In 2008 The Pensions Regulator published a record-keeping guide, in the hope to motivate pension schemes to initially review their data and come up with a basic plan to address issues around so-called common data (names, addresses and other basic information), although the Regulator will also require clean conditional data by December 2012.

The conditional data focus on questions such as what pensions should be in payments, what increase it should get and what spousal
pension the spouse should get if the member dies.

The problem of unclean data is still not solved, although many schemes are working on it. The Regulator said: "As a result of our research, and from sampling 689 schemes from our database throughout 2009, 85% of schemes surveyed said they had, or planned to have, a means of measuring common data. The equivalent figure for conditional data was 79%."

Richard Murphy, partner at LCP, explains why cleansing data is not as easy as it sounds. "Schemes obviously run a very long time, and people who worked for the company in the 70s and 80s and left at that point, may only have paper records of what benefits should be paid to them and even those may be incomplete. Many schemes have that issue, while others have just been poorly administered."

There are many problems that trustees face when it comes to
data cleansing. One example is additional voluntary contributions (AVCs) members have made, which are sometimes overlooked by the administrator if the member transfers or retires, meaning that there is some orphan money there.

Another example is where some of the benefits have been insured but when the administrators have not told the insurer about the death of a member, which means the insurer keeps paying money to the pension scheme. A large payment is then due back to the insurer, because they have been paying pensions of people who already passed away.

Avgi Gregory, director at Muse Advisory, says: "There are huge gaps in the data, which are very difficult to be traced. The data has to be extracted from data fiches or from scan documents, some of which might be very tough to read, as they are fading away. The sheer workload of cleansing data is very costly and time-consuming."

This brings up the issue of the costs, which is something the industry hasn't yet found a solution for. Gregory says: "It is one of the most expensive exercises that many trustee boards are now facing. One of the biggest challenges is who pays for it: the trustee, the sponsoring employer or the third party administrator?"

Murphy adds: "The costs vary enormously, but it's worth thinking that even at £100 per individual, this would be a £500,000 exercise for a scheme with 5000 members. Very often it is a six figure sum to address data issues, which can be a real shock to trustees who thought they were paying an administrator to keep the records in order."

The Regulator believes that it depends on the scheme type and the reason why the data needed to be cleansed who will be paying the bill. "It is fairly scheme specific as to who might be most liable for the costs. If the maintenance of scheme data falls under the remit of an administration provider's contract or service level agreement, then trustees should not be expected to incur further costs for data cleansing. Trustees should be having robust discussions with their administration providers to ensure they are getting value for money on behalf of their members."

Gregory points out that technology might pose a future risk to clean data, as some HR systems have four address lines, while others have five, meaning that one line, often the postcode, gets lost when a transfer to another system takes place.

Communication is also something that is important when it comes to data cleansing. Gregory says: "If a communication exercise is to be done, what is the message, how do you communicate and when do you communicate? Let's say a problem has come up in a fairly long program of data-cleansing. Do you write to that member and say a problem has come up on this particular element of your data or do you wait until the whole process gets completed?"

Share Story: