1.6. The intended automated compliance regime for PAYE is in abeyance (and has been since 2013) because HMRC is unable to verify whether employer reporting is either timely or accurate so there is no baseline of validated data values for HMRC to ensure employer RTI data error rates improve over time.
1.7. The current rate of RTI ‘LMI’ data error Lord Freud acknowledged to the Committee on 8 February 2017 of ~5% indicates a potentially more significant rate of employer data error. HMRC has never published, or to my knowledge assessed, the employer error rate in PAYE data (at employer scheme level).
1.8. While HMRC processing errors affect only employees tax liabilities and tax coding, not their UC entitlement, errors in HMRC’s own RTI data are frustrating effective and targeted HMRC compliance intervention against the minority of employers that report (~5%) RTI data incorrectly.
1.9. The current low volume of in-work UC claimants is atypical - DWP are unlikely to be able to manage data error at the current rate when UC is fully rolled out unless HMRC implement reliable mechanisms to improve employer reporting error.
2. Technical Overview – Use of RTI Data by HMRC and DWP
2.1. To understand the potential scale and impacts of poor RTI data on UC it is necessary to understand how both departments use of RTI data differs.
2.2. DWP uses RTI data reported by employers in real time and so the earnings data used by DWP contains only employer data error. DWP requests RTI data from HMRC systems for UC claimants in real time, four times a day, which HMRC makes available unprocessed in the identical format as reported by employers. For tax purposes, however, HMRC processes its own separate copy of employers’ RTI data in its departmental systems much later in batches, not in real time. Errors therefore in HMRC’s RTI systems for tax and NI contain both the employers’ data errors – present on submission – and, potentially, HMRC’s own data errors introduced to the individual UC claimant’s tax and NI records as it processes RTI earnings data in its core systems.
2.3. HMRC has acknowledged its processing of RTI data (for tax and NI purposes) is capable of introducing error: it has admitted the problem of duplicate employments, where its systems duplicate employments and earnings. But it is important to point out that additional errors introduced by poor or inaccurate HMRC processing will not impact UC claim calculations directly.
2.4. However, additional error in HMRC’s systems makes it more challenging to identify and target employer error – by compounding issues with the data.
2.5. The greater the additional RTI quality issues experienced by HMRC the harder it is for HMRC to identify the smaller sub-set of employers who are genuinely failing to report
RTI properly. This in turn makes any intervention by HMRC to improve the error rate identified by DWP impossible.
2.6. The scale of error in HMRC’s RTI systems is critically important to the success of UC, albeit indirectly. Without HMRC’s efficient and accurate administration of PAYE it cannot deliver, for DWP, the timely and accurate reporting of RTI by employers. The link between employer reporting failure and proactive compliance and intervention by HMRC (that is correctly targeted) is broken.
2.7. 
The true level of RTI data quality in HMRC’s tax systems appears to be much worse than the level of error acknowledged by Lord Freud. HMRC RTI data processing errors have been widely commented on (ICAEWi, LITRGii & ATT submissions to HMRC’s Post Implementation Review of RTI, of April 2016).
2.8. Because an employer’s RTI return is made by employee, HMRC determines the employer’s liability to PAYE and NI by processing (the total of all) the individual employee pay calculations reported. If there are errors in the employer’s PAYE scheme (business tax) account in HMRC’s systems, as is frequently reported by employers and agents, this means that the personal tax and NI records of the individual employees are also wrong.
3. HMRC Administration of PAYE - Data Issues
3.1. The simplest benchmark of RTI’s success is the number of employer PAYE schemes correctly reconciling (at the year-end) indicating HMRC and employers’ PAYE records match. However, HM Treasury gave a written answer to Stephen Timms MP on 19 December 2016 saying “HM Revenue and Customs does not keep this data.”
3.2. PAYE Regulationsiii were amended in April 2014 creating a discretionary power for HMRC not to take action to enforce collection of employer unpaid PAYE balances owing under £100 per PAYE period, potentially amounting to a cumulative scheme discrepancy of some <£1200 in a fiscal year, which may mask scheme errors.
3.3. From a Freedom of Informationiv answer dated 24 May 2016, sent to me from a payroll bureau, HMRC acknowledge the number of employer PAYE schemes in dispute is some 7.9% but the volume of schemes that do not reconcile may be higher still.
3.4. RTI has not reduced the volume of P800 annual personal tax reconciliations for an under or over-payment of tax since RTI began in April 2013 as confirmed in a written answer to Stephen Timms MP of 6 December 2016.
3.5. The Tax Credits (Income Thresholds and Determination of Rates) (Amendment) Regulations 2016, which came into force 7 March 2016 provide for HMRC to use RTI data automatically to adjust tax credit claims, although denied by HMRCv, may have allowed poor RTI data to have contributed to the volume of tax credit claimant contacts contributing to the (customer service) failure of the HMRC Concentrix contract.
3.6. The error and inaccuracy of data in HMRC’s PAYE systems frustrating an effective HMRC employer RTI compliance regime may be significantly greater than the rate of error DWP currently experiences.
4. Accuracy of In-Work Claimant UC Awards
4.1. The basis of UC is that the value of the award paid to claimants changes automatically as they move in and out of work, topping up their earnings from the workplace in real time. As HMRC explained in October 2014, DWP may use RTI data the same day it is reported by an employer in calculating a claimant’s awardvi.
4.2. On joining UC a claimant is given a monthly claim period when their UC award is paid each month, the date of which is fixed, and which is personal to them. This claim date is not aligned to the claimant’s employer(s) payroll cycle(s). Any earnings from the work place reported by an employer within this personal UC monthly claim period will be taken into account automatically in the calculation of the claimant’s award.
4.3. Because UC has been designed to use employer RTI data in real time, RTI error entering the system puts claimants at risk. Late, missing or incorrect RTI data may all impact the claimant’s award. UC’s design assumes that there will be no error in the earnings data reported by employers, and for this reason there is no mechanism in the design of UC to allow for retrospective adjustment of claims and awards if RTI data is challenged and consequently revised. But as Lord Freud acknowledges, RTI data is not error free. As HMRCvii told the Committee after the Concentrix enquiry it should “shift the focus of compliance activity to … preventing error and fraud at the point of entry to the system.”
4.4. The late reporting of RTI by an employer, or RTI returns with a different pay date to the actual date pay credits the claimant, means the claimant is at risk of either too much money one month and too little the next. This risks undermining claimants’ confidence in the system and putting them at risk of financial difficulty and debt.
4.5. HMRC and DWP’s requirements of RTI data are fundamentally different. Because DWP cannot remedy error after the event, RTI data has to be right for DWP first time, every time. But earnings data is cumulative for tax: HMRC can always revise retrospectively the in-work claimant’s liability to tax to recover underpaid PAYE by changing the claimant’s tax code. Subject to the way in which UC awards take account of the claimant’s historic tax debt, it is possible that DWP will pay higher UC awards to in work claimants because data error in HMRC’s PAYE systems means it previously allocated an incorrect tax code to the claimant. In effect, data errors in HMRC’s management of PAYE would mean the system is ‘robbing Peter to pay Paul’, as DWP is required to pay more to correct for additional tax owed by the claimant. If the tax debt is recovered the difference to the public purse should ultimately be neutral but the claimant’s experience is one of financial uncertainty and stress making the risk of further personal debt greater.
4.6. HMRC does not have any means of correctly identifying employer RTI data error, beyond a failure to submit a return or a return submitted later than a previous return date. HMRC can only rely on the reporting values and dates it expects from employers, based on their historic pattern of reporting, to identify issues. But this is too generalised to be accurate: it cannot determine correctly when an employer actually paid their employees and so when they should have reported RTI. RTI effectively allows the employer to self-certify their compliance with the ‘on or before’ (the date of contractual payment) reporting rule, simply by reporting a different date in RTI to the actual date they credited their employees bank accounts with their net pay. There is no
active compliance regime for RTI, because as HMRC’s Ruth Owen acknowledged in March 2014viii “We haven’t been able to target them [400,000 automated GNS messages] as sharply as we hoped and they went to people who had complied”. The infrastructure HMRC built to enforce its employer RTI compliance regime of automated electronic employer notifications (the Generic Notification System, ‘GNS’) and penalties for late or incorrect RTI reporting and late payment are all in abeyance. HMRC have also extended a 3-day late reporting easement since August 2014 because HMRC don’t have the ability to identify and so respond to employer late reporting failures.
4.7. A 5% rate of RTI data error, as indicated by Lord Freud, equates to 9,000 in-work UC claims a month (of 180,000 claimants in work based on the UC Statistics Data to 9 February 2017 release) affected by or exposed to employer only RTI error.
4.8. Error in HMRC’s own PAYE RTI systems does not impact employers’ RTI data returns used by DWP in UC claim calculations. But it (and HMRC’s inability to validate employer reporting values on entry) does prevent HMRC from identifying actual employer RTI reporting error to allow appropriate compliance interventions to be targeted only at those employers in default of their obligations.
4.9. The current rate of employer RTI error DWP experiences is almost certainly unsustainable at higher claimant volumes. A ~5% error rate with, for example, 4m UC claimants equates to 200,000 claims at risk of incorrect awards each month. The current volume of complex UC household claims is limited and single UC claimants entering work may predominantly be employed by agencies and large national employers whose payroll procedures are likely to be highly accurate. As the volume of UC claimants in work climbs and complex UC claims also increase, exposing UC household claims to the RTI data of smaller SME employers, the exposure of UC claims to employer data error is likely to increase. While Lord Freud acknowledged the ~5%
‘LMI’ error rate, in his evidence to the Committee of 8 February 2017, DWP’s written answers to Stephen Timms MPix appear to indicate no significant issues with RTI data.
This may suggest DWP is re-processing employer RTI data to improve its quality, by comparing the in period earnings value with the cumulative in year total, also present, as well as using the DWP BACS hash validation process to correct employer RTI data error. If DWP have developed additional or alternative processing of employer data in its own systems, to manage the current rate of employer error, it is unclear if retrospective real time data processing by DWP can work at scale once UC is fully deployed.
5. Conclusions
5.1. There is significant complexity in the way DWP and HMRC’s use of RTI data differs; the impacts of employer error on DWP’s UC award calculations; and the consequences of HMRC’s passive management of RTI data in its own systems. But for UC claimants in work their experience of UC ultimately depends not on DWP but indirectly on HMRC’s accurate and efficient management of RTI and its inability proactively to identify and target accurately those employers reporting RTI late or incorrectly. DWP RTI data is not
sufficiently accurate and does affect claimants’ UC award calculations, as oral evidence heard by the Committee on 23 January 2017x from London Borough of Croydon’s Welfare Director and the Child Poverty Action Group has confirmed. HMRC’s failure to manage employer RTI issues efficiently frustrates effective compliance against the minority of employers who do not comply with their RTI reporting obligations.
5.2. The assumption of no RTI data error in UC’s design makes its success contingent on the quality of RTI data reported by all employers. Should HMRC remain unable or unwilling to achieve greater accuracy of RTI data in its own systems (for tax purposes) to allow it correctly to target non-compliant employers, employer RTI data quality cannot be improved for DWP.
5.3. HMRC’s RTI compliance regime is in abeyance because it lacks the means of correctly identifying those employers that do not report RTI accurately from those that are, so there is currently no possibility of RTI data accuracy improving. Making Tax Digital has potentially further delayed the introduction of an effective compliance regime until April 2018 at the earliest when HMRC intends to introduce a new compliance regime, even though MTD further increases the urgency of RTI data improvement.
5.4. HMRC’s intention to minimise the administrative burden of RTI on employers comes at the cost of saddling UC claimants and DWP with an unworkable level of data error. This approach is self-defeating as employers currently have the worst of both worlds as HMRC’s failure to engage with RTI data issues in its core systems exposes employers to challenge and recourse from HMRC (because HMRC believes erroneously the employer’s PAYE account is not up to date) and employees who dispute and challenge their UC award calculation.
5.5. The overwhelming majority of employers comply with their RTI reporting obligations as PAYE regulations require, but HMRC’s own processing errors (for tax not UC) contribute to the level of query and challenge they encounter from HMRC and increasingly employees, undermining trust and confidence. HMRC is unable to identify with any accuracy that small (but systemically important to UC) subset of employers that are not correctly reporting RTI or doing so ‘on or before’ they actually pay their staff.
5.6. Unless HMRC can deploy an equitable compliance strategy capable of correctly identifying employer reporting errors, the success of UC will always be at risk from an unmanageable volume of incorrect UC award payments, breaking UC’s commitment the claimant will always be better off in work. UC’s success is contingent on HMRC’s
willingness and capability to deliver a level of accuracy and efficiency in its administration of PAYE it has not previously attained.
5.7. Disappointingly there has been no willingness from HMRC to engage with employer organisations or payroll software suppliers, contrary to the assurances given by DWP, to improve the system since HMRC disbanded its own RTI User Group in July 2013. Despite its requests for input from external stakeholders to its mandatory internal Post Implementation Review of RTI, HMRC has so far not published its findings.
6. Recommendations
6.1. Ask DWP how it manages RTI data errors for claimants in work and what, if any, additional measures it has deployed that explain how it manages a ~5% rate of employer RTI data error and how will this, or greater rates of error, be sustained at higher claimant volumes in future.
6.2. Draw the Committee’s attention to HMRC’s original 2010 PAYE consultation document Improving the operation of Pay As You Earn (PAYE): Collecting Real Time Information where HMRC and Ministers made clear “Consideration will [only] be given to further developments of PAYE once RTI has bedded in and been evaluated” if it did not deliver the necessary improvement in RTI data quality.
6.3. Invite HMRC to the Committee to explain the original RTI compliance regime it implemented and what actions it took when it suspended its automated RTI compliance systems in 2014 to ensure RTI data accuracy could be improved.
6.4. Request HMRC set out in detail an effective employer compliance regime that avoids all unnecessary contacts with employers and inaccurate targeting of employers to ensure only employers with proven breaches of their obligations are correctly targeted and sanctioned.
6.5. Hold HMRC to account for the success of its RTI compliance regime and the quality of RTI data presented by employers.
6.6. Recommend that the Committee takes account of the total dependency of UC’s design on HMRC’s administration and management of PAYE and ask HMRC and DWP to explain how the burden of overpaid UC (or historic tax credit or tax debt,) arising from the claimant’s underpayment of tax due to HMRC’s operation of the incorrect tax code for the claimant, is managed.
March 2017
References & Notes
i ICAEW additional comment: https://ion.icaew.com/taxfaculty/b/weblog/posts/paye-in-real-time-post- implementation-review
ii LITRG additional comment: http://www.litrg.org.uk/latest-news/submissions/160426-rti-post- implementation-review
iii “New regulation 67L makes provision for a tolerance of £100 between the total of all the amounts that the employer is due to pay to HMRC for the tax month”
iv Freedom of Information Act request to HMRC, May 2016, from Mr M Portt, submitted a as follows:
“How many instances have HMRC been asked to deal with where the employer disputes the PAYE/NI charge and your HMRC software gives another.”
The request was answered on 24 May 2016 (by Jan Beckett, HMRC Audit & Briefing Adviser, 2nd Floor, Parkside Court, Hall Park Way, Telford FT3 4LR):
“I am answering under the terms of the FOIA and can advise that we currently have in excess of 2m employers using Real Time Information (RTI). In the three years since its inception in 2013-14, 97,648 (less than 1.5%) employers have queried discrepancies between the charge calculated by their payroll software and that recorded on HMRC’s RTI system.”
HMRC’s definition of “employers using RTI” is all businesses registered for PAYE operating RTI including ‘those with no employees where a self-employed owner-manager(s) or company with one employee, assumed to be an employee director’ according to the ONS. HMRC’s percentage of less than 1.5% is not explained and is presented as a cumulative not annualised rate of data error, but PAYE scheme reconciliation errors are carried forward into every subsequent fiscal year unless corrected.
Single employee PAYE schemes typically use HMRC’s free Basic Payroll Tools RTI software and operate a quarterly or annual PAYE cycle to draw income from their businesses: their simplicity, lower volume and reporting frequency expose them to a significantly lower level of data error.
ONS data sets for early 2016 show 1,229,330 employers (public and private) in the whole economy employing 2 or more employees. See https://www.gov.uk/government/statistics/business-population-estimates-2016 https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/559220/bpe_2016_detailed
_tables.xls Table 2: 97,648 PAYE scheme discrepancies amounts to a total error rate of 7.9% of all UK employers with 2 or more employees, if one-employee ‘employers’ are excluded.
v Financial Times, 7 November 2016: comment to FT “[HMRC] denied that RTI played a role in the incorrect withdrawal of tax credits from hundreds of people by Concentrix … RTI did not play any part in the Concentrix mistakes.”
vi HMRC Employer Bulletin 50 October 2014: “Universal Credit claim dates are personal to claimants. So payments you make today, could be used in a Universal Credit assessment today.” p.13
DWP: Universal Credit and employers: frequently asked questions, March 2015 p.4 [polling HMRC systems 4 times daily]
vii HMRC’s intention to "shift the focus of compliance activity to a greater emphasis on education and preventing error and fraud at the point of entry to the system" https://www.parliament.uk/business/committees/committees-a-z/commons-select/work-and-pensions- committee/news-parliament-2015/concentrix-government-response-16-17/
viii AccountingWEB 3 March 2014 http://www.accountingweb.co.uk/tech/tech-pulse/hmrc-clarifies-generic- notices-situation
ix Universal Credit: PAYE: Stephen Timms MP, Written questions to DWP 52312 52527 & 55898
x Oral Evidence to Work & Pensions Committee – 23 January 2017 Q68- Q70 Universal Credit Update
