Pension data accuracy

25 October 2018

This article was featured in the November 2018 issue of the magazine.

A roundtable was held on 13 September in Committee room 2A House of Lords. Mike Nicholas reports

The roundtable, which was hosted by Baroness Ros Altmann on behalf of the CIPP, pensionsync and Reward Strategy, was attended by an extensive list of individuals representing the payroll and pension industries. A reception event held after the roundtable evening brought more luminaries.

Ros opened the roundtable explaining the concerns about pension data accuracy and how the pensions of millions of savers could be affected by errors and omissions in the data. Over time, as employers cease, schemes windup, people change pension funds creating multiple pension pots, and master trusts consolidate, lack of data accuracy will become critical unless action is taken soon. The Department for Work and Pensions estimates that by 2050 the number of abandoned pension pots could reach fifty million.

Ken Pullar FCIPP (CIPP chief executive officer (CEO)), Jill Smith MCIPPdip (CIPP policy manager), Neil Esslemont (The Pensions Regulator), and Will Lovegrove (pensionsync CEO), all addressed the meeting commenting on the issues. 

Attendees were then invited to join the discussion focusing on the causes of inaccurate automatic enrolment (AE) data submissions, the impact on the pensions market and setting out ways to improve data reliability and security. 

In her closing remarks Ros confirmed that further discussion will ensue to develop and carry forward measures to address data accuracy issues.


Discussion issues

The requirement to send or refer data back to its source implies that the administration of workplace pensions is inefficient, resulting in increased costs for employers, payroll firms, book-keepers, other third-party providers and also for pension providers in terms of customer support. With high error rates commonplace, the costs of pension administration and the time taken to submit the information to pension providers is much greater than necessary. 

It is also worrying that some administrators ‘fudge’ pension information by editing it within spreadsheet or word processing software to get the job done rather than addressing the root causes of the data errors within the source (typically payroll) software. In situations where payroll software is not updated to reflect changes made by pension providers to data requirements, administrators may have little choice but to adjust the data in a spreadsheet or key data directly into a pension provider’s web portal.

The AE declaration of compliance does not require confirmation that the pension contributions and employee pension records have been robustly verified as accurate, with little if any checking of member data accuracy by most providers. Indeed, many providers could not even check whether contributions are correct if they wanted to, because they do not collect the relevant information on employee pay.

With some pension master trusts in the process of consolidation, the risks of pension records already being incorrect are worrying. Master trust authorisation did not require robust checks on data accuracy or proof that there are proper processes to discover and correct errors. It is essential that processes are urgently introduced to regularly check pension data are accurate and reliable because over time and if schemes merge it will be much more difficult to reconcile past records. 

Though it will never be possible to ensure 100% accuracy, having automation processes in place which constantly check and allow errors to be corrected promptly, will bring pensions administration into the 21st century. This can also ensure greater reliability, security and lower costs.


Research findings

Attendees were given copies of the research by pensionsync which exposes high error rates in AE data for small firms. The legislation only checks employers are paying into a scheme, not the accuracy of information held which begs the question whether pension records are already inaccurate under AE. 

Though employees expect their employer and/or pension provider to ensure accuracy in the amounts being recorded, there are continuing failings even with new pension schemes.

Analysis of data representing contributions to over 10,000 schemes reveals that data sent on behalf of employers to pension providers have a high error rate causing returns to employers for correction. Such rates suggest the need for greater attention to be paid to data accuracy by pension administrators and/or employers themselves. (See Success rates and errors, below.)

Among the common errors found by pensionsync are: contribution amounts being too high or too low or being made for workers who do not belong to the scheme or have opted out; and incorrect pay period dates. Other examples of pension administration errors stem from employers or their agents incorrectly understanding or believing a pension scheme operates tax relief-at-source rather than the net pay arrangement, and vice versa, and so confuse gross and net figures.


CIPP survey

Jill Smith MCIPPdip, CIPP policy manager, provided details of a survey the Institute had recently conducted ahead of the roundtable. The CIPP gathered the thoughts, views and experiences of our members and also the wider payroll profession by asking the following questions:

  • What problems are you encountering on a day to day basis with the operation and managing of an accurate data flow between payroll and your pension provider?

  • Are you experiencing any recurring issues that are due to inaccurate AE data submissions? Does this raise concerns for you, employers and employees?

  • If you could create a wish list of improvements to workplace pensions what items would you like?

The results of the survey support the statements that AE is cumbersome and a burden due to complexity, too much choice when choosing a pension, and nowhere to go for clear advice. Payrollers would benefit from standardisation in all aspects of admin processes, terminology, communication, and implementation of AE by pension providers.

The following are examples of areas for improvement: processing opt outs, reconciling pay periods, successfully uploading accurate contributions to providers. It would be good to have a standard handover procedure so that when a client leaves one payroll provider the new provider has instant access to data to be able to process the payroll.

The common theme was very much around accuracy and consistency of communications.

In her final comment to the roundtable, Jill observed that the “meeting has identified the problems of interfacing between payroll and pension provider of inaccurate data but we have only touched the tip of the iceberg, there is a lot more to explore. Also, though we have concentrated on the smaller employer the problems are happening in the larger organisations too so they not specific to the smaller employer only. It would all need to be looked at.”


Success rates and errors

  •  The table shows how the success rate of data submissions to pensionsync increases month on month, as users improve the quality of data held in payroll software (based on feedback from the pensionsync platform and staff). The data was collected between August 2017 and July 2018, for all data submissions made to pensionsync in month one, and where users continually submitted data each month for the next eleven months. The sample data set comprised of 11,177 pension schemes, submitted by 5,931 users of 13 software providers.

  • The chart show the most common errors recorded by pensionsync using data submitted by 291 users of nine software providers for 581 pension schemes. The volume of errors decrease over time as pensionsync interacts with users. (The ‘Contribution amounts’ reflect only the accuracy for two master trusts.)