Room document 6

MEETING OF THE HIGH LEVEL GROUP ON STATISTICS

PARIS 13 JUNE 2002

SALLE 1

Agenda Item 2.5

THE OECD DISSEMINATION POLICY FOR STATISTICS: BACKGROUND INFORMATION

Room Document 6

Room document 6

2.5 THE OECD DISSEMINATION POLICY FOR STATISTICS: BACKGROUND INFORMATION

A. Introduction

1. The OECD Secretariat collects and compiles a wide range of statistics for its ongoing work of monitoring developments in Member countries and in key non-Member countries. Since the Organisation is called upon to develop recommendations in emerging areas of policy interest to governments, it is often requested to develop new types of statistics, new statistical standards and statistical systems. These activities are normally developed in co-operation with national statistical offices (NSOs), central banks, ministries and other government agencies and international organisations.

2. The OECD also disseminates a very large amount of statistics to external users and the most accessed theme in the OECD web site is the “statistics portal”, with about 25.000 visits each month. The external demand for OECD statistics grew significantly over the last decade and the Organisation developed new instruments (for example, SourceOECD) to disseminate statistics as part of a more general renovation of its dissemination and communication policy. These instruments improved the accessibility to OECD statistics and helped to reach new groups of users.

3. At the beginning of 2001 the Statistics Directorate (STD) launched a new statistical strategy and several initiatives have been undertaken in the area of communication and dissemination, in co-operation with the Directorate for Public Affairs and Communication (PAC). In April 2002 the OECD Council has revised the Organisation’s publishing policy and recognised that:

- one of the objectives is to “enhance the credibility of the OECD as a source of timely, relevant analyses, high quality statistics and policy prescriptions”;

- there is a need to improve the value-added of statistical publications in order to safeguard the reputation and credibility of the Organisation;

- dissemination of statistics requires special consideration and called the Secretariat to present concrete proposals for improvements.

4. The adoption of the new publishing policy provides an ideal opportunity to define a new strategy for the dissemination of statistics, taking into account changes that are occurring in this field at national level, and in other international organisations. In particular, the development of the “information society” has changed dramatically the “market” for information, raising the expectations of users about statistics (especially their availability and accessibility), and the competition among private and public data providers. On the other hand, the development of e-government policies has influenced the dissemination policies for statistics in several countries, while new technologies pushed data providers to explore new communication instruments, and to differentiate their policies for different target clients. Finally, the “economics” of dissemination has changed and now the marginal cost of electronic distribution of statistical information is close to zero.

5. This paper is organised as follows. The next section outlines some basic facts about the current OECD dissemination policy for statistics. The third section reviews policies adopted by

Room document 6

NSOs and other international organisations to evaluate the evolution of the “market”. The fourth identifies key technical and organisational issues that currently limit the adoption of a more modern policy in disseminating statistics. Proposals for the development of a new OECD dissemination policy are presented in the final section.

B. The current status of the dissemination of OECD statistics

6. The first issue is the identification of what are regarded as “statistical publications”. In fact, an extensive volume of statistics is disseminated in publications focused on policy analysis or analytical studies. For example, The Economic Outlook has a very large statistical annex, The Science, Technology and Industry Scoreboard also contains a wide range of indicators and analysis based on these, as does The Environmental Outlook. At the same time, some important new data are published in reports which are quite exclusively focused on policy analysis, or in booklets which are not classified as “publications” (see, for example, Transition at a Glance). In addition, the OECD disseminates several methodological publications (containing national metadata) and statistical guidelines and recommendations (Frascati manual on R&D, Oslo Manual, manual on capital stock, etc). Even though these books may not be regarded as statistical publications, they nevertheless represent a key output of the Organisation and are an important adjunct to statistical output.

7. According to the classification used in the framework of the new web site, a publication is regarded as:

- “Statistics, data and indicators” if more than 70% of its content is devoted to statistical tables, data and indicators. Documents concerning Manuals, Sources and Methods, and other publications including projections, forecasts and outlooks are not included unless the 70% threshold is reached;

- “Manuals, sources and methods” if devoted 100% to explain how to use statistical databases (manuals) or how they are compiled (sources and methods) and techniques used in their measurement.

