House of Commons - Treasury

Select Committee on Treasury Sixth Report

2 Overpayments of tax credits

9. Except where the context otherwise requires, when we refer to "overpayments" in this report, we mean both excess payments and overpayments:

an "overpayment" is an amount owed to HMRC after the end of the tax year, which HMRC ordinarily recovers in the following tax year;

an "excess payment" is an in-year reduction which HMRC makes to tax credit payments, with the aim of avoiding an overpayment at the end of the tax year.

Level of overpayments

10. The Government set out its proposals for the new tax credits in a paper published alongside Budget 2002, The Child and Working Tax Credits: The Modernisation of Britain's Tax and Benefit System.[9] In this paper, the Government explained that the new tax credits regime had been "designed to minimise the scope for substantial overpayments", before setting out its estimates of the number of families whose awards might need to be reassessed as a result of rises or falls in income.[10] The Government was "reasonably confident" that, "in steady state":[11]

around 1 million cases would see their awards change as a result of a fall in income;[12]
around 750,000 individuals and couples a year would see their awards change as a result of a rise in income;[13] and
exceptionally, for the 2003-04 financial year, around 1 million awards would be reassessed as a result of rises in income, as tax credit awards for the year beginning in April 2003 would be based upon income information for the tax year 2001-02, not 2002-03.[14]

The paper stated that those people who experienced a rise in income "would be advised to ask for an adjusted award during the year to reduce the risk of overpayments".[15]

11. Data released by HMRC shows that the level of overpayments for 2003-04 was significantly higher than that predicted by the Government in April 2002. About one-third of all awards paid—nearly 1.9 million awards—were overpaid, at a cost of nearly £2 billion. In December 2005, the Paymaster General indicated that, but for the reforms announced to the scheme in the Pre-Budget Report 2005, "initial estimates" had suggested that subsequent years' overpayments would have been "of broadly the same level as in 2003-04".[16] The tables below summarise numbers of awards made in 2003-04, and the amount of money paid out in tax credits awards in 2003-04. HMRC hopes that equivalent data for 2004-05 will be available shortly.[17]

Table 1: Tax credits awards, 2003-04

Numbers of awards

Awards underpaid 713,000

Awards neither under nor overpaid 3.078 million

Awards overpaid 1.879 million

Total 5.670 million

Source: HMRC, Child and Working Tax Credits Statistics, Finalised awards 2003-04, table 1

Of the 1.879 million claimants who received overpayments, about 41,000 received overpayments of £5,000 or more.[18] Half of the total overpayments related to some 283,000 families who had been overpaid by £2,000 or more.[19]

Table 2: Tax credits awards payments, 2003-04 (£ million)

Entitlement Net paid Net overpayment
at 5 April 2004

Awards underpaid 2,191 1,727 -464

Awards neither under nor overpaid 5,332 5,332 0

Awards overpaid 4,539 6,470 1,931

Totals 12,062 13,529 1,468

Source: HMRC, Child and Working Tax Credits Statistics, Finalised awards 2003-04, table 1

12. Evidence received from the National Audit Office discussed the reasons for overpayments, and explained why the design of tax credits necessarily results in overpayments:

A tax credit award is provisionally based on a family's income and circumstances from the preceding tax year. The award is finalised after the end of the tax year once income and circumstances are known for certain. The final award will be lower than the provisional award where incomes increase, although the first £2,500 of any income increase is disregarded.[20]

However, the NAO also states that "further unforeseen overpayments" have occurred.[21] Full recovery of overpayments from 2003-04 is expected to take at least 5 years.[22] To date, the Government has written off some £95 million of overpayments and has made provision for a further £961 million to eventually be written off.[23]

EXTENT OF INCOME VOLATILITY AMONGST LOW-INCOME FAMILIES

13. We explored the question of why the actual level of overpayments had been so much higher than would have been expected, based on the Government's predictions about the number of claimants who would experience a rise in income.

14. Mike Brewer, Programme Director of the Direct Tax and Welfare sector at the Institute for Fiscal Studies, commented that:

In the Revenue's defence … there was not a good source of data for them to design the initial [tax credits] policy, when they were thinking about this in 2001-02. Basically, that remains the case even now, except the Revenue now have data from one or two years of operation of tax credits.[24]

He explained that "large household surveys", used by researchers such as himself, did not generally collect information about respondents' annual income, as opposed to weekly or monthly income, and that therefore "there just was not a good source of data out there about families' annual income and how it changes over time".[25]

15. Rt Hon Frank Field MP referred to the Paymaster General's statement that the recent package of reforms to tax credits, announced in the Pre-Budget Report 2005, should "reduce the value of overpayments by around one third".[26] Mr Field observed that, given that the increase in the disregard threshold to £25,000 should mean that overpayments due to income changes virtually cease, this suggested that "it is household composition changes which are the basis of the [remaining] two-thirds of overpayments".[27] He commented that, if his suggestion was correct, it said "something extraordinary about these changing circumstances of lots of our constituents at the bottom end of the income scale".[28]

16. This lack of information about the financial circumstances of lower-income families appears to be a continuing problem for HMRC. For example, in October 2005, the Paymaster General told the House that information about the number of changes of circumstances relating to tax credits entitlement reported to HMRC in 2003-04 and 2004-05 was not available.[29] What evidence there is suggests that this number is likely to be high. For example, research carried out on behalf of OPF between May and July 2004, based on a sample of 100 lone parents receiving tax credits, found that, in 2003-04, nearly half (47%) had experienced between two and seven changes of circumstance.[30]

17. Some indication of the extent to which some tax credits claimants' circumstances are subject to frequent change is given by a recent study, the initial results of which were reported in March 2006 by the Economic and Social Research Council Research Centre for Analysis of Social Exclusion in Tracking income: how working families' incomes vary through the year.[31] The study was financed by HM Treasury and the Inland Revenue (now HMRC) and aimed: "to reveal the range of income patterns across a whole year for a group of particular policy interest, low- to middle-income working families with children", information which "has not been collected in the UK before".[32] Its results were based on data for the financial year 2003-04, for a total of 4,800 weeks of income from 93 families, all of whom were receiving the Working Families Tax Credit in the winter of 2002-03. The average total net family income for the 93 families was £17,000, and most of the families had total net incomes in the range between £12,000 and £22,000.[33] The report's authors described it as shedding light on three key issues:

how the distribution of incomes across a whole year compares with those measured over a short period;
patterns of income mobility at a finer level than observed before; and
the extent to which state transfers (social security benefits and tax credits) smooth incomes over the year.[34]

18. The authors concluded that "patterns of income mobility … involve considerably greater volatility of income within the year (for this particular kind of working family) than many might have expected".[35] For instance:

only seven of the 93 cases had incomes in the 13 periods that varied within a range of +/- 10% of the case's annual average;
a quarter of the families had at least four periods with incomes outside a range of 85% to 115% of their annual average;
generally speaking, those families with the greatest volatility of income were those with lower incomes (total net incomes for the year below £15,000), and a higher proportion of lone parents and tenants had more variable income;
high degrees of variation affected some cases from all of the family types studied, and some of the families had patterns of income receipt that were very variable indeed.[36]

19. Of particular relevance are the report's findings on the extent to which benefits and tax credits "smoothed" the net incomes of the families studied. The study found that both benefits and tax credits reduced inequality between the total net incomes of the 93 cases, and did so to the same degree. However:

while both social security benefits and tax credits reduced the variability of individual families' incomes within the year, benefits did so to a greater extent than tax credits, even though the amounts of benefits involved were less than half the amount on average than that of tax credits. In nearly a third of cases income was more variable after including tax credits than before doing so …[37]

20. The report's authors suggest that the difference between the income-smoothing effects of benefits and tax credits lies in the different design of each: "some social security benefits are based on circumstances over short periods with benefits adjusted immediately, and pound for pound. Tax credits are generally intended to reflect the position over the year as a whole, with their payments adjusted to achieve this for the year as a whole, not in any particular week or month."[38] They conclude:

