House of Commons - Treasury

Select Committee on Treasury Second Report

III. NATIONAL STATISTICS: SPECIFIC ISSUES

6. When we heard oral evidence from the Economic Secretary in March 2000, it was unclear whether the initial scope of National Statistics would be significantly larger than the current work of the ONS.[20] In fact, the Framework Document shows that a large volume of the statistical work undertaken in Government departments other than the ONS will bear the National Statistics label and the Minister told us that she expected the scope of National Statistics to "grow in time".[21] We welcome this. However, we were unable to form an assessment of how many Government statistics and statistical processes have been excluded from National Statistics because "there is no established comprehensive list of statistics produced by Government departments".[22] Mr Cook said that the ONS was "preparing a comprehensive list of National Statistics" which would help "identify ... significant statistics that are not National Statistics".[23] It is important that the scope of National Statistics expands over time, so that Government policy-making, and the scrutiny of it by Parliament and the public, is increasingly well informed.

7. Neither the Statistics Commission nor the National Statistician have a formal role, specified in the Framework Document, in relation to the scope of National Statistics. Both said that they did have a role, however. Mr Cook, for example, said "the National Statistician's role on scope is as an adviser to Ministers ... certainly I do not think you should assume that any national statistician will be inactive".[24] The Minister emphasised that decisions on scope could only be taken by departmental Ministers because of the resource implications which might arise from ensuring that a statistical series met the quality and other criteria for inclusion in National Statistics.[25] We welcome the intentions of the Statistics Commission and the National Statistician to advise on scope, and recommend that, in its reply to this Report, the Government acknowledges and clarifies their roles in this area.

8. Our concerns remain about the role of Ministerial discretion in determining the scope of National Statistics and we prefer the scope of National Statistics to be an independent decision for the National Statistician, in consultation with the Statistics Commission.[26] Two significant inconsistencies in the types of statistics included by departments in the initial scope of National Statistics have been immediately apparent. Firstly, the Royal Statistical Society has emphasised that "there is considerable inconsistency over the inclusion of surveys" and "no department has been brave enough to make a statement in the initial scope to the extent that any future survey or other statistical work undertaken to research an issue of national concern, even if it is a one-off project, will be included within National Statistics".[27] Secondly, the distinction between National Statistics and statistics produced for internal management purposes can be blurred, particularly in the case of management data linked to Government priorities, such as hospital waiting list statistics. At the moment, quarterly hospital waiting list figures are included in National Statistics but monthly figures are not, although they are often used as a key indicator by the Government.[28] Users of statistics, particularly in Parliament and the media, should pay careful attention to whether or not Government statistics carry the National Statistics badge. We hope the media will adopt a practice of reporting whether or not a particular Government statistic is a National Statistic.

Statistics Commission

9. The Framework for National Statistics stated that the new Statistics Commission "will play a key role in advising on the quality, quality assurance and priority setting for National Statistics, and on the procedures designed to deliver statistical integrity". We were concerned that this formal role might constrain the Commission from commenting on other matters such as the scope of National Statistics. Sir John Kingman, Chairman of the Commission, assured us, however, that the Commission would use its "one power ... to report and advise" on "anything which is relevant to statistics", including scope, the resources available to the ONS and the Commission, the calibre of Government statisticians, the status of the retail price index and the ways in which statistics are used by the Government, as well as issues of quality and integrity.[29] Sir John also said that "any advice we give is necessarily public because Ministers have told us that we have to be transparent in our proceedings".[30] The Statistics Commission could complement Parliament in scrutinising the work of the ONS, the statistical work of other Government departments, and the ways in which the Government uses statistics. We hope that our successors in the next Parliament use the Statistics Commission's reports to continue our scrutiny of National Statistics.

National Statistician

10. We identified in our 1998 Report the difficulty of combining the role of Director of the ONS with that of head of the GSS, particularly as the latter role lacked the powers necessary to ensure that the activities of departmental statisticians were coordinated. We recommended that "the Government accord a high priority to ensuring that the 'Head of National Statistics' role is managerially workable and supported by sufficient powers to coordinate the production of official statistics as well as ensuring the Head has the required freedom from political interference".[31] By placing the responsibility for the professional quality of National Statistics, the design and implementation of the Code of Practice, and the resolution of interdepartmental disputes over statistical integrity and methodology on the National Statistician, the Framework for National Statistics could facilitate the coordination of the work of statisticians throughout Whitehall. It is by no means certain, however, that these changes will make the devolved system of Government statistical activity more manageable.

11. In oral evidence on 1 March 2000, the Economic Secretary told us that "the National Statistician will have a right of access in terms of integrity issues directly to the Prime Minister should he wish to exercise that at any time".[32] In the Framework for National Statistics, however, it was stated that the National Statistician will "have the right of access to the Prime Minister, through the Head of the Home Civil Service, on matters concerning the integrity and validity of official statistics including on resources where he/she believes they impact on the integrity and validity of official statistics".[33] When challenged to reconcile these statements she said "all the people who head up Government departments or parts of Government who are effectively Permanent Secretaries ... have an initial contact before anything else with the Head of the Home Civil Service ... I believe that gives the National Statistician direct access under normal arrangements through to the Prime Minister, if that is what he so wishes".[34] Irrespective of the Minister's understanding that direct access to the Prime Minister is usually indirect, through the Head of the Home Civil Service, in this case, where the National Statistician is responsible for guaranteeing the integrity and quality of National Statistics, we recommend that direct access should mean direct, not through the Head of the Home Civil Service.

Retail Price Index

12. In 1998 we said that we saw "no justification" for the Chancellor of the Exchequer to retain control of the scope and definition of the Retail Price Index (RPI) under the new arrangements for National Statistics.[35] In March 2000, the Economic Secretary told us that the Chancellor would retain responsibility for the scope and definition of the RPI "because of its importance" and to provide "a more direct link with accountability" through parliamentary questions and debates.[36] She emphasised, however, that the ONS was responsible for "the compilation and indeed the presentation and publication of the index", a point reaffirmed by Mr Cook in November 2000.[37] Sir John Kingman told us that "the RPI is quite a complex situation and it is one which I think the Commission will want to do some serious work on".[38] We welcome the early attention which the Statistics Commission will pay to the division of responsibilities for the Retail Price Index (RPI) between the National Statistician and the Chancellor of the Exchequer. We have 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 and we reiterate our earlier recommendation that the Chancellor should not control these matters.

Reliability of Statistics

13. The reliability of the Government's statistical output is an important issue which can have ramifications on policy-making throughout Whitehall and beyond, as our 1998 Report showed in relation to problems with the reliability of the average earnings index.[39] Sir John Kingman told us that one of the Statistics Commission's first decisions had been to tell "the National Statistician that we would like to discuss with him and his colleagues ways in which the statistical output from ONS and National Statistics at large can be accompanied by indications of reliability and accuracy".[40] He said that "the sort of 'health warning' [that should accompany National Statistics] will depend on the sort of output and that is something we shall want to discuss".[41] It is promising that the Statistics Commission has paid early attention to the reliability of National Statistics and we look forward to the development of clear, consistent guidance for users on this issue.