Unfortunately, given the existing monitoring system, the Organisation is not able to provide a clear and detailed picture of publishing activity according to the definitions quoted above. In very rough terms, for 2002 PAC is forecasting gross revenues for publications and other products at M€12 (excluding International Energy Agency). Statistical products (publications, databases, CD-ROMs, etc.) could yield a gross revenue of around M€4 (33% of the total). In particular, 25% of this revenue is expected to be derived from SourceOECD, 25% from CD-ROMs and diskettes, 30% from STD paper publications and the remaining 20% from data re-sellers (Bloomberg, DRI-WEFA and others) and paper publications from other Directorates.

8. The generic user has several options to subscribe to OECD statistical products (one single product, families of products, etc.). Compared to other international organisations, the OECD offers a very wide range of options. The only missing component in OECD “direct” dissemination practices is the possibility of extracting single series (or groups of series) from databases without being obliged to buy the entire database, but this opportunity is offered by several re-sellers.

9. In the framework of the new OECD web site, STD has created a “statistics portal” (Annex 1) in co-operation with PAC, the service for Information Technologies and Network (ITN) and substantive Directorates. In particular, it is now possible to navigate across all

Room document 6

statistical pages following a unique classification, derived from the UN classification of statistical themes (the same classification is also used in the presentation of the OECD Statistical Programme). Full implementation of the “portal” (scheduled for the half of 2002) will permit access to a selection of available statistics for each theme. 10. Two important gateways to access OECD statistics are OLISnet and SourceOECD. These two instruments cover two different audiences: OLISnet the network through which authorised users (including the national statistical offices) access OECD published and unpublished documents, books and data sets; and SourceOECD, mainly designed for “non-governmental” institutional users. For the moment, SourceOECD only contains data files in Beyond 2020 format, while OLISnet contains all data files published by the OECD (Beyond2020 files, MS Excel files, Access to SQL databases, MS Access databases and simple .csv files). OLISnet in particular permits the download of entire files in a format suitable for the bulk updating of a database. In addition, OLISnet permits the presentation of OECD data according to a thematic classification while SourceOECD simply lists the datafiles. Also, the OLISnet interface provides keyword and full text searching for statistics, both leading the user directly to the statistical tables/information requested. Finally, OLISnet also permits authorised users to access “reserved” areas, where Directorates place preliminary or unofficial data to invite comment and to prepare for Committee meetings.

11. The pricing policy for statistical publications has been fixed according to the general policy adopted in 1996 and now has to be revised according to the new publishing policy1. The main target of the OECD publishing activity is to disseminate as widely as possible and in the most cost-effective manner the results of work carried out within the OECD on issues of significant, recognised interest and relevance so as to:

- help build support and understanding in Member countries and, as appropriate, in other countries, for policy approaches and standards that Members are pursuing within the framework of the Organisation’s programme of work; and to,

- enhance the credibility of the OECD as a source of timely, relevant analyses, high quality statistics and policy prescriptions broadly reflecting economic, environmental and social performance within the scope of the Organisation’s Work Programme2.

12. In addition, it is stated that:

- “publishing in the OECD should be consistent with the objectives of maximising dissemination while respecting the necessity for efficient management and cost effectiveness, in accordance with the financial and budgetary regulations and requirements of the Council. Maximum use should be made of online possibilities;

- the Secretariat may distribute selected publications free of charge;

- the decision to publish for sale is based on clearly identified potential or existing markets. The Organisation should set clear targets regarding the level of recovery of costs and investments related to its publishing activities;

1 See document C(2002)80. 2 The creation of a “Fund” for improving quality of OECD publications has been proposed. The Fund should be created using part of the revenues obtained through the publishing activity.

