Audit report uncovers PAYE coding fiascoby
The scale of HMRC’s difficulties in introducing a new system to manage PAYE codes has emerged in an appendix to the department’s 2009-10 accounts.
A report prepared by the Comptroller and Auditor General to accompany the annual HMRC accounts (2.7MB PDF) devotes several pages to the problem, and estimates that if 18.2m unresolved “open” cases are taken into account, a net figure of £1.6bn in repayments may still be outstanding going back to 2007-08 and preceding tax years.
There has been some controversy in recent weeks over the decision not to publish a full annual report for HMRC. In the government’s view, dropping the annual accounts would save money and was not necessary because many of the targets on which they reported related to initiatives that were no longer relevant. The CIOT’s John Whiting, Grant Thornton’s Mike Warburton and PKF’s John Cassidy all complained to The Observer about important information for departmental transparency and for tax advisers that would no longer be available.
As in previous years, the NAO has qualified its audit report with regard to Tax Credit errors and fraud. In 2009/10 the department paid a net £27.3bn in Child and Working Tax Credits, and estimates that error and fraud resulted in incorrect payments of up to £2.27bn, 8.3% of the total.
But for long term students of HMRC’s IT disasters, the Auditor General’s report gives an insight into the problems that continue to plague the department’s computing infrastructure. The problems with the PAYE system have been evident for years, and paved the way for a major overhaul to standardise numerous separate systems into a single database for employee records.
The National Insurance and PAYE Service (NPS) went live in a phased implementation that started in June 2009, more than two years’ behind schedule. The decision to delay the project was made to protect the integrity of the PAYE system, but built up a huge backlog of PAYE returns in 2007-08 and 08-09.
The deferral cost the department an extra £33m and prevented it from realising some £55m of planned efficiency savings during 2008-09 and 2009-10. Further changes, including improved security measures, added a further £78m to the cost of the project. Based on its revised figures, HMRC expects to save around £532m over five years, and to recover its investment of £389m by 2013.
In spite of introducing the new system, PAYE processing errors increased in 2009-10, amounting to £132m in underpayments (a 15.7% increase) and £238m in overpayments (a 148% increase). The department discovered that going live with a high volume computer system that wasn’t adequately scoped or tested was a sure way to foul things up even more.
Before migrating 54.3m live taxpayer records from 12 regional databases into 45.4m employment records on the new database in June 2009, HMRC neglected to validate the integrity of the information, assuming that its new exception processes would be able to correct inaccuracies as the system received new data from employers.
“Eevidence from the initial operation of the new Service suggests that the Department did not fully appreciate the extent of risk from data inaccuracies or its implications for the delivery of PAYE,” observed the NAO report.
The capability to reconcile many of these discrepancies within the new system was not available until the third phase of the project in April 2010, which meant that more than 7m over- and underpayments were unreconciled when it came to run the 2010-11 annual coding exercise. These records will now be processed from August 2010, although it is not yet clear how many cases will clear automatically and how many will be left for manual working.
When the new system encountered an employee that it could not match to existing employment data, it automatically generated a new, erroneous employment record. So many new items were appearing that when HMRC processed 2008-09 data on the system, it exceeded its 12.5m capacity for open items. The 7m work items arising from 2008-09 returns had to be removed from the workflow queue rescheduled for processing in August 2010.
“The Department found that the new service had produced significantly more amended tax codes than expected, with the potential to generate up to 25.8 million coding notices, almost double the amount anticipated. A significant number of the codes generated were incorrect. “
While its computer was churning out coding notices, HMRC only really became aware of the problem as the volume of calls mounted to its contact centres – which peaked at more than 18,000 calls per day about erroneous coding notices.
In January 2010 HMRC put in place a recovery programme to identify and prevent the issue of any further incorrect coding notices to individuals and ensure that corrected coding notices were issued to employers in time for the 2010-11 tax year. At its peak, the department assigned a total of 3,000 staff to the project, including 2,400 who were switched from their usual work supporting personal tax customers.
At the end of June 2010, estimated a further 2m records were at risk and needed review and will begin bulk processing of 2008-09 reconciliation this month. The department is also undertaking a review of the annual coding exercise to identify lessons learned and further actions not addressed by its recovery programme, the NAO reported.
Rob Durrant-Walker, tax manager at UHY Hacker Young, commented: “Even though HMRC’s new system should lead to more accurate codes in future that can cope with increased job mobility, these figures are quite shocking. We have seen a steady increase in PAYE errors over the past year. For the amount of tax collected in error through PAYE to jump 148% in one year is simply unacceptable.”
A spokesman for HMRC apologised for the errors and acknowledged the processing of PAYE returns and coding notices was “not acceptable”.
“The new system raises the bar in terms of data quality and will in the medium term significantly improve overall accuracy, reducing both under and overpayments,” he added. “Our contact centres are able to quickly correct inaccuracies, when contacted by the taxpayer, in part because the new system has for the first time created a single taxpayer record which the contact centre operator can access and amend.”
But all the anguish and half-billion pound costs of the PAYE system upgrade could be rendered redundant within a few years, if the new government’s proposals for reform are put into practice. Rather than trying to reconcile its data with that supplied by employers in their annual returns, HMRC is looking to move to real time information, with the added possibility of taking on responsibility for calculating and deducting tax due centrally in a second phase.
Given HMRC’s lamentable track record on recent PAYE computerisation projects (remember NIRS?), many practitioners have called into question the department’s ability to handle centralised deductions. But perhaps the five-year catalogue of woe compiled by the NAO encouraged ministers to look at the alternatives.
You might also be interested in
AccountingWEB’s interim Editor in Chief has been with the site since 1999 and returned to the editorial hot seat in March 2020 to lead the hunt for a long-term successor... Send a DM if you're interested! When not tending to the needs of AccountingWEB members and geeking out on their technology habits, he devotes much of his time to his oddball...