Duncan Watson, managing director of products and services, EQ Paymaster at Equiniti, talks to Francesca Fabrizi about the importance of conditional data to good scheme governance and how to get there
What constitutes conditional data and why should trustees be thinking about it now?
Watson: There is some confusion around what conditional data actually is, so let’s start with a very simple definition of the difference between common and conditional data. For me, common data includes the things that identify an individual – so their date of birth, their sex, things that would make sense outside of the pension scheme, that make sense in normal life.
Conditional data is the data that links that person to that particular pension scheme. For example, their pensionable salary, the date they contracted out and so on. So it makes sense within the pension scheme environment – it’s the data that the pension scheme then uses to calculate the benefits that are due to that particular individual, in very simple terms.
It is important because it really underpins the running of the pension scheme. Without good quality conditional data, the pension scheme can’t calculate and pay benefits; they can’t undertake annual exercises such as pension increase exercises; they can’t communicate with their members about their entitlements in the scheme; and they can’t really conduct the annual valuation.
Why should looking at conditional data be more important for trustees than their other financial priorities?
First of all, you can’t really run the basic pension scheme without good quality conditional data – that’s my first reason.
I’ll break my second reason into three different sections. Trustees and scheme sponsors undertake proactive activities with regard to their pension scheme over the course of the life of the scheme. So there’s the day-to-day running of the scheme. Then there are other things they might choose to do alongside this. For example, they may choose to engage their membership, they may choose to introduce self-service modelling tools so that their membership can log onto the pension scheme website and plan their future. They can do all that, but without good quality conditional data underpinning that, those modelling tools are frankly next to useless. So it hampers their drive to engage with their membership.
Secondly, in trying to drive down the efficiency or the cost of running the pension scheme, they may ask their administrator to introduce more automation. Again, you can automate the life out of a pension scheme, but if the data underpinning the scheme, the data relating to the members is not credible and not whole, then there’s no point in having automation, because the automation will just spit out the wrong results.
My third reason is that many schemes are looking at the lifespan of a scheme, their funding plan, their flight plan and also possibly at de-risking the pension scheme over a period of time – through buy-in, buyout, through pension increase exercises or ETV exercises. Again, without that fundamentally strong conditional data, those exercises are really difficult to execute. In fact, without good quality data, an insurer will not transact with the pension scheme. Without good quality conditional data it’s just a nonstarter really.
So the trustees have got a choice. They can either look at their conditional data in a piecemeal fashion – so they decide to do an exercise, to look at the conditional data that relates to that exercise, fix it there and then, then move onto the next exercise. So doing it reactively or piecemeal is an option.
The option I would recommend is that trustees adopt a data strategy. So, running alongside all of the other things they need to do, the day-to-day running, all of the ‘nice to have’ exercises, they have running along in parallel a data strategy, a conditional data strategy so it almost becomes a habit to be looking at that data. Is it right, is it whole, is it appropriate, is it consistent? Then fixing it as part of their business.
Can we delve deeper into the risks that trustees face if they have poor quality conditional data?
I’d categorise risk into three areas. There’s a reputational risk for having poor quality conditional data. There’s a financial risk and there’s a lost opportunity risk.
From a reputational risk, we know the regulator isn’t happy with the way schemes are looking at their data and managing their data. They said so at the end of 2016; and from 2018, from the start of next year, they’re asking trustees in their scheme returns to comment on the quality of their conditional data. This is something that they’ve never done before, so it’s a sign that the regulator is starting to get tough on data. For me, if schemes don’t adhere to that, then the regulator, with its powers, will intervene and that could damage the reputation of the scheme if that gets into the pensions press.
The other area of reputational risk is at an individual member level. If you have poor conditional data, there is a risk that you will pay benefits incorrectly. That could be on specific high profile cases, or on a number of cases at the same time. That could impact the reputation of the pension scheme and the sponsoring employer.
Then there’s financial risk. Poor data could lead to inappropriate funding, expensive funding – you could be paying benefits too highly, paying people who have already died. There’s also the cost of fixing that data – if a problem is discovered, then data rectification exercises, in my experience, are extremely expensive and take a long time to play through.
Going back to the liability exercise point that we raised earlier, if a scheme does a buy-in or a buyout transaction, for example, and the insurer deems their conditional data to be of poor quality, they will either ask the scheme to go away and fix it, which takes time, and removes the opportunity to transact; or they’ll add a data risk premium onto the price, so the scheme ends up paying more than they need to in order to make that transaction. So there could be a real financial penalty for not having great conditional data there.
The third reason I highlighted was missed opportunity. So again, using the liability exercise example, if the trustees and the sponsoring employers see an opportunity to transact and see a price they like, if the data’s not right there’s no way they’re going to be able to transact straightaway. Insurers will make them, as I said earlier, clean their data. That may take so long that they may miss the opportunity to transact.
There’s another potentially missed opportunity as well. If, for example, the administrator of that scheme has got some really clever technology, some really engaging technology for members to take advantage of, of self-service, of modelling, of automation of quotes, et cetera, that may be great but if the data’s not there then the trustees can’t take that opportunity to give the members the experience they deserve.
So what should schemes do?
There are three first steps that a trustee or a sponsor can take. The first one is admit there’s a problem, so take it seriously. Whilst they may not be able to allocate a lot of budget and time to it straightaway, at least allocate an element of budget just to do a feasibility or a scoping exercise. So make it a priority and put a plan around it.
The second thing they can do straightaway is to pause and look at what they’ve already got. Most administration systems will have some sort of inbuilt checks around the quality of data, or some reporting mechanism that the trustees can have a look at and see immediately if there are any obvious areas for concern.
The other area is the annual or the triennial actuarial valuation. In my experience, the actuaries will look at the conditional data. They will check whether dates look inconsistent. They will almost certainly play that back to the trustees or the administrator and say ‘these are the things we found in your data’. So the question to ask is: did the scheme actuary play it back to the administrator and then did the administrator fix it? So there are some things that the trustees and sponsors can do there for which they already have the information.
Having done that, the next step is to then decide how they’re going to make the assessment of the remaining gaps. So are they going to ask one of their existing providers, the administrator or the actuary to do some forensic reporting? Are they going to go and buy a solution elsewhere from a third party? There are third party tools out there in the marketplace that will do an assessment of the conditional data.
But I think the main thing is to recognise that they want to do something to develop a strategy and to devote, albeit a little bit of time and money just to do a proper scoping exercise before they launch into it.