Room document 6

- a flexible pricing policy, set out in the annual publications programme submitted by the Secretary General to Members accommodates the different audiences of the Organisation in order to maximise dissemination while allowing scope for the generation of revenues. Pricing policy is not intended to offset the intellectual costs of a publication, which have been paid by Member countries under the work programmes independently of a decision to publish, or the cost of information activities of the OECD Centres;

- the Secretariat assures a selective free distribution of all priced publications to Delegations, the General Secretariat, specialists indicated by Author Directorates, the media and, subject to reciprocal arrangements, International Organisations. Quota systems established and applicable in all cases are reviewed on a two-yearly basis. Maximum use of online distribution of free publications should be encouraged wherever possible”.

13. The OECD currently applies a restrictive policy in terms of free distribution of statistics to “non-governmental” users. With the launch of the new web site a new policy has been proposed by STD, and endorsed by the Secretary General and PAC. In particular, it has been decided that:

- for each page created in the web site, a sample subset of available data has to be provided free. These data are provided following standards fixed by STD, and utilising software already available at the OECD. To maximise the efficiency of the process, and in agreement with PAC and ITN, STD has chosen a solution mainly based on Beyond 20/20 (the software used for SourceOECD), which allows users to select some variables, length of time series, etc.;

- the amount of data provided free must be limited a maximum of 10% of the overall available data.

The implementation of the “10% initiative” is under way and should be completed by the end of June 2002.

C. The dissemination policies of national agencies and other international organisations

14. The development of Information and Communication Technologies (ICT), the creation of the so-called “information society”, and the launch of national e-government policy initiatives are key factors which explain the main changes to dissemination policies adopted by national statistical authorities and other international organisations. These policies have altered radically over the last decade, with accelerated change to a new vision in this area taking place in the last two-three years.

15. From a technical point of view, the development of the Internet, the increase in the performance of servers and PCs, and the development of software capable of managing very large relational databases, have completely changed the previous technical and organisational environment. All statistical data providers today consider the Internet to be the most important instrument for disseminating statistics, which really comprise two elements: data and statistical methodological information outlining national practices (metadata). Guidelines have been proposed for the dissemination of data and metadata through the Internet and new languages have been adopted or are under development to ensure the best use of this dissemination medium.

Room document 6

16. The availability of powerful servers and PCs has provided new opportunities to place large databases on the Internet. In the meantime, new software has been developed to efficiently create and manage very large databases, containing millions of elementary data or time series. These databases are efficiently accessible through the Internet and data can be extracted using different formats, according to user’s preferences. Private data providers offer tailored products, updated in real time, using a wide range of statistical tools to elaborate elementary data and to release “ready-to-use” products.

17. The development of the “information society” has obliged private and public bodies to adapt their strategies to the new environment. Today, an incredible amount of information (including statistical) is made available by private sources. In particular:

- specialised research institutes and companies collect data on specific areas (market trends, large enterprises balance sheets, agents’ expectations, etc.);

- timeliness is the crucial element and several agencies in all OECD countries produce early estimates (or forecasts) of all main economic indicators;

- through re-organisation of their information systems, large and medium enterprises are able to use internal operational information (sales, orders, etc.) to produce reports and to derive strategies;

- business associations have begun to manage and disseminate statistical information to their members and associates, merging different sources, making estimates, etc.;

- the Internet has dramatically increased the accessibility of information, though very often users still have difficulty in distinguishing between “good” (or relevant) information and “bad” (irrelevant or biased) information.

18. In this context, the usual distinction between “official” and “unofficial” statistics has lost importance. The national “public” data providers (NSOs and other government bodies) and international organisations have to compete for attention on the information market with other data providers in terms of quality. Unfortunately, most users do not consider in a balanced way the various quality dimensions of statistics. For example, timeliness and accessibility for economic statistics are often given perhaps undue preference to other dimensions of data quality (such as accuracy, comparability, coherence and interpretability). For this reason, international organisations and statistical offices have undertaken some initiatives to establish guidelines and minimum standards in disseminating statistics (see, for example, the IMF initiative on Special Data Dissemination Standards). The issue of timeliness has been addressed by all statistical bodies and that has resulted in an impressive improvement in several fields. The issue of accessibility of statistics also needs to be investigated in the context of competition with other providers of data, official and non-official and from the point of view of the final user.