It might be argued that it is achieving the correct position over the year as whole by its end that matters—which is what the new tax credit and older PAYE systems are designed to do—rather than income smoothing within the year. However, our interviews suggest that families with incomes at these levels budget on a much shorter-term basis than over the whole of a tax year—over a month or less—so such adjustments may come too late for them. Given the generosity of the new tax credit system, making up more than a quarter of the sample families' total net incomes, the ways in which credits are paid obviously have major effects on their income flows through the year, and their design has to be carefully considered and monitored in the light of findings of the kind reported here.[39]

OUR CONCLUSIONS

21. Recent research published by the Economic and Social Research Council Research Centre for Analysis of Social Exclusion indicates that the month-to-month incomes of low- to middle-income working families with children are considerably more volatile than might have been expected. We welcome the fact that such research has been undertaken: it offers valuable data to those involved in designing tax credits policy. We are pleased to see the Government funding such a study, which seems to us to offer precisely the sort of information the Government needs to enable it to provide State assistance in a way which fits with the income patterns of those targeted. We recommend that the Government direct additional resources to funding research likely to inform the formulation of tax credits policy.

22. We are particularly struck by the Research Centre's findings that such families tend to budget over a month or less, rather than over the whole of a tax year, and that, in nearly a third of cases studies, income was more variable after including tax credits than before doing so. These findings suggest that the tax credits scheme, which is designed to deliver the correct amount of state assistance over the year as a whole, rather than over any shorter period, could be aligned more closely to the financial needs of such families. End-of-year adjustments in tax credit entitlement may come too late for such families, and any demand for reimbursement is felt very keenly by them. We suggest that there is evidence that determining awards over shorter time periods would reflect the needs of lower income families more accurately than annual awards.

Causes of overpayments

23. We sought to establish how much HMRC knows about why individual overpayments have arisen. In April 2005, HMRC acknowledged, in a letter to the Ombudsman, that it could not "easily identify the reasons why an overpayment arose".[40] HMRC indicated that it did not intend to examine each individual award because this "would be prohibitively expensive in terms of resources", and concluded that, as a result, it did "not yet have sufficient data to allow [it] to publish information on the causes of overpayments."[41]

HMRC'S ACCOUNT OF CAUSES OF OVERPAYMENTS

24. In December 2005, the Paymaster General issued what seems to be the Government's most complete statement yet about the causes of overpayments. The Paymaster General told the House:

Analysis of overpayments suggests that they result from a number of factors: income rises from one year to the next; families overestimating the extent to which their income has fallen when they seek extra support during the year; provisional payments made at the start of the tax year, which are based on out-of-date information that is subsequently updated when the award is renewed; and delays in reporting changes in families' personal circumstances to HMRC.[42]

25. During our visit to the Tax Credits Office (TCO) in Preston, we were told by TCO management that the 'top three' reasons for tax credits overpayments were the first three of the four factors listed by the Paymaster General, above. TCO management estimated that these three factors accounted for about 70% of overpayments, and that about 40% of overpayments were attributable to the first reason alone. We subsequently took oral evidence from HMRC officials who repeated these three factors as the 'top three' reasons for overpayments. Officials noted that these three factors:

will not always operate in isolation and you will quite often have two or more of them operating in combination. Our best estimate … is that if you take those three factors together they probably account for about two-thirds of total overpayments. But it is very difficult, with the best will in the world, to actually precisely identify the component that each of them makes because they are not always in unique operation.[43]

26. Shortly before we agreed this report, HMRC told us that, from the data available to it, it was not possible "to produce reliable estimates showing the relative importance of these factors in explaining overpayments".[44] The Department explained that this was because there are "significant interactions" between the different factors: for example, a family could have a rise in income of more than £2,500, as well as a change of circumstances that reduced entitlement, and both factors could contribute to an overpayment.[45]

Our conclusions

27. From the above figures it follows that about 30% of all overpayments are due to delays in reporting changes in families' personal circumstances, such as family breakdown or the establishment of new partnerships, or a child going to school or leaving home. However it is clear from the oral evidence from HMRC officials that the focus is on the other three categories of overpayment. We believe that enough attention has not been paid to the problems caused by families' changing circumstances and the difficulty of adjusting tax credits to reflect these. We therefore recommend that more research is undertaken into this, especially in view of impending changes in reporting requirements.

