Dear David,

PUBLIC SECTOR OUTPUT AND PRODUCTIVITY STATISTICS

I wrote to you on 27 April about public sector output and productivity statistics following a report in the Sunday Times. Since then, on 30 June, ONS has published improved estimates of health output. On 18 October we shall be publishing a paper on productivity in the health service, the first in a series of articles on productivity in the major public services drawing on national accounts and related estimates. On the same date we shall also be publishing a further analysis of the health output estimates. Given the public and political interest in this area, this letter explains the background to, and handling of, developments since April.

In my letter of 27 April I said that "a wider and improved range of health output indicators is currently being subject to peer review for possible inclusion in this year's Blue Book". I also informed Oliver Letwin of this possible change at a meeting with him on 4 May, offering a further meeting with him or his economic advisor to discuss the details of these changes. (This offer was reiterated on 30 June and a meeting held with his advisor on 8 July.) Subsequently on 12 May we gave users advance notice of the potential revision in an article on Revisions to GDP planned for the 2004 Blue Book in the May issue of Economic Trends. This said:

"ONS is also in the process of quality assuring new estimates of the real output of government health services which could affect GDP in the closed period. These new estimates will be incorporated into the 2004 Blue Book provided ONS is satisfied that the methodology, and other aspects of data quality, are robust".

Further notice was given in the UK output, income and expenditure First Release on 26 May. In a section on Future revisions paragraph 7 of the Notes said:

"ONS is also in process of quality assuring new estimates of the real output of government health services which would affect GDP growth rates in the closed period. These new estimates will be incorporated into the 2004 Blue Book provided ONS is satisfied that the methodology and other aspects of data quality are robust."

These announcements fulfilled our obligation under the Revisions Protocol to give advance notice of planned future revisions. It was exactly the procedure that in February, when announcing the inclusion of additional material on revisions in First Releases, we said we would follow. I was disappointed therefore by comments that we had given insufficient notice and had surprised our users. Although whatever we do will seem inadequate to some, we will repeat such messages in the case of politically sensitive outputs. One step we have already taken is that from time to time the Office's Director of Macroeconomic and Labour Market Statistics will write direct to key users to draw their attention to announcements on Forthcoming Developments in Economic Statistics that they might have missed. His first letter was dispatched in the week beginning 13 September. A copy was sent to the Commission.

The timing of the changes in health output used in the National Account was determined by the availability of improved data sets from the Department of Health, the benefits of expertise and resources unlocked by the Atkinson Review, and the precedent that changes to National Accounts for 'closed' years are normally made, when needed, at the time of the annual Blue Book. The changes went through ONS standard peer review and quality assurance processes. These are clear grounds for refuting several rather unpleasant media comments that the timing of the change was driven by political influence.

Looking forward, the article to be published on 18 October will present the best available information on the output and productivity of the National Health Service, primarily from the National Accounts and complemented by information from other relevant sources. It will also include a critique of the sources and methods used in compiling the figures in order for readers to be aware of the basis, and the limitations, of the analysis. As proposed by Sir Tony Atkinson in the interim report of his review of the Measurement of Government Output and Productivity for the National Accounts, the article also includes a section on "triangulation", which consists of independent corroborative evidence on NHS productivity.

ONS will also publish at the same time a paper which provides a further analysis of health output figures produced using the new methodology introduced earlier this year in the National Accounts. This complements the methodological paper published on the National Statistics website in June 2004, at http://www.statistics.gov.uk/articles/nojournal/Healthcare_Output.pdf.

Both articles will be available in printed copy as well as on the National Statistics website. To launch this series, and to present the findings from the first article, ONS is holding a seminar at 14:00 on 18 October, to which we have invited academic experts, government officials, representatives of Opposition spokespersons and of the Treasury and Health Select Committees. Journalists with an interest in health, social policy and economics have also been invited.

When launching the new series I shall be emphasising that the estimates should be interpreted with care and that there are clear limitations of measurement in this area. The seminar is intended to help us explain this and to put the paper in context. This is why we have chosen to publish the paper at 1:30, just ahead of the seminar (people arriving for the seminar from 1:30 will be given copies at which point it will effectively be in the public domain). However we recognise that the media need reasonable time to digest the paper. Consistent with the Protocol on Release Practices, we shall therefore be making the paper available under embargo to journalists at 10:30 am . Paragraph 5(10)a of the Protocol provides for accredited journalists to be given embargoed access to complex (non-market sensitive) statistical reports where it is seen as "necessary to provide them with a period of time to assimilate and comprehend the data, in order to provide for informed comment at the time of release". The journalists in question have also been invited to the seminar and will therefore have an opportunity to ask questions about the article and seek clarification of any points in it.

Privileged early access to the articles has been strictly controlled and documented, consistent with Protocol on Release Practices. A limited number of experts in HM Treasury and the Department of Health have been provided with early drafts of the article for quality assurance purposes. These are not market sensitive statistics, and ministers were sent the articles on 13 October, three working days prior to release, consistent with the Code of Practice.

In order to assist me in assuring the quality of the productivity article, I invited a number of experts to participate in a "Quality Assurance Board". The Board was chaired by John Pullinger, Executive Director of Economic and Social Reporting in ONS. Members of the Board were John Fox, Director of Statistics at the Department of Health, Peter Goldblatt, Director of Health & Care Division at ONS, Joe Grice, Executive Director Atkinson Review, Graham Jenkinson, Director of National Income and Expenditure Division in ONS, Peter Smith, University of York and Prabhat Vaze, ONS's Chief Economist.

Like the improvements to estimates of health output published on 30 June, the work on health productivity has been undertaken in conjunction with the Atkinson Review team. The Review has been the vehicle for unlocking expertise and data in departments, including the Department of Health, without which progress on measuring health output and productivity would not have been possible. However final responsibility for the estimates and analysis lies with ONS and, as National Statistician, I have approved both papers to be published on 18 October. As soon as I became confident that the article would be ready for publication, ONS announced the release date for the article in an Operational Note, dated 23 September 2004.

In due course there will be further articles on health and the series will be extended to cover other public services, starting with education. A publication date for the first education article will be pre-announced in the usual way, but is unlikely to be before the new year.

For the next article on health, we intend to develop the sources and methods, in particular reporting on: development of the deflators used with the current price NHS input expenditure figures and the offsets to NHS inputs; further research into corroborating information to improve the triangulation section; and a report on developments following the publication of final report of the Atkinson Review. We will also take account of feedback on the first article.

I shall be making this letter publicly available at 1:30 on 18 October to coincide with publication of the articles.

Yours sincerely,

Len Cook

This page last revised: Monday, 18 October 2004