirements of the scheme and its administration.

47. When undertaking a check of member records, administrators should make clients aware of the data that’s being checked and of any data items that won’t be checked, and the reasons why. Trustees need to be given sufficient information to enable them to understand any constraints on the record checks being undertaken.
48. Targets for the standards of conditional data should be set by the trustees in conjunction with its administrators, or by the providers. This scheme-specific approach recognises that schemes may have unique characteristics, but we emphasise that this data is still essential, particularly for the calculation of entitlements, and should be treated no less seriously in plans.

Table showing illustrative list of conditional data (PDF)

Helpful numerical information

49. For trust-based schemes, some numerical information should be made available to trustees, to supplement the basic scheme membership numbers which are frequently reported. This will enable them to check that the scheme's data is broadly consistent with their own understanding of the scheme and its history. The suggested items to be reported are set out in the table below. We would normally expect to see much of this in an administration report anyway. Reporting on the numbers will help put some of the other measurement results into perspective.

50. For contract-based schemes, we would strongly recommend that the tests are completed and measured in a way that accurately reflects the number of members. Where there are extremely large numbers of members involved, sampling may be appropriate. Testing should be by product, scheme, or system to recognise their different characteristics and the total numbers should be reported alongside the test results.
Some examples of the numerical information which would assist understanding of results are shown in this table, but clearly if an item is not relevant to a scheme then it need not be reported upon.

Examples of numerical data

Test Brief test description  Reasons for reporting
Membership statistics
Number of benefit categories Number of benefit categories that give rise to a separate benefit structure (eg staff/works, different section, previous acquisition etc) eg could have different benefits for different groups etc.
Status Membership statistics (active, deferred, in payment)
Consistency with other reporting
Pensioner type
Pensioner membership by type (member, spouse/dependant/civil partner, child)
Specific data statistics

Number of active members with part-time service history
Aid to understanding of the scheme
Number of DC active and deferred members with a lifestyle investment strategy
Number of active and deferred members who have payment of AVCs recorded
Transfers in
Number of active members with a transfer in recorded
Divorce (pension sharing)
Number of members who have had a pension sharing order
Divorce (earmarking)
Number of members with an earmarking order
Members with primary or enhanced protection
Number of members with primary and/or enhanced protection

Checking calculation results

51. Even when a member's data is complete and consistent it doesn’t guarantee that any benefits recorded for that member have been calculated correctly. The proposals are concerned with the presence of the records required to calculate and settle benefits, rather than the calculations themselves. Nevertheless, benefit calculations are also self-evidently crucial, and we expect that providers and trustees will reassure themselves about the integrity of their data by implementing checks of these calculations.


52. We expect that administrators and providers will wish to develop their own new reporting formats, in their own style.

53. During consultation we saw several different examples from providers of their adapted reports. For ease and illustration we’ve included in Appendix 3 a simple example of how results might be shown, but would encourage the development of alternative styles which suit the client/provider yet still provide the same level of assurance.

54. For trust-based schemes, we would expect the trustees to be the main recipients of reports but they may wish to share these with employers who are the scheme sponsors and the route through which much data is received.

55. For contract-based schemes:

  • we would expect organisations to be reporting within their own control and assurance processes. Where employers have a management committee or similar arrangement in place with their pension provider, then we would expect such reports to be available and discussed to enable any process improvements to be made;
  • where sampling techniques are used in very large schemes they should be capable of being demonstrably representative of the population as a whole, statistically robust and comprehensive; and
  • we would expect to be able to see and discuss the results of such tests with providers, but we don’t require them to be routinely sent to us.

56. Where results indicate further investigation or data improvement is needed, action plans should be developed to address gaps and weaknesses according to the risks involved, by the end of 2012. Clearly such plans will involve re-testing to demonstrate progress.

57. Schemes that show few problems should re-do these simple tests on an annual basis to evidence the effectiveness of internal controls, as part of their risk assessment processes.

Legal background

58. There’s a significant body of law which requires good records to be kept and which applies variously to trustees, managers and providers of pension schemes. Legal obligations to keep records about members and their benefits arise from many sources, including trust law, primary and secondary legislation, tort, contract law and European law.

59. The Pensions Regulator is not responsible for enforcing all these obligations. However, it does have the power to enforce the requirements of pensions legislation using its statutory powers. These powers include the use of Improvement Notices, Third Party Notices, and the imposition of Civil Penalties.

60. Under section 249A of the Pensions Act 2004, the regulator can also take action against schemes for failing to establish and operate specific adequate internal controls; schemes must be administered and managed in accordance with the scheme rules and the requirements of the law. This is a wide-ranging duty that goes right to the very heart of good governance and administration. A failure to keep basic records about members and the existence and nature of their entitlement (for example, the common data items) would be wholly inconsistent with discharging this duty properly.

61. In addition, in terms of specific statutory obligations related to record keeping, schemes are required to comply with many detailed disclosure obligations. These can only be met if proper records are kept in the first instance – otherwise it would be impossible for the correct information to be disclosed to the correct person at the correct time.

62. Statutory record-keeping requirements are set out in:

  • The Pensions Acts 1995, 2004, and 2008
  • The Occupational Pension Schemes (Scheme Administration) Regulations 1996
  • The Occupational and Personal Pension Schemes (Disclosure of Information) Regulations 2013
  • The Occupational Pension Schemes (Internal Controls) Regulations 2005
  • The Data Protection Act 1998
  • The IORP Directive


63. This guidance details our approach to the testing and measurement of member record data. In particular, the approach to:

  • common data – of relevance to all schemes
  • conditional data – dependent on scheme type, structure and system design
  • numerical data – to check the overall reasonable of the test results.

64. Since we issued guidance, we’ve focused on educating and enabling those responsible for member record-keeping and those who administer pension arrangements, with the objective of improving the standard of record-keeping across the industry.

65. We now intend to supplement this by strengthening our regulatory approach. We will take enforcement action where the evidence we gather indicates a breach of pensions legislation. Where record-keeping problems are so severe as to indicate a failure to maintain adequate internal controls, resulting in a failure to administer the scheme in accordance with the scheme rules and the broader requirements of the law, these schemes will be prime candidates for investigation.

66. However, consistent with the regulator’s approach of working with trustees, providers and administrators where possible, we would be less likely to enforce breaches of legislation where trustees and providers are in the process of implementing realistic plans to address their identified data problems.

67. We’ll continue to review progress within the industry and will publish a further update in 2011. This will include the results of data audits of a sample of schemes which we will select this year. This will enable trustees and providers to see how their results compare against others.

Appendix 1: Scheme change events

1. The following table shows those events that we consider give rise to an urgent need to review record-keeping, irrespective of any longer-term plans. When one of these events occurs, we will expect providers and trustees to give high priority to reviewing and cleansing the scheme’s data.

Circumstances Scheme type  Reason for validation
Wind-up (full or partial) All work-based
Accurately secure benefits
Entry to Pension Protection Fund assessment period or the Financial Assistance Scheme
Accurately secure benefits
Change of administrator
All work-based
Risk of data and knowledge loss. All data becomes legacy data
DB Accurately measure liabilities and secure benefits

2. The following table shows those events that we believe offer providers and trustees a convenient opportunity to review record-keeping. We believe they increase the risks of errors arising if data is incorrect. Some of them also present a risk that the event itself will give rise to a deterioration in data quality.

Circumstances  Scheme type  Reason for validation 
Valuations DB
Accurate measurement of liabilities.
Scheme closure to future accrual etc
Accurate measurement of liabilities.
Ceasing to contract out
All work-based
Accurate assessment of contracted-out liabilities and reconciliation with NISPI.
Scheme rationalisation
All work-based
Risk of data and knowledge loss. Benefit entitlement may ‘crystallise’.
Merger & acquisition
All work-based
Can trigger many of the other circumstances. Accurate measurement of liabilities, risk of data and knowledge loss.
System change by employer, administrator or investment manager
All work-based
Risk of data and knowledge loss.
Change in investment manager
All work-based
Potential for reconciliation issues. (Note – this also applies to DB scheme AVCs).
Insurance products closed to new business
Lack of support for product.
Benefit statements issued
All Opportunity to verify some data items

Appendix 2: Specific matters raised in consultation

Data excluded from our proposal

1. To avoid over-complication, not every data item is included in our illustrative lists but the matters listed below were specifically raised in feedback during our consultation. Administrators and providers need to ensure that all relevant data is being recorded in a way that will facilitate paying the right benefits to the right people at the right time, and support other scheme functions such as valuations, reporting and accounting, transfers etc.

Member engagement

2. We appreciate that maintaining the accuracy of some data also requires member engagement, such as address details. We expect providers and trustees to use reasonable endeavours to ensure that data is both accurate and up-to-date. They should in any event be aware of the scale of any gaps or deficiencies in their data. Our new targets for legacy data are intended to reflect our expectation that addresses should be kept up-to-date as much as possible. We would be interested to hear about systems and methods which particular schemes are using to achieve this successfully.

Salary records

3. Scheme design, as well as the record keeping system used, will dictate the type of salary information that needs to be recorded. Salary information needed may include basic pay, gross earnings, pensionable earnings etc. We haven’t been prescriptive about which salaries have to be recorded, other than to specify that the relevant salary history is required. Dialogue between provider, administrator, and client is essential.

DC considerations

4. We set out below some internal control considerations for DC processing. In all of these, some dialogue is also essential between all the interested parties. DC processing requires robust processes, including clear audit trails that enable trustees to be assured that data is complete and accurate, that transactions are being actioned promptly and correctly, and that transactions are accurately recorded in member records. The process needs to ensure that contributions are allocated to the correct funds, and that changes to members' choice of funds are correctly actioned. Trustees need to be aware of the risks inherent in DC processing and understand how the investment process works. Trustees are unlikely to have this level of knowledge initially, and we believe that administrators have an important role in explaining these issues to them.

5. Switching - trustees need to be sure that there are adequate controls in place to ensure that all transactions resulting from investment and switching activity have been actioned both promptly and correctly.

6. Lifestyling - trustees should seek reassurance that providers have robust processes for ensuring that any lifestyle and phased investment options are correctly recorded for individuals. We recommend that, when record checks are conducted, the records of all members who have taken a lifestyle/phasing option are checked to ensure that their investments correctly reflect their position within the lifestyle/phasing process.

7. DC roll-up – trustees have a responsibility to their members to keep records of all their transactions, and providers and trustees need to be aware that members sometimes have queries about their past contribution and transaction history. Many computer systems regularly roll-up DC contribution and investment data to give current totals. As a result, for many schemes it may be difficult to check historical transaction records for each member on a month-by-month basis. However, trustees should consider how they will access historic information many years in the future.

8. Contributions records - there will normally be a full record maintained of all contributions received during the current scheme year, and sometimes for the previous scheme year. We therefore expect that member records are checked for the presence of contributions during these periods. We accept, however, that this will not always be practical in some schemes in certain circumstances.

9. Risks of inadequate controls – it’s particularly important that trustees of DC schemes are aware of the risks of inadequate DC processing controls. In particular, we believe that administrators need to draw the attention of trustees to their operational procedures, the way in which contribution and member data is processed, and the investment and reconciliation process followed by the administrator. We believe that this is the only way in which trustees can become fully conversant with the operational and data recording risks, and how their responsibilities in this area are being met by the administrators. There are huge risks in simply assuming the presence and accuracy of all member records.

10. Remitting contributions - in trust-based schemes it’s normal for contributions to be deducted by the employer's payroll department and paid to the trustees' bank account (or directly to the investment provider) by the company. Trustees will wish to assure themselves that the processes for calculating and remitting contributions are thoroughly tested and secure. It’s important that trustees are aware of their responsibility for ensuring that contributions are handled correctly.

11. Unit reconciliation - although not strictly a check on individual member records, we regard it as of fundamental importance that trustees regularly reconcile the total number of units held by investment managers with the units allocated to members. We believe that this reconciliation should ideally take place whenever there’s an investment or disinvestment, and in any event should be conducted monthly. This provides one measure of the integrity of member records in totality, albeit not at an individual member level.

Partial benefit crystallisation

12. Not all computer systems have yet caught up with legislative changes during the past few years, eg the ability to take phased retirement. Nevertheless, computer records must be able to identify members in this category and record all relevant information needed to perform future benefit calculations, as well as to pay those benefits that have crystallised.

Use of 'remark' and 'comment' records

13. Free-format records for remarks and comments are a feature of most computerised systems. Historically, such records have been used to store many types of data, either temporarily (whilst computer systems were being enhanced to cater for changes in requirements) or permanently. While such records have a valid function, most computer systems are unable to take account of their content in any calculations or reports, and this needs to be taken into consideration when considering the overall quality and usability of this data. Where fields of this type are used to record critical data (ie data required to calculate a member's benefits) providers and trustees need to be aware, and give consideration to any risks posed.

PPF and scheme wind-up issues

14. Some may query why particular items of data are included in our list, eg employing company details for trust-based schemes, spouse's/civil partner's/dependant's name and date of birth. These details become acutely relevant in certain contexts – eg when a scheme is winding up or entering a PPF assessment period. This information is essential, and it’s been included in the conditional data items for this reason. Further details of the PPF's requirements are available on their website. Where data tests are being considered for a scheme in an assessment period, we recommend advice is taken from the PPF case manager.

Appendix 3: Example report on data quality