19. Initiatives have been undertaken in all OECD countries to develop public policies to address the issue of the management of an information society. In particular, e-government policies have been launched to support the evolution of national societies, to improve the efficiency of the public administration, to minimise the “digital divide” across citizens, etc. In this framework, statistics play a fundamental role in filling the information gaps between different parts of society and in improving the decision-making processes of public bodies, businesses, individuals. In several OECD countries plans to create unique “portals” to statistical information

Room document 6

released by different parts of the government have been launched. In some cases, national statistical offices have been called upon to co-ordinate these efforts, imposing standards in data and metadata dissemination. More generally, the use of statistical tools, languages and protocols has been encouraged to connect public administrations, share databases, etc.

20. From the perspective of information being a fundamental instrument to improve individual and collective wellbeing, the character of official statistics as a key “public good” element has been reinforced. This vision provided a catalyst for statistical offices to substantially enlarge their distribution of free statistics on the Internet. Even in those countries where rigid and strict rules were applied in disseminating statistics, almost all “headline” statistics are now disseminated free on the Internet. In some countries, this change has been presented as part of the more general national e-government policy and the loss of revenue for NSOs due to this passage has been fully compensated by governments, through an increase of public allocations to statistical offices.

21. In 1999, Statistics Denmark carried out a review of dissemination policies in seven European countries: Denmark, Finland, France, the Netherlands, Norway, Sweden and the United Kingdom. The main conclusions of this comparative study are the following:

- in all surveyed countries publications that can be downloaded from the web are given free of charge;

- in the Netherlands, Norway and Sweden steps have been taken to give first priority to dissemination of publications via the web, second priority to paper publications. This means that the web version is regarded as the “real” publication, the paper version is an extra service;

- PDF, or portable document format, is generally considered to be the most suitable format for disseminating large and static publications containing complex tables and graphs;

- there is a tendency to substantially enlarge the availability of databases free of charge. Databases can be protected by password even if they are free of charge;

- in Sweden (where statistics on users are available), the number of users moving towards free of charge databases increased from 500 (December 1999) to 10,000 (December 2000), and the yearly retrievals increased from 72,000 to 160,000. Preliminary figures show similar tendencies in Denmark (50,000 retrievals in the first month the free access database was launched)3;

- a remarkable change in the composition of users occurred in Sweden after the decision to adopt free dissemination. In particular, ministries, local governments, etc., were the main users during the period of the paid service, while more than 50% of all users are now private enterprises;

- metadata are very important, especially when users retrieve data themselves.

22. Several of the conclusions raised for European countries (Eurostat has recently launched a new survey of all EU countries) are also true for other OECD countries. In particular, experiences in Australia, Canada, New Zealand and the United States show similar results, and strategic plans to adopt new dissemination policies based on the web are under development.

3 In 2001 about 75,000 users accessed SourceOECD.

Room document 6

Finally, it is noticeable that in several countries, in the framework of e-government policies, NSOs have been given responsibility for horizontal projects to improve the electronic dissemination of data produced by other public bodies, following the same standards and policies already applied by statistical authorities.

23. From the standpoint of international organisations in mid-2001 STD reviewed Internet dissemination policies adopted by members of the ACC4. The review showed that all ACC members disseminate at least some statistics via the Internet, though there is considerable variation in both the amount of data disseminated, and the amount of data disseminated free of charge. In particular:

- the International Labour Organisation (ILO), World Bank, UNESCO, World Trade Organisation, UN Economic Commission for Africa (UN/ECA), UN Economic and Social Commission for Asia and the Pacific (ESCAP), UN Industrial Development Organisation (UNIDO) all apply a general policy of free distribution of statistics on Internet;

- Eurostat, Economic Commission for Latin America and the Caribbean (ECLAC) and the

FAO provide extensive information free via limited selection/extraction facilities;

- UNICEF and the UN Statistical Division provide an extensive range of information free via readily accessible static pages, while others provide extensive information in the form of pdf files;

- in a number of instances the Internet site also serves as a portal to extensive subscription specific on-line databases (Eurostat, ILO, FAO, UN Statistical Division, World Tourism Organisation, UNIDO);

- the majority of methodological information (statistical guidelines and recommendations, publications outlining current statistical methodologies used by member national agencies, reference documents, etc.) are normally disseminated on Internet free as a matter of policy. The trend is to make more of this type information freely available on this medium in the future, due to its high public good component, particularly with regard to the promotion of statistical transparency, best practice and the use of comparable statistical methodologies5.

24. Even in those countries which disseminate statistics free on the Internet, books and other more complex statistical products are normally for sale. In some countries, a print on demand mechanism for books and CD-ROMs has been put in place. Furthermore, differentiated prices are applied to different groups of users (public administrations, universities and schools, businesses, etc.). Countries which have begun dissemination of free data on the Internet have also had a simultaneous increase in demand for free data which has spilt over into increased demand for other products, though the net impact of these recent trends on total revenues is at yet unclear.

25. In conclusion, a clear tendency towards the major use of the Internet to disseminate statistics, and towards a policy of free data dissemination through this medium is emerging both

4 The Administrative Committee on Co-ordination is a UN body where all international organisations active in statistics are represented. The STD report commented on the current Internet statistical dissemination practices of 23 international agencies. 5 The OECD already disseminates all methodological information free on the Internet.

Room document 6

at the national and international levels. This tendency is encouraging all NSOs and international organisations to carefully re-think their existing policies6, and to apply the most advanced techniques to improve the accessibility of statistics (and related metadata), as well as the efficiency of the process.

D. Some existing technical and organisational constraints for the OECD

26. As already mentioned, the current OECD dissemination policy is based on the use of several dissemination media. Although the Organisation has implemented some very powerful instruments, some have quite limited flexibility in technical terms, and this has to be considered in discussing possible changes to the Organisation’s dissemination policy7.

27. First, the underlying design of several OECD electronic products have evolved from pre-existing paper publications. Sometimes the electronic database (or CD-ROM) is a by-product of the paper publication, and not vice-versa. As a result, when the original paper “publication” is ready it is forwarded to PAC for printing, and the related electronic products (CD-ROMS, databases, etc) are finalised at the same time in the author Directorate. Although this approach may be suitable for some annual products it is not the case for most. For example, short-term economic databases should be updated in real time (i.e. on the day the data are made available by the national agency), while until now they are only updated for external users once a month or quarter. The same mechanism is needed for datasets where data are updated on a country-by-country basis during the year. It would be better to be able to have the possibility of providing updates for individual countries once they are finalised, rather than waiting for all countries to be finalised. The technical possibility of implementing continuous updating of some datasets for external users already exists, and the adoption of this policy would significantly increase the usefulness of OECD data in competition with other data providers.

28. Secondly, in SourceOECD and in OLISnet it is not currently possible to navigate across datasets, and users are obliged to exploit each single dataset to extract the series of interest. Search engine facilities currently available for OLISnet are able to list all datasets where requested series are available, but the extraction must be done dataset by dataset. This approach can become very cumbersome for users who are looking for a number of relevant series. In addition, because datasets can be created from different sources or at different points in time there is a risk that different estimates be available for the same variable or that users do not find the most updated or accurate estimate for a certain variable. Those problems come mainly from the fact that, at the OECD, dissemination of statistics follows the producers’ point of view rather than that of users.

29. More generally, because each dataset is created independently by author Directorates, there is the possibility of apparent inconsistencies in data available in different datasets. In most cases these apparent inconsistencies are due to statistical reasons, but in other cases are due to the lack of appropriate metadata. The absence of a general “data catalogue”, a number of IT constraints within software currently used, and a shortage of resources are obstacles which make the resolution of this problem difficult8. 6 Eurostat has recently created a “Reflection Group on the future of statistical dissemination in Europe” and the OECD has been invited to join the Group as an Observer. 7 This paragraph is based on the debates carried out at the OECD Statistical Policy Group, as well as the questions put forward by external users to STD Contact. 8 STD has organised, with the involvement of other relevant Directorates, a Task Force to address these issues. Some recommendations have been formulated to improve the presentation of statistics in SourceOECD and to inform users of the sources used to create the different databases, etc.