14. Prior to both of our evidence sessions on National Statistics in 2000 we received written memoranda from the Greater Manchester Low Pay Unit drawing attention to apparent deficiencies with job vacancy, and other labour market, statistics. The Greater Manchester Low Pay Unit was particularly concerned with discrepancies between the number of vacancies advertised in job centres and the figures published for total job vacancies in the region. The ONS multiplies the number of vacancies advertised in job centres by three in order to estimate the total number of vacancies in the economy, a measure which the Treasury said "despite well-known limitations and incomplete coverage ... still provides a useful measure for analysis".[42] The multiplication factor of three is derived from occasional employers' surveys and, according to Mr Goldsmith, of the ONS, "a little bit of thumb in the air". Mr Goldsmith went on to assure us that "appropriate health warnings" are attached to job vacancy statistics whenever conclusions are drawn from them.[43]

15. In February 2000, the Treasury produced a policy paper entitled "The Goal of Full Employment: Employment Opportunity for all Throughout Britain", which included a discussion of the level of job vacancies and trends in the ratio between job vacancies and unemployment. The document did not contain any information about the reliability of job vacancy statistics, although the Treasury later told us that "it has been assumed that it would be clear that the estimate of 'around one million' vacancies is only broadly indicative".[44]

16. Since we began exploring the reliability of job vacancy statistics, the ONS has reconsidered its method of estimation. In a footnote submitted to the oral evidence we took in November 2000, the ONS said that it regarded the multiplication factor of three as "insufficiently robust for ONS to incorporate into an adjustment procedure for its published vacancy data" and that "a new survey approach, designed to estimate the number of vacancies in total directly" is being developed.[45] The ONS has also recently declined to provide data on the unemployment to vacancies ratio in different parts of the United Kingdom, on the grounds that the ratio between vacancies notified to job centres and total vacancies could vary across the country, even though such figures were published by the Treasury in February 2000 and were the basis for its conclusion that recently "every region in Britain has seen sharply falling unemployment and rising levels of vacancies".[46] The ONS's reassessment of the reliability and use made of job vacancy statistics is a positive indication of the impact of the new arrangements for National Statistics. We recommend that the Statistics Commission, in its work on the reliability of National Statistics, pays close attention to labour market statistics.

17. When the Minister and her officials appeared before us in March 2000, we asked her about the improvements made to the average earnings index (AEI) to improve its reliability. Mr Grice, of HM Treasury, told us that the improvements to the AEI would take "a period of months to implement" and that he could not describe the series as "absolutely right".[47] The Minister, however, said that she was "certainly confident about the quality of the AEI".[48] It was reported in December 2000 that the National Statistician had described the AEI as methodologically sound but possibly "irrelevant in a new policy context" because of recent changes in the pattern of economic activity.[49] We recommend that, in reply to this Report, the Government publishes in full its thinking on the future development of average earnings data.

ONS Resources

18. Table 1 shows the future expenditure plans of the ONS, following the completion of the 2000 Spending Review.

Table 1: ONS Expenditure Plans 2000-04

	2000-01	2001-02	2002-03	2003-04
Cash terms (£m)	136	186	139	126
Cash terms, excluding census (£m)	91	97	123	120
Real terms (£m)[50]	136	181	132	117
Real terms, excluding census (£m)	91	95	117	111
Real growth (%)	-	+33.1	-27.1	-11.4
Real growth, excluding census (%)	-	+4.4	+23.2	-5.1

At first glance, the ONS's spending review settlement would appear to be disappointing, with planned expenditure in 2003-04 lower than in 2000-01. Taking this point up, the Statistics Users' Council warned us that "the change from 'official' to 'national' statistics, if it is to have any real meaning, will require extra expenditure" and we probed reports that the ONS was one of the least well funded statistical offices in the developed world.[51] In a supplementary memorandum, the ONS explained that, excluding expenditure on the census, its expenditure in real terms was planned to increase to 2002-03, with a slight fall thereafter, as table 1 shows.[52] £20 million, in cash terms, of this increase is earmarked for developing ward level data as part of the Cabinet Office's PAT18 social exclusion project.[53] Mr Cook told us that "the important element of that project is that not only the current outputs of ONS will change, but it will be one of the significant opportunities to change the quality of the capital base of ONS ... Its real value is not so much that it is going to create new data, but that it will allow very much more powerful access to data, which we have locked away in systems which are rather old now and not very flexible."[54]