Select Committee on Treasury Tenth Report


2  The Framework for National Statistics

National Statistics

18. The concept of 'National Statistics' was introduced as part of the non-statutory Framework for National Statistics in 2000. Essentially, National Statistics accreditation acts as a quality kite-mark for official statistics: only those data produced in accordance with the National Statistics Code of Practice receive the accreditation. Although all National Statistics are official statistics, not all official statistics are National Statistics. The Code applies to all datasets which the ONS is responsible for producing and to National Statistics produced within government departments. The National Statistician has overall responsibility for the professional integrity and statistical quality of all datasets designated as National Statistics, although within government departments, ministers are responsible for deciding which statistics should be National Statistics.[26]

19. The Minister told us that the term 'official statistics' refers to an unquantifiable number of statistical outputs including "databases", "management data" and "one-off research projects" produced by Government officials who are not necessarily members of the GSS. He told us that, currently, around 1,450 datasets are produced by members of the GSS. Of these, approximately 1,180 are designated as National Statistics, with around 250 of these produced by the ONS.[27] Figure 1 outlines the relationships between these figures.Figure 1: Categories of official statistics in the United Kingdom

figure in here

Source: Q 240

20. Consequently, approximately four in five GSS-produced statistics are National Statistics. However, the undefined nature of the wider 'official statistics' category means that we are unable to determine what proportion of all statistics across Government are National Statistics. At the departmental level, we discovered from the Home Office that, of its 165 statistical outputs, 20 (that is, approximately 12%) were National Statistics. The Home Office cautioned that these figures were subject to some definitional uncertainty, with some questions about what constituted a separate statistical series. Obviously, the situation in the Home Office is not necessarily representative of other government departments.

CRITICISM OF THE NATIONAL STATISTICS SYSTEM

21. The existence of National Statistics accreditation was criticised by some of our witnesses. The RSS believed that the public was unable to distinguish between National Statistics and other statistics, explaining that the public found it "very hard to understand why this one is a National Statistic and this one is not".[28] The RSS described National Statistics accreditation as creating "a two-tier system" which was "something of a patchwork at the moment":

[For example] we have the situation where monthly waiting lists are not National Statistics but quarterly waiting lists are. It is very difficult for anyone outside the system to understand the logic of that.[29]

The Statistics User Forum described the concept of National Statistics as "confusing" even to professional users:

the great majority of users in the wide world … do not draw any distinctions as to where statistics come from within Government; it is just Government official data and they do not have any subtleties about which departments in particular.[30]

22. Lord Moser expressed concern about the role of ministers in determining which data sets would be treated as National Statistics. He told us that, given the chance, he would "abolish the concept of national statistics" and replace it with "what we have always had and what every other country has: official statistics".[31] He argued that ministerial control of National Statistics undermined trust:

The great mistake is to have a category of statistics which are left totally to the ministers' hands. It is a formula for lack of trust, because anybody who looks into it can see that the minister has decided that those particular things do not go anywhere near ONS; they are totally for him or her to decide on. That, to me, is a very basic flaw.[32]

23. The Statistics User Forum argued that all statistics should be produced in the same way according to a code of practice. It told us that this did not necessarily mean that the statistics must be "highly accurate", simply that they should be as accurate as possible and that any shortcomings should be made clear.[33] Professor Alison Macfarlane of City University London supported this idea. She told us that she wanted any code of practice "to apply as widely as possible". She argued that currently it was difficult to have any faith in numbers quoted by politicians which had not been produced in accordance with the existing Code of Practice.[34]

24. Despite these criticisms, the Government proposes that the National Statistics system should continue.[35] The Minister's argument appeared to be that it was better to have some official data produced to the higher National Statistics standard than to have all data produced to an identical, but lower, standard. He commented that, given the difficulty of defining and quantifying statistics produced outside the GSS, "it is probably quite hard to have a code of practice that covers absolutely any statistics, any statistical outputs, or any data that may be generated within Government".[36]

PUBLIC DOMAIN DISTINCTION

25. Simon Briscoe suggested that statisticians working in government departments could "quite happily work for their ministers", but that once figures approached the public domain they should "cross a dotted line" and become "the responsibility of the National Statistician and the [ONS]".[37] The RSS echoed this opinion when arguing that statistics produced for internal management purposes should be treated differently from those produced for public consumption. It told us that statisticians working in departments frequently prepared "alternative 'What if?' scenarios" as part of a department's process of determining what policies to pursue, and that such calculations were unlikely to be of interest to the public:

Clearly, only one policy is going to be arrived at and that process … is not going to fall into the public domain, unless all evidence to ministers is publishable … What we are talking about are things which are used by the public to monitor Government performance.[38]

26. The RSS assured us that, in most instances, it would be possible to determine which statistics would be prepared for public consumption in advance. It told us that:

The vast majority of the statistics which come into the public domain are produced on a regular basis. We know they are coming out, we know when they are coming out and we know what they are based upon. It is easy to identify those as being the statistics which are being … produced largely for public consumption.[39]

OUR CONCLUSIONS

27. As the Minister explained, there are effectively four categories of statistics in the UK: official statistics; GSS-produced official statistics; National Statistics; and ONS-produced statistics (see Figure 1). We accept the Minister's point that it can be difficult to define and quantify data produced outside of the Government Statistical Service and we recognise that some data is produced for internal use only. For this reason, we accept that it may not be feasible to apply a code of practice to all official data. We believe, however, that it would be helpful if a clearer distinction could be made between data which is produced for the public domain and data which is used for internal purposes within departments.

28. We note the evidence we have received from witnesses who described the National Statistics system as confusing to both professional users and the wider public. We agree that the current model is confusing. We are concerned that the creation of National Statistics has resulted in the emergence of a two-tier system, which has tended to undermine public confidence in official statistics. We recommend that the Government seek to address the problem by setting down clear criteria for what categories of statistics should be classified as a 'National Statistic'. We discuss below who should have responsibility for determining which statistics should be classified as National Statistics.

29. We note that the Home Office has been able to provide figures for the proportion of all its official statistics which are designated as National Statistics. We recommend that the Government publish the equivalent figures in respect of all government departments, in order to inform parliamentary scrutiny of the expected legislation.

The decentralised statistical system

30. The Government's consultation paper proposes retaining the current decentralised system, on the basis that "the strengths that flow from decentralisation outweigh the potential benefits from centralising all statistical activity in one office".[40] The Minister told us that having a cadre of professional statisticians based in separate departments meant that:

officials in those departments have ready access and influence with those statisticians in the way that they go about their work.[41]

He said that the arrangement also had advantages for the statisticians themselves, arguing that they were "closer to the data sources", "closer to the consumer" and "closer to the policy imperatives for which statistical systems need often to be devised".[42]

31. The ONS supported the Minister's position, arguing that "real added value" was gained by having statisticians in government departments and that the arrangements ensured that policy issues were "founded on evidence-based policy".[43] The Director of National Statistics and Planning told us:

The ability for the statistician and the head of profession in each department to have some kind of protection through a statutory code is a perfectly acceptable way of working, and I think if you actually take the statisticians out of that environment, they are going to be far less valuable and far less effective than they are now.[44]

32. Other witnesses expressed reservations about the current decentralised system. The RSS agreed that there were "strong arguments" for maintaining the decentralised system but argued that "compensating mechanisms" had to be found to ensure that there was public confidence in the production of statistics.[45] The RSS assured us that statisticians in Government departments were capable of acting differently depending on whether they were producing and analysing internal management data, on the one hand, or statistics for the public domain, on the other:

It is not unrealistic to say that one of the things which we need to put in place if we are to retain the benefit of a decentralised system is a system of Chinese walls whereby, professionally, the statisticians producing these numbers within policy departments as public consumption figures should actually have a responsibility to the National Statistician.[46]

The Chief Statistician of Canada argued that such 'Chinese walls' could be formalised in the form of statistical "institutes" within Government departments.[47]

OUR CONCLUSIONS

33. We acknowledge the benefits of the current decentralised statistical system, as set out by the Government and endorsed by the Office for National Statistics and others in the statistics community. Given the apparently high levels of support for the current system, we do not suggest bringing all of the Government's statistical operations together into a single office.

34. However, we note the Minister's acknowledgement that the decentralised system affords government departments, and potentially ministers, more influence over statisticians than would a centralised system. We agree with the Royal Statistical Society that a decentralised system risks perceptions of political interference and that "compensating mechanisms" are therefore required. We recommend to the Government that, although statisticians should remain close to policy colleagues in departments, they should have formal responsibility to the National Statistician for any statistics they produce which are intended for the public domain. We also recommend that the Government examine the adequacy of the 'Chinese wall' arrangements which are currently in place in departments—between departmental statisticians and the rest of the department, including ministers—and that it puts in place improved arrangements, if necessary. We point the Government to the example of the Government Legal Service, which appears to enable lawyers to work within their individual departments in the interests of those departments while still retaining a professional duty to offer impartial, objective advice which may not be politically welcome.

Scope of the proposed legislation

35. The Government's consultation paper proposes that the new independent board would have responsibility for assessing statistics already designated as National Statistics against the new statutory code of practice, but that ministers would remain "wholly responsible" for statistics produced by their departments which are not National Statistics. It would be for ministers to decide whether these statistics should become National Statistics, "subject to the board [subsequently] assessing and approving them against the code".[48]

36. This aspect of the Government's proposals was criticised by witnesses on two counts: first, that the scope of the legislation would be too narrow, with the focus being on the ONS and not the wider system; and secondly, that the independence of the system would be undermined by ministers retaining control over what should be designated as National Statistics within their departments.

37. On the first count, the RSS considered that the Government's consultation paper focused on the ONS and gave "not as much attention to the wider statistical system as it should".[49] The RSS argued that if statistics produced by departments on topics such as "crime, education, health and social security" were omitted from the scope of the legislation, then public confidence would be eroded rather than enhanced.[50] The Chief Statistician of Canada, Dr Ivan Fellegi, told us that:

The role of the proposed new board is very limited outside of the ONS. While it would have statutory authority to enforce a new code of practice outside of the ONS, its only tool to do so would be an audit function and the power to "name and shame"—which is really not much different from the tool currently available to the current Statistics Commission.[51]

38. Lord Moser argued that the ONS was already the best part of the statistical system and was therefore the part which "least needs any kind of reform".[52] The Statistics Commission told us that it felt that many of the concerns about statistics expressed by the media and among the public related not solely to the ONS but "much more frequently" to other parts of the system, and argued that the Government therefore needed to "address the totality of the system".[53]

39. On the second count, the Statistics User Forum argued that the proposals would "perpetuate the unsatisfactory division between National Statistics and other official data", and that allowing ministers to decide what should be designated as National Statistics would "undermine the whole idea of statistical independence".[54] Dr Fellegi told us that:

Under the proposals, the scope of the National Statistical System is left for individual ministers to determine. They would decide whether "their" statistical activity ought to be part of the National Statistical System. I would suggest that you may want to ask yourself what incentive Ministers would have to favour opting in. Surely, the likelihood that, should they do so, their statistical activity would be subject to audits is not a very strong incentive to opt in.[55]

40. Dr Fellegi felt that these two issues of scope and ministerial control, combined with the public's inability to differentiate between statistics originating from the ONS and statistics coming from other Government departments, would mean that the Government's proposals would have little impact on public confidence. He told us that, if the Government did not extend its proposals beyond the ONS, and if the proposals failed to remove responsibility for designating National Statistics from individual ministers, then the exercise would amount to no more than "tinkering" and would fail to solve the problem of public trust in statistics.[56]

41. The Minister responded to these criticisms by telling us that the Government had chosen to concentrate on the 1,450 National Statistics products, in the first instance, because they represented "the most important sources of data that tell us what is going on in the economy and in society".[57] The Minister defended the right of ministers to decide whether a particular dataset should be designated as a National Statistic, arguing that it was for ministers to be responsible for outputs, the operation of the departments producing them and the allocation of resources.[58] He suggested that the proposals, if implemented, would in fact give ministers added incentive to designate data as National Statistics:

with the added status that I think this independent process will give to the nature and the confidence in National Statistics, there will increasingly be—for ministers who are concerned about the confidence people can have in the statistics that cover the key areas of activity their department is responsible for—an incentive for them to want to see them within the system.[59]

OUR CONCLUSIONS

42. The Government's consultation paper proposes that the new independent board would have responsibility for assessing statistics already designated as National Statistics against the new statutory code of practice, but that ministers would remain responsible for other official statistics produced by their departments. In addition, ministers would be responsible for designating which statistics within their departments became National Statistics.

43. We note the concerns of several witnesses that the Government's consultation paper fails to address the independence of statistics outside the ONS. We are not convinced by the Minister's assertion that National Statistics "represent the most important sources of data" for explaining "what is going on in the economy and in society", because they do not include some of the most frequently-quoted data on health, crime and education. For example, monthly Department of Health figures on NHS Inpatient and Outpatient Waiting Times are not National Statistics, nor are Home Office statistics on Race and the Criminal Justice System, and nor are Department for Education and Skills figures on Revenue Funding per school pupil.[60] We agree with the Chief Statistician of Canada that the public is unlikely to distinguish between statistics originating from the ONS and statistics coming from other government departments. We are therefore concerned that, by addressing only the independence of the ONS in its consultation paper, the Government may have missed an opportunity to improve public confidence in official statistics. We recommend that the Government examine including protocols in its forthcoming legislation that would be applicable to all official statistics.

44. We recommend above that the Government should specify clear criteria for what constitutes a National Statistic based on the principle of distinguishing between data generated or collected for the purposes of public consumption and data generated or collected for the purposes of internal management. We acknowledge the Minister's argument that ministers are ultimately responsible for the outputs of their departments and that they may therefore have some incentive to see their departmental statistics awarded the National Statistics kite-mark of quality. We are nevertheless concerned that retention of this control by ministers would undermine the perceived independence of the system. The promise of a more thorough audit would not appear to offer a strong incentive for ministers to designate statistics within their departments as National Statistics. If the Government proposes to retain the basis of the current National Statistics system, we recommend that decisions about designation should rest with the independent board, not with ministers, in order to ensure that decisions about what constitutes a National Statistic are made objectively and consistently by a body external to the government departments concerned.

Code of Practice

45. Under the current Framework for National Statistics, the National Statistics Code of Practice is the foundation document, intended to ensure quality and integrity across National Statistics. The Government's consultation paper proposes the "continuation of the National Statistics system with a set of highest quality statistics produced in line with the published code".[61] The Government proposes to give the code statutory backing "by legislating to make the development and maintenance of the code the specific responsibility of the board".[62]

46. The Statistics Commission has previously described the existing Code as being poorly understood within Government, observing that "the day-to-day interpretation of the Code can vary widely between the numerous bodies that produce official statistics".[63] Reporting on the financial year 2004-05, the Commission noted that:

Through the year, the Commission's attention was drawn to various suspected breaches of the Code. None of these was particularly serious in itself, and in total they relate to an extremely small percentage of the mass of statistical releases. Taken together, however, they do raise a question about how well the Code is understood across government.[64]

47. The Commission also described the Code as being difficult to enforce:

It is important to note here that only a few parts of the Code are of a kind that allows an unequivocal judgement to be made about adherence to it. Much of the Code is aspirational in style and requires no specific evidence of adherence. Thus the evidence [of breaches] that does exist often relates to transgressions against less fundamental aspects of the Code.[65]

