Andrew Cox on the importance of good data management

One of the essential tasks of a pension scheme is to pay the right amount of pension at the right time to the right person. Getting this basic procedure right is an essential part of keeping members happy - and will help the trustees sleep soundly at night.

But poor data management that hinders this basic process could have unhappy consequences for scheme trustees. The Pensions Regulator (TPR) has signalled its expectations for data management by schemes and issued a statement of good practice for record-keeping in December 2008. This said, schemes should be aiming for 100 per cent compliance as the benchmark for accuracy.

Just over one year on from the initial statement, TPR is gathering information on data standards at schemes and if it finds widespread evidence of poor data management, standards of data management could become mandatory.

Poor record-keeping at pension schemes could therefore result in future sanctions from the watchdog. This could include the public naming and shaming of the worst non-compliant schemes.

Pension schemes should think carefully about what they can do now to improve their data management.

To clean or not to clean?
TPR has listed events that he would expect to trigger data analysis and cleaning - these could include a change of administrator, or a merger or acquisition. In future, if a scheme fails to act following such events, this could be held against it. Even now, TPR may ask why data was not looked at following such an event.

Schemes should also practice good data management as part of their liability management duties. Even the cleanest data degrades over time if it is not maintained, as members get married, have children, divorce, go part-time, leave the scheme, move house and ultimately die. As a result, regular data analysis is needed to ensure that a scheme's liabilities continue to be accurately stated.

Indeed, poorly maintained data means that schemes will have to pay more for de-risking solutions such as a pension buyout or a longevity swap. One insurer has estimated that the additional cost of poorly maintained scheme records could add 5-10 per cent to buyout premiums, or roughly £2.5-5m on a £50m transaction.

Such additional costs might prevent an employer from taking de-risking measures. Further, enhanced transfer value exercises cannot be undertaken if members cannot be contacted.
Inadequate records could also prevent schemes automating their systems. Because membership data may be incomplete and therefore unsuitable to code the calculations, trustees might not be able to automate administration processes and reduce costs. A scheme will also struggle to reap the benefits when changing administrators if data is poorly maintained. Poor data also increases the risk of fraud.

There may be other issues for members arising from poor data management. Member education is extremely important, especially in a DC environment. But a scheme cannot communicate with its members if it doesn't have their current contact details.
Alternatively, the benefits and cost savings of member self-service might preclude online member access if data deficiencies were exposed.

Planning for the future
With the increased emphasis on DC pensions, trustees should therefore think carefully about the importance of taking action to improve their data management. This will necessitate a change in mindset.

In a defined benefit (DB) scheme, by recoding an individual member's leaving date, joining date and salary history, one can normally calculate benefits at retirement. In a defined contribution (DC) environment, every single contribution and fund movement needs to be accurately recorded. Rebuilding a DC scheme's records to correct past errors can cost hundreds of thousands of pounds, not to mention the additional out of market investment risk.

Other common historic causes of poor data include administrators inheriting poor data, recurring data interface problems, or systematic issues arising from faulty procedures or incorrect calculations. Poor administration - for example, the use of notes fields on administration systems or 'gone away' entered in the address field when correspondence is returned - degrades data accuracy;

Given these problems, the scrutiny of TPR and the fact that clean data saves money, schemes should consider taking action to understand and address any potential data problems. This will help them stay on the path towards the 100 per cent compliance benchmark set by TPR. Clean data cannot be achieved overnight, but schemes need to be able to show how they intend to attain it.

Practical steps to improve data
There are a number of steps that can be taken to improve scheme data.

Firstly, create a plan to analyse scheme data, then address the specific issues raised. For example, member tracing exercises conducted by specialist tracing agencies are invaluable ways of verifying and updating member records. This could cost between £2.50 and £10 per record and a 'no find, no fee' charging basis is possible. Mortality screening can also be conducted to ensure pensions are not being paid to deceased members. These screenings should be carried out on a regular basis (insurance companies test their books every quarter to ensure their records are up to date) and can often pay for themselves, especially if de-risking exercises are being considered.

Take specialist advice
Some trustees prefer to use independent consultants to conduct data analysis, project planning and cleaning of their members' records. This removes the problem of incumbent administrators facing potential conflicts of interest if the data they have been maintaining is in a poor condition. Once the consultants have reported on the state of the data, the costs and benefits of cleaning it can be assessed and a prioritised recovery plan devised.

This should set a timetable, allocate responsibilities and deliver regular progress reports for trustees. Once data is cleaned, it needs less subsequent work to maintain it. Administration costs may be reduced and service levels improved.

For trustees, it is vital that they can show that they are taking the issue of data management seriously. Rather than relying on subjective assurances from their administrators, trustees need to obtain regular objective reports and monitor actual results against targets, with remedial action taken where necessary.

Data management is a key ongoing priority for pension schemes. We can expect to see further announcements from TPR; particularly if it feels that its good practice guide has fallen on stony ground.

Ultimately, it is in the best interests of pension schemes to practise good data management, but they may have no choice anyway, with the industry's watchdog wanting to be sure that this is happening.

Andrew Cox is a senior consultant in the Employee Benefits Practice at Lane Clark & Peacock LLP

Data management case studies
1. Incorrect payment of spouses' pensions
Situation: A pension department administering DB and DC plans for approximately 1,800 members needed to find out the impact and cause of why, on the death of a pensioner, the surviving spouse was given half of the post-commutation pension, instead of the pre-commutation pension. The client called for the development of a robust technical solution to a sensitive and potentially expensive issue, which it could communicate without de-motivating the administration staff and tying up valuable time.

Approach: An exercise was carried out to discover the scale of the problem and the cost of resolving it. It was also necessary to investigate what had gone wrong and to find a solution to prevent any future re-occurrence. To tackle the problem, the original paper files were needed in order to obtain the pre-commutation pension at exit. This enabled the calculation of the pre-commutation spouses' pensions at retirement plus increases to the present time or date of death to update member records. This provided trustees with sufficient information to exercise their discretion regarding the re-adjustment of spouses' pensions in payment to the correct levels.

2. Data cleansing
Situation: A DB scheme with approximately 3,800 members held incomplete data records on a number of members they suspected to be deceased. It wanted to sensitively carry out a mortality screening exercise, so that it could obtain full death registration details and copies of death certificates.

Approach: On behalf of the trustees, a project was initiated to work with an external tracing agency on a mortality screening exercise. The scope of the exercise was agreed with the scheme, analysing member data, providing project management skills, reviewing the results, providing a cost benefit analysis, resolving queries and reporting back to the client.

The exercise identified a number of deceased members and copies of death certificates were subsequently obtained. Solutions were suggested for recovering overpaid pensions and further courses of action for those members who could not be traced. As a result, the trustees had at hand all the information they needed to decide on what they should do going forward with the data.

Liabilities were reduced, member records were updated and an effective solution provided - much more quickly than the trustees expected.

Share Story: