My Lords, I rise to speak to Amendments 54 and 65, both of which are on the same topic. I beg the Committee’s indulgence. This is such an important issue that I want to expand on some of the areas involved and my reasons for tabling these amendments.
Accurate and complete member data is surely an essential prerequisite for the success of any pensions dashboard. I was struck by the Minister remarking that pension information is the lifeblood of the dashboard, which is absolutely right. These are probing amendments; I do not claim that they are the perfect answer to the issues that I am raising, but I have tried to insert into the Bill specific requirements that must be imposed on trustees, managers and administrators of occupational pension schemes to ensure that the information submitted to the dashboard has been checked regularly for accuracy. I have suggested that errors must be corrected within six months. That may not be a reasonable timeframe, but it is a start. That is Amendment 54.
Amendment 65 seeks to do the same kind of thing for personal and stakeholder pensions which Amendment 54 is seeking to do for occupational schemes. I am not sure whether I need to mention this each time I speak in Committee, but I draw noble Lords’ attention to my interests as set out in the register. Auto-enrolment has been a great success as all UK employers have set up pension schemes for their staff. Workers will be building towards a better retirement, which is a force for good, but it cannot be right that there are currently no formalised requirements that data records are verified as accurate regularly.
In the past, pensions have been plagued by data problems. Recently a number of pensioners have had to repay some of their pensions and face future pension cuts as they have been told that past errors in their pension entitlement have been discovered, many decades later in some cases. Records were not regularly updated or corrected. In the past there was manual record-keeping, which was prone to human error, and failure to ensure robust data reconciliation had been regularly carried out meant that errors were not discovered promptly, and they persisted over time without people knowing.
For any dashboard initiative to work, consumers have to be able to trust that their pension contribution records are accurate. This is a particular problem because the complexity of pension rules makes it almost impossible for individuals, especially workers enrolled in auto-enrolment schemes, to know whether the amounts being paid in on their behalf are correct. The complex calculations must calculate the employee contribution, the employer contribution, the tax relief, and potentially the national insurance relief as well. The member would naturally assume that their employer or their pension provider was ensuring that the amounts being recorded on their behalf were accurate, but unfortunately this has not been the case in the past and it is still not the case for new pension schemes. For example, a study I was involved in last year which analysed data representing more than 1 million contributions from more than 100,000 schemes—these were small employers —showed that the data had a 50% initial error rate. Some 50% of some aspects of the information was incorrect and had to be sent back for correction. Those error rates did not persist, but the data was not necessarily checked as thoroughly as it could have been.
Pension administration is the Cinderella of pensions. It is the low-margin end. It is not the sexy end. It is under cost pressure, and administrators seem to have been expected to absorb often very complex changes. Sometimes pension providers change their data requirements and their payroll software is not updated to reflect the latest version, so administrators then manually adjust spreadsheets to try to make sure that they have some data recorded. Data includes incorrect contribution amounts, contributions made for workers who did not belong to the scheme or who had already opted out, wrong identifiers for the pension scheme, inaccurate postcodes, incorrect pay period dates and so on and, for example, incorrectly believing that a pension scheme operates on a relief at source basis when it is net pay or the other way around so the amounts are simply not right.
Unfortunately there are no regulatory checks to ensure that data is verified for accuracy. What we have seen in legacy schemes is the detriment that this can cause to pensioner members, and if we have a pensions dashboard that people are relying upon to make their retirement plans, it is not good enough that administrators will just to try to make sure that by the time people reach retirement and get their pension all the errors are corrected because people will need that all along. For example, the auto-enrolment declaration of compliance does not have accuracy checks built in. Employers are asked to confirm that they have paid the right amount but nobody ensures that that is the case.
If they want to check, many pension providers currently do not collect the information that they need to verify because they are not getting the pensionable pay data sent over to them; they just get an amount of money and are told that it is correct, and that is that. We are in the middle of pensions master trust authorisation. Again, there is a risk of records being incorrect but the authorisation does not entail robust checks on data accuracy or proof that proper processes are in place to discover and correct errors.
I was trying to put into the Bill a mechanism whereby we can draw a line at a point in time and make sure that the pensions dashboard data has been through a process of cleansing and verification as a requirement for submitting the data. I am not saying that this will be simple or easy, but as more schemes emerge it will be more difficult to go back and try to reconcile past records. We have an opportunity now to put that sort of requirement into the Bill.
I quote from the Pensions Administration Standards Association, a body that oversees pensions administration and has been directly involved in some of these areas:
“Data cleansing is costly, so in low margin operations there is little appetite to invest in either clean data or in digitisation which depends on the quality of data. There is no incentive to do better than your competitor, as you are all in the same boat. Customers do not demand improvements and where they do trustees choose to ignore the calls either because of cost, resource constraints or other priorities for the scheme.”
It goes on:
“There is an expectation that introducing mandatory data provision as part of the dashboard project will act as an incentive to schemes and providers to clean up data that has been in a poor state for decades … The uncomfortable truth is that while compulsion will encourage some clean up, it will only be to the minimum level needed to show some data in a field, which essentially means that the presence of a data item will take precedence over the accuracy of it. Schemes already report 90% compliance with common data standards set by the Pensions Regulator. This should mean 90% of schemes will be able to present data that identifies an individual, but of course we know this is not reality, because it has been self-reported and not robustly checked.”
We have an opportunity to recognise the poor quality of data. This is not a blame game; it is about trying to put into the legislation a mechanism through which providers and everyone involved in the dashboard know that they can no longer rely on other people not correcting their data and no longer not attend to this themselves.
Of course, it will never be possible to ensure 100% accuracy, but having processes in place that constantly check and which allow errors to be corrected promptly is urgently required. Random regulatory checks, mystery shopping and systematic accuracy verification by an independent body would be of value and is surely a vital ingredient of any dashboard on which consumers are expected to rely.
As I said, I am not suggesting that the wording of the amendments is appropriate, in the right place or expressed correctly, but I hope that my noble friend the Minister can give us some information on and consideration of whether this could be built into a dashboard requirement. I beg to move.
6.15 pm