28. The factors cited by the Paymaster General and her officials as contributing to the problem of overpayments do not appear to us to give a comprehensive account of the reasons which overpayments have arisen. While we are not in a position to draw up our own comprehensive list of factors resulting in overpayments, it is obvious to us that the Paymaster General's account makes no reference to causes of overpayments which have arisen as a consequence of the Department's own processes—for example, official error and IT system error. Rather, the Paymaster General has referred only to those causes of overpayments which can be attributed to claimant error or omission, or to the design of the tax credits regime, or a combination of both. We examine these other possible causes of overpayments below.

29. It seems self-evident to us that HMRC cannot take steps to improve the way in which it administers tax credits without first identifying, and developing a detailed understanding of, the factors which cause overpayments and the extent to which each individual factor has contributed to the overall overpayments problem. We recommend that, as a matter of priority, the Government provide a detailed breakdown of as much of this information as is currently available.

OFFICIAL ERROR

30. As OPF pointed out to us, "what is interesting in the Paymaster General's analysis of overpayments [of 5 December] is that she does not include official error [as a cause of overpayments], and we do know there has been a huge amount of official error. Looking at the figures, they wrote off, for example, £24,000 in July [2005] due to official error."[46] OPF carried out research into tax credits between May and July 2004, based on a study of 100 lone parents. This research showed that "28% of people [surveyed] had been overpaid and, of those, 80% was due to official error", although OPF pointed out that that figure reflected the early years of the scheme, so was probably unusually high.[47] OPF described it as "very difficult" for anyone outside HMRC to establish levels of official error. Although figures are available about the number of overpayments HMRC has written off due to official error, those figures represent only those cases in which HMRC has accepted that it was reasonable for the person to have believed their award was correct, "so you lose a big chunk of the official error figures".[48]

31. It is clear that official error has been a cause of overpayments in a significant number of cases. The National Audit Office told us that "Departmental error can also lead to many overpayments".[49] The Comptroller and Auditor General's standard report on Inland Revenue's 2004-05 accounts referred to an example of the consequences of official error:

The unexpected problems at the time of the introduction of tax credits in 2003 caused the Department to issue 500,000 manual cheque payments totalling £170 million, with the supporting documentation being inadequately completed in some cases. In 2004-05, the Department wrote-off £33 million because the records were not good enough to cost-effectively match them to claimants.[50]

32. The NAO also noted that HMRC has improved the accuracy with which it processes information received from claimants: 78.6% of information was processed accurately in 2003-04, compared with 96.5% in 2004-05.[51] Sir David noted that "error tends to create re-work, which tends to create more havoc, so the more accuracy we can get into the system … the better we are at that the less we will need to put resources into fixing problems which we have helped create".[52]

33. The Paymaster General has said that "no complete analysis exists of official error causing or contributing to overpayments".[53] When we asked her if she was able to give us an estimate of the amount of overpayments caused by official error, she replied:

No, I cannot, but if you look at the interim findings it was 3.4%, and I think it will be a little higher than that. You do not have to believe me … look at the NAO Report and the information there on this study …[54]

It appears that the Paymaster General was here confusing the NAO's interim finding of the level of claimant fraud and error with official error. HMRC officials appeared to be similarly confused: they confirmed that 3.4% was the provisional figure for claimant error and fraud but then, when asked what contribution official error made to overpayments, said they could not remember "whether it is claimant or official error which is the bigger element within that, but it is a significant part of the 3.4[%]." Officials appeared to confirm that figures for both official error and for claimant error and fraud should be available in Spring this year.[55]

34. Subsequently, shortly before we agreed this report, HMRC indicated that it was still considering "whether it will be possible to give a meaningful breakdown of the estimates of error between official error and claimant error for future enquiries."[56] The Department explained that it was "difficult" to establish, on the basis of tax credits records, "whether the claimant or the Department is responsible for a particular error, for example, whether information was incorrectly reported or incorrectly recorded".[57]