Room document 6

30. In the past some investments have been made to implement innovative ways to disseminate data. The existing situation is that SourceOECD, OLISnet and the Corporate Data Environment (CDE) currently being developed use different software which obliges author Directorates (and PAC) to follow different processes to disseminate the same data.

31. In the framework of the new OECD strategy for statistics, STD is working with other Directorates to improve the co-ordination in disseminating statistics from a methodological point of view. STD is also working with PAC and Information Technology and Network (ITN) Service to identify the best technical solutions to improve the coherence of, and access to, OECD statistics. It is important to recognise that, even if substantive Directorates agree on the principle of the co-ordination, the available resources for statistical activities are so limited they cannot consider the strict adoption of standards as a priority. Similar problems arise with reference to the analysis (and the resolution) of apparent inconsistencies between data coming from different sources, which other Directorates consider mainly a task for STD. This task is very difficult because today each substantive Directorate is free to disseminate data and metadata using various formats (CD-ROM, Excel files, databases, etc.), even though ITN has tried to limit their number through a process of “moral persuasion”.

32. In conclusion, some of the tools currently used by the OECD are very advanced, others are obsolete and do not ensure the maintenance of a competitive position in the information market. Furthermore, the OECD needs a medium-term strategy to update the tools used for the dissemination of some products, taking into account the behaviour and plans of other data providers, as well as, and most importantly, the users’ point of view on the global OECD data offer. For example, others data providers have developed (or are developing) new database systems which permit better navigation across datasets than existing OECD data dissemination systems. There is therefore a high risk that the OECD will lose its current position (and related revenues) because of a lack of investment.

E. Towards a new policy for dissemination of OECD statistics

33. In order to address strategic and technical issues in the dissemination of statistics an “Ad Hoc Group” was established in February 2002 with the participation of several Directorates. The Group identified key problems for each phase of the dissemination process and its conclusions were discussed at the OECD Statisticians’ general meetings. Some of these problems should be resolved through guidelines that will be provided in the “OECD Quality Framework”, while others require technical and organisational improvements. Table 1 summarises the conclusions of the Group and solutions already identified.

34. From a strategic point of view, the OECD dissemination policy for statistics should meet three different objectives:

- disseminate as widely as possible the statistics collected and elaborated autonomously by the Organisation, adopting high quality standards to facilitate their accessibility and interpretability;

- enhance the credibility of the OECD as a source of high quality statistics reflecting economic, environmental and social performance in Member countries and in selected non-Members;

Room document 6

- contribute to the development of a culture of “informed decision making” at national and international levels, both in governmental and non-governmental bodies.

35. In meeting these objectives, the statistics dissemination policy has to be conducted:

- in the most cost-effective manner, in accordance with OECD general publishing policy and with the financial and budgetary regulations and requirements of the OECD Council. Maximum use should be made of online dissemination possibilities;

- ensuring that the user community in general can have free access to basic statistical information collected and/or originally produced by the Organisation;

- maximising the co-operation with other national and international data providers. In particular, free access to all statistical products has to be given to all national governmental bodies (included national statistical offices), as well as, subject to reciprocal arrangements, to international organisations.

36. To ensure a wide dissemination of the OECD statistics the following actions can be envisaged:

- pursue the free dissemination of “OECD basic statistics” through the web9;

- allow user access to all electronic products (and books in electronic format) from a certain time after their publication (e.g., 12 months for annual publications). Free access should be allowed through the statistical portal;

- develop, by the end of 2002, a general catalogue for all statistical products;

- establish agreements with National Statistical Offices10 to:

include a reference to OECD statistics in existing webs and catalogues. The OECD could draft a mock-up of an ad-hoc html page (in English and French) and a “box” for paper catalogues, while NSOs could translate them (if necessary) into other languages;

promote access to OECD products through the national data shops managed by NSOs;

- establish agreements with national business associations to promote access to OECD products. In particular, promotional html pages (in national languages) could be posted in their websites;

- define a more flexible pricing policy for academics and NGOs;9 Include commonly published statistics for each subject matter. The precise definition of “basic statistics” for each topic has to be done by individual OECD Directorates. In order to make data presented as useful as possible, STD has proposed the following guidelines for the selection of data to be presented on the web:- widest geographical coverage possible;- most relevant subsets of indicators;- longest historical coverage possible, in time series format and seasonally adjusted if relevant and

available, with a minimum of 36 months, 12 quarters or 3 years of data, if possible. 10 This initiative should involve not only NSOs of OECD members, but also those of non-Member countries that are involved in co-operation activities.