48. The Commission argued that the proposed statutory code of practice should be "drafted by the National Statistician and approved by the Governing Board and/or an appropriate Parliamentary Committee", without the interference of ministers.[66] The Commission commented that ministers "should of course be consulted", but that their agreement to the terms of the code should not be a prerequisite.[67] The Minister confirmed that this would be the case, telling us that the preparation and drafting of the code would "in all likelihood" be led by the National Statistician, but that it would be "for the board to direct that [and] for the board to approve that", and that, subsequently, it would be for the board to modify the code "as it thinks fit, depending on how it works".[68] He said that none of these functions would "be a matter for ministers".[69]

49. The Statistics Commission told us that it was planning to prepare a draft code of practice in the coming months in order to allow any new board to "hit the ground running, rather than having to create a new code of practice over many months".[70]

OUR CONCLUSIONS

50. We note the Statistics Commission's concerns regarding the clarity and enforceability of the existing Code of Practice. We therefore welcome the Government's proposal to establish a statutory code of practice, and its assurance that drafting the code would be a matter for the independent board without involvement from ministers. The introduction of a new statutory code of practice would offer the opportunity to establish a code which is unambiguous, able to be understood by a wide range of readers and sufficiently precise as to be readily enforceable. We are pleased to hear that the Statistics Commission intends to put forward proposals for a new statutory code of practice in the next few months.

Retail Prices Index

51. The Retail Prices Index (RPI) has a special status among statistics in the UK. The Framework for National Statistics provides that the National Statistician will be responsible for "developing and maintaining statistical standards, definitions and classifications", except in the case of the RPI where special arrangements apply:

the National Statistician will take the lead in advising on methodological questions concerning the RPI but the scope and definition of the index will continue to be matters for the Chancellor of the Exchequer.[71]

52. The Chancellor has been responsible for the scope and definition of the RPI since 1989, when the CSO took over responsibility for the production of the RPI from the Department of Employment. Previously, the Secretary of State for Employment had responsibility. In determining the scope and definition of the RPI, the Chancellor of the Exchequer refers issues to an RPI Advisory Committee which he convenes as appropriate. Membership of the Advisory Committee typically includes experts on the matters being discussed and representatives of users of the RPI. The Advisory Committee makes recommendations to the Chancellor who, in theory can accept or reject them. However, since the first Advisory Committee was established in 1904, no recommendation has ever been rejected.[72]

53. The Treasury Committee has previously considered the anomalous treatment of the RPI. In 1998, the Committee said that it saw "no justification" for the Chancellor of the Exchequer retaining control of the scope and definition of the RPI under the new arrangements for National Statistics. In 2001, the Committee concluded that it had heard "no sound argument why the scope and definition of the RPI, like the features of any other important statistical series produced by the ONS, should not be under the control of the National Statistician".[73] In the course of our present inquiry, both the Statistics User Forum and the Society of Business Economists submitted that the anomalous position of the RPI should end.[74]

54. Countries with statistical systems comparable to the UK do not treat the RPI in the way in which the UK treats it. Table 1 sets out the governance of national indices of prices in a selection of other countries and shows that Sweden is the only country other than the UK in which a senior government statistician does not have responsibility for the definition and scope of the index. In Sweden's case, however, it is Parliament rather than the Government which is responsible for definition and scope of the index, and eight out of the nine members of the permanent advisory committee, the Index Board, are appointed by the Chief Statistician.