Our conclusions

35. The Paymaster General has said that no complete analysis exists of official error causing or contributing to overpayments. This is a significant gap in HMRC's understanding of the reasons why overpayments arise. If HMRC is to succeed in improving the administration of the tax credits regime, the first thing it needs to understand is what is going wrong within its own processes, before it looks to problems elsewhere. As Sir David himself acknowledged, the more accuracy HMRC can achieve, the fewer resources it will need to put into fixing problems which it has helped to create. We recommend that the Government undertake a complete analysis of the incidence of official error and the extent to which it causes or contributes to overpayments, and that it publishes that analysis. If it is unable to carry out such an analysis, it should explain why.

IT SYSTEM ERROR

Background

36. It is also clear that IT system, or software, error has been another significant source of overpayments. According to the National Audit Office, the administration of payments "suffered from the serious problems with the computer systems during the introduction of tax credits in April 2003", which "both delayed the processing of claims and led to incorrect payments being made".[58] The NAO told us that software errors resulted in overpayments of £184 million in 2003-04 and 2004-05, and that HMRC was "continuing to investigate the reasons for other incorrect payments caused by system miscalculations."[59] The Comptroller and Auditor General's standard report on Inland Revenue's 2004-05 accounts summarised the cost of writing off overpayments arising from software error as follows:

In my 2003-04 report I noted that certain software errors had resulted in overpayments calculated as £94 million in 2003-04 of which the Department had written-off overpayments of less than £300 totalling some £37 million. I also noted that the Department expected further write-offs in respect of the balance of £57 million. In addition, the Department calculated and wrote off other overpayments of some £2 million that it considered had been caused by software errors. The Department calculated that [these] software errors … resulted in further incorrect payments in 2004-05 of £7.9 million. Various other incorrect payments have also resulted from other system miscalculations.[60]

37. In recognition of these difficulties, the Paymaster General undertook, in her Written Ministerial Statement of 26 May 2005, that HMRC would "improve the speed with which it identifies IT system problems and processing errors so that they can be resolved more quickly."[61]

Difficulties with the IT system

38. At the time the new tax credits regime was set up, it was intended that its administration would be wholly IT-based. It is evident that the administrative process currently requires significant human intervention. For example, when we visited the TCO in Preston, we heard that only about 25% of all claims went straight through the automated system without the need for manual intervention. About 80% of new claims required intervention; about 30%-35% of claims for renewal required intervention. The Public and Commercial Services Union (PCS) told us that the Government:

initially intended that the 'rapid data capture' process (the conversion of written information from application forms into electronic data) would handle around 90% of claims without further need for human intervention. Our estimate is that only 10% of the information is captured by this method.[62]

39. HMRC is involved in ongoing work to improve the IT system. In October 2005, Sir David told us that HMRC had:

been trying to stabilise—and I think have been successful—the IT system. It is still potentially fragile and, therefore, every time we have to do something, I approach it in the spirit that we do not want to lose any progress that we have made in stability.[63]

More recently, Sir David told the Committee of Public Accounts that HMRC still did not feel it knew "enough about the system to be clear of its resilience … it is a very complicated system".[64] In the course of the present inquiry, the Paymaster General updated us on the performance of the system:

IT performance has been significantly improved … a significant new software release was introduced without a hitch in November. Largely invisible to people outside, this will deliver real improvements in operational performance. In total there have been 300 improvements made to the system since April 2005.[65]

40. We received evidence from both PCS and the voluntary sector that the IT system was continuing to cause difficulties both for staff and claimants. PCS acknowledged an improvement in "identification of IT system problems", but felt that the resolution of processing errors had not similarly improved:

The computer system is not user friendly and continues to be often unavailable at key times … Our members report continuing problems with the interface between the [tax credits] computer system and treatment of data that HMRC operatives are forwarding for inclusion (such as a claimants' change of circumstances).[66]

41. Representatives from the voluntary sector referred to many examples, generally involving the tax credits helpline, where claimants had been disadvantaged by what appeared to be software error—although, from the perspective of helpline users, it is often difficult to tell whether it is official error or software error which is the problem. A recurring theme was the IT system's lack of flexibility, and the difficulty of correcting a mistake once it had been (erroneously) entered into the system—staff may accept that information is wrong, but still be unable to correct the information.[67] The system also appears to lose claimants' records periodically, as a representative from a Citizens Advice Bureau explained:

I used to think [the problem] was a lot administration, a little bit computer, but I personally now believe it is a lot computer and a little bit administration, particularly with this issue of losing clients. There are two times of year where a lot of clients drop off the system, particularly around the beginning of October … I was led to believe they could not get them back on the system. I am now told that about a quarter go back but they cannot tell you why, but the others are all left with manual payments … one of the suggestions at a meeting was a to change people's national insurance numbers because that way you could put them back on the system with a new national insurance number.[68]

42. Mr Field mentioned an example of a claimant he knew of who, when completing an application form, had put a line through a section of the form to be filled in by claimants with disabled children—to indicate that this section was not relevant to her. Her form was read electronically as indicating that she did have a disabled child.[69] The Chartered Institute of Taxation (CIOT) commented on the fact that the IT system had, on occasion, produced award notices containing internally inconsistent information—for example, where the number of children has differed in different parts of the notice.[70] The CIOT considered that the IT system should operate in such a way that "errors like these are isolated for checking, or are just never possible in the first place".[71]

43. The Paymaster General rejected PCS's statement that the IT system was often not available at key times:

The computer has downtime in order for it to have the software put on to it … I have no information … that the computer was repeatedly going down. The only incident I can find in all of the last 12 months is that there was a period of maybe two hours, and I cannot remember when it was, when the system did not perform as it should have done. This idea that it is repeatedly not available and you cannot get access I disagree with.[72]

Terms of settlement between HMRC and EDS

44. The tax credits IT system went live in April 2003, under a contract with Electronic Data Services (EDS). This contract ended on 30 June 2004 and was replaced by a contract with Cap Gemini.[73] On 22 November 2005, the Paymaster General announced that HMRC had reached a settlement with EDS:

the aggregate settlement is £71.25 million … including an up-front payment and payments of additional amounts over time. Details of the settlement are commercially sensitive and therefore bound by a legal confidentiality agreement as is normal in agreements of this nature.[74]

The Paymaster General had earlier stated that, as at 31 August 2005, the "identifiable costs" paid to EDS and, subsequently, Cap Gemini for running the IT system were £236 million, exclusive of VAT.[75] Between 2002-03 and 2004-05, the Department spent £19.7 million on IT consultants in relation to tax credits.[76]

45. HMRC has so far received £47 million from EDS; Sir David told us that the payment has "had the effect of being as if it was £47 million cash", but was "funded in different ways through different streams".[77] £24.25 million remains outstanding. Since the Paymaster General's November announcement, further details of the settlement have emerged. Subsequent to our 19 April hearing with Sir David, the Committee of Public Accounts reported to the House that "staged payments of up to £26.5 million are contingent on EDS winning new business with the United Kingdom Government … there is however no guarantee that EDS will win sufficient new business to trigger payment of the full amount".[78] Prior to our 19 April hearing, however, the substance of the Committee of Public Accounts' report to the House was leaked to the media. We therefore discussed with Sir David the conditions attaching to the payment of the remaining £24.25 million.

46. Owing to the confidentiality agreement which formed part of the HMRC-EDS settlement, Sir David felt constrained in what he was able to tell us about the terms of the settlement. By way of context, he explained that the cost of taking EDS to court over the matter was estimated at £20 million and expected to take two years.[79] In reaching the settlement, HMRC considered it had a choice:

we could have settled for a lower sum of money and been certain or [taken] this mechanism [of staged payments contingent on EDS winning new business with the UK Government] against the threat that we did not reach a full and final settlement. There will only be full and final settlement with EDS when we are paid the £71,250,000.[80]

Prior to agreeing the settlement, HMRC had inspected EDS's order book and found that it "many, many times covered" the amount of the contingent payments. Sir David said he would be "extraordinarily disappointed if EDS did not honour this obligation to pay the remaining amount of money".[81]

47. We raised with Sir David the propriety of making part of the HMRC settlement contingent on EDS's future business with other departments of the UK Government. He stressed that nothing HMRC had done was meant "to influence in any way any decision [by] anybody in government to procure services from EDS".[82] He also contended that the terms of the settlement would not in fact influence any decision by government as to whether to award a contract to EDS, on the basis that:

[government] departments are under an obligation under EU law to make value for money declarations in terms of contract procurement … You are asking, 'When you come to make a value for money consideration in the National Health Service, would you factor into your calculation that there was a benefit for HMRC?' Answer, no.[83]

48. We also questioned Sir David about who in Government made the decision to accept the settlement with EDS. He described the situation as follows:

… we went specifically to the Treasury and to the NAO, who went all through this in great detail … this is an accounting officer deal. This is an accounting officer [that is, Sir David himself] who is responsible … We talked at the working level to the Treasury, but this was a matter for us to determine what we thought was in the best interests of [HMRC] … I did not seek ministerial approval … The decision to accept this deal was the decision of the Commissioners of the Revenue. It has nothing to do with the Treasury.[84]

Our conclusions

49. On the basis of the evidence we have received, the rate of error within the IT system seems to us to have been significant. Just as HMRC appears to have attempted no complete analysis of the contribution made by official error to overpayments, so we have seen nothing from the Department attempting to assess the contribution made by IT system error. Again, it seems obvious to us that HMRC must acquire a thorough understanding of the problems which have arisen if it is to succeed in improving the administration of the tax credits regime. We recommend that the Government undertake a complete analysis of the incidence of IT system error and the extent to which it causes or contributes to overpayments, and that it publishes that analysis.

50. The settlement of £71.25 million agreed by HMRC and EDS appears to have provided for staged payments of up to £26.5 million which are contingent on EDS winning new business with the United Kingdom Government. It is clear that the IT system which EDS delivered for the running of new tax credits was unsatisfactory in a number of respects. We have grave concerns about the wisdom of an agreement which then makes the payment of compensation to the affected government department by the provider of the unsatisfactory service contingent on that provider winning other contracts with government. Our concern is not that we believe the contingent payments will influence future decisions by government departments to award contracts, but that it will be impossible to be sure that they have not. The agreement has the appearance of impropriety, if not the fact.

51. We also draw the attention of the House to the confidentiality agreement which formed part of this settlement, and which so constrained the Chairman of HMRC in his ability to respond to our questioning. We are extremely concerned that HMRC appears to be claiming to have effectively 'contracted out' of its obligation to be publicly accountable for its administration and expenditure, by virtue of having entered a private contract. The existence of such a confidentiality requirement also makes it impossible for the House to assess what happened in this particular case, and to seek to draw broader lessons from it about the problematic area of government IT contracts. We recommend that the Government ensure that this particular settlement does not indicate the start of a trend on the part of public bodies towards agreeing such confidentiality requirements. We further recommend that the procurement, design, project management and delivery of the tax credits process and systems be independently examined by the National Audit Office, regardless of this agreement.

9 HM Treasury and Inland Revenue, The Child and Working Tax Credits: The Modernisation of Britain's Tax and Benefit System, April 2002; available at www.hm-treasury.gov.uk. Back

10 The Child and Working Tax Credits, para 4.48 Back

11 The Child and Working Tax Credits, paras 4.43 and 4.49 Back

12 The Child and Working Tax Credits, para 4.43 Back

13 Ibid. Back

14 Ibid. Back

15 The Child and Working Tax Credits, para 4.43 Back

16 HC Deb, 5 Dec 2005, col 57WS Back

17 Qq 561-562 Back

18 HC Deb, 10 Oct 2005, col 324W Back

19 Comptroller and Auditor General, Standard Report on the Accounts of the Inland Revenue 2004-05, 7 October 2005; available at www.nao.org.uk Back

20 Ev 159 Back