55. The Minister explained the Government's justification for continuing the anomalous treatment of the RPI:

[The RPI] has traditionally been the responsibility directly of the Chancellor, essentially because of its unique place and its unique role. It is used for policy, for legislative, for contractual purposes. It is used for up-rating pensions and benefits. It is used for indexing tax thresholds […] Up to this point, there has been a very strong view that … the degree of government exposure from changes to the RPI made it appropriate to leave the ultimate say on any changes to the Chancellor.[75]Table 1: Responsibilities and governance of national indices of prices
CountryDefinition and scope of index Methodology Advisory Committee Appointment of Advisory Committee
CanadaChief Statistician
Chief Statistician PermanentChief Statistician
FranceDirector General, national statistics office Director General, national statistics office None
NetherlandsDirector General, national statistics office Director General, national statistics office PermanentDirector General, national statistics office
New ZealandGovernment Statistician Government Statistician PeriodicGovernment Statistician
SwedenParliament
Index BoardPermanent Chief Statistician[76]
United StatesCommissioner, Bureau of Labor Statistics Commissioner, Bureau of Labor Statistics PermanentCommissioner, BLS and Secretary of Labor

Source: House of Commons Library

OUR CONCLUSIONS

56. We have considered the Minister's justification of the current anomalous treatment of the Retail Prices Index, whereby the Framework for National Statistics gives the National Statistician responsibility for "developing and maintaining statistical standards, definitions and classifications" of all statistics other than the RPI. We question whether the degree of Government exposure from changes to the RPI is such that the Chancellor of the Exchequer needs to retain control of its scope and definition, particularly given that equivalent indices in comparable countries are not treated as 'special cases'. We invite the Government to explain more fully why its considers that the RPI should be treated differently from other key macroeconomic statistics.


26   Independence for statistics: A consultation document, para 2.8; Office for National Statistics website, http://www.statistics.gov.uk/about/national_statistics/introduction.asp  Back

27   Q 240 Back

28   Q 37 Back

29   Ibid. Back

30   Ev 16; Q 81 Back

31   Q 216 Back

32   Ibid. Back

33   Q 83 Back

34   Q 20 Back

35   Independence for statistics: A consultation document, para 4.14 Back

36   Q 241 Back

37   Q 6 Back

38   Q 60 Back

39   Q 59 Back

40   Independence for statistics: A consultation document, para 4.8 Back

41   Q 262 Back

42   Ibid. Back

43   Q 190 Back

44   Ibid. Back

45   Q 58 Back

46   Ibid. Back

47   Ev 79 Back

48   Independence for statistics: A consultation document, para 4.15 Back

49   Q 35 Back

50   Letter from the President of the Royal Statistical Society to the Chancellor of the Exchequer, 26 January 2006 Back

51   Q 101 Back

52   Q 206 Back

53   Q 120 Back

54   Ev 96 Back

55   Q 101 Back

56   Qq 103 and 115 Back

57   Q 240 Back

58   Q 250 Back

59   Ibid. Back

60   www.dh.gov.uk/PublicationsAndStatistics/fs/en; www.homeoffice.gov.uk/rds/; www.dfes.gov.uk/rsgateway/index.shtml  Back

61   Independence for statistics: A consultation document, para 4.14 Back

62   Ibid. Back

63   Statistics Commission, Annual Report 2004-2005, July 2005, p 7 Back

64   Ibid. Back

65   Ibid. Back

66   Letter from the Chairman of the Statistics Commission to the Chancellor of the Exchequer, 20 January 2006 Back

67   Ibid. Back

68   Q 247 Back

69   Ibid. Back

70   Ibid. Back

71   Office for National Statistics, Framework for National Statistics, para 3.4 Back

72   Office for National Statistics, The Retail Prices Index: A Technical Manual, 1998, paras 1.7-1.7.1 Back

73   Treasury Select Committee, First Report of Session 1998-99, Office for National Statistics, para 48; Treasury Select Committee, Second Report of Session 2000-01, National Statistics, HC 137, para 12 Back

74   Ev 96; Ev 76 Back

75   Qq 265-266 Back

76   The Chief Statistician appoints eight of the nine members of the advisory board; the Government appoints the ninth member. Back


 
previous page contents next page

House of Commons home page Parliament home page House of Lords home page search page enquiries index

© Parliamentary copyright 2006
Prepared 26 July 2006