Room document 6

- establish an agreement with Eurostat to enhance the distribution of OECD products through the network of Eurostat data-shops. The OECD could also promote the access to Eurostat products through the data-shops managed by statistical offices on OECD non-EU countries.

The following table summarises the approaches that could be used for disseminating OECD statistics and their related pricing policies.

Type of data Updates Channels Pricing policyBasic statistics Concurrent Statistics portal Free

Standard statistics Concurrent SourceOECD Priced (variable for different categories of users)

Concurrent OLISnet (only for the OECD network11)

Free

Old issues Statistics portal Free

Concurrent Paper publications Priced

Methodological documents

Concurrent Statistics portal, OLISnet and SourceOECD

Free*

Paper publications Priced

Customised statistics

Concurrent Service centre Priced

37. To improve the credibility of OECD statistics clear rules of conduct have to be established in the context of the “OECD quality framework” and made available to the public. Confidence by users is built over time and one important aspect is trust in the objectivity of the data. This implies that the data are perceived to be produced professionally 12 in accordance with appropriate statistical standards, and that policies and practices are transparent. For example, data are not manipulated, nor their release timed, in response to political pressure. In the OECD context, the Secretariat has to decide if the publication of poor quality data received from countries affects the overall credibility of the OECD as a high quality data provider. If the answer is affirmative, the Secretariat should refuse to publish the data. Furthermore, it must ensure that, once an agreement between the Secretariat and countries has been reached on collection of specified data, the data subsequently collected cannot be withdrawn in response to political pressure.

11 Included NSOs and other international organisations.*Already applied.12 Principle 2 of the UN Principles of Official Statistics (1994) states: “ to retain trust in official statistics, the statistical agencies need to decide according to strictly professional considerations, including scientific principles and professional ethics, on the methods and procedures for the collection, processing, storage and presentation of statistical data”.

Room document 6

38. An important instrument to build confidence by users is the publication of an advanced release calendar13. A publication schedule may comprise a set of target release dates (at least for key economic indicators) or may involve a commitment to release data within a prescribed time period from the their receipt14. In the OECD context, a publication schedule would help:

- external users, by improving their capacity for timely use of OECD statistics;

- internal users, by enhancing their capacity to plan their work based on the released dates;

- the Secretariat, by enhancing its capability to resist pressure to tamper with release dates for political reasons.

On the other hand, there may be occasions where the OECD cannot adhere to its schedule, for example due to changes in the priorities of the Organisation. These changes should be clearly communicated to users.

39. To improve accessibility and interpretability of OECD statistics specific guidelines will be defined in the context of the “OECD quality framework”, as well as to improve cost-efficiency of data and metadata dissemination activities. In particular, the targets to :

- present data linked to metadata;

- provide full navigability across different datasets;

- provide clear explanations of the reasons for apparent inconsistencies between different data sources15 and support to users in identifying the best data source for their specific purposes;

- minimise the use of different formats for accessing different data sources;

will be pursued.

40. Co-operation with national statistical agencies and other international organisations in data and metadata dissemination has to be reinforced in three main areas:

- because of ICT developments, in a few years it will be much easier to retrieve data directly from the web sites of individual national sources for compiling multi-country statistical tables. Contradictory dissemination and pricing policies adopted at national and international levels risk crowding out one of the two data providers. Therefore,

13 The currently published OECD press releases for key economic indicators already contain the date for the next release.14 Here “release date” refers to the date on which the data are first publicly made available, by whatever medium, typically, but not inevitably the web site.15 Coherence implies that the same term should not be used without explanation for different concepts or data items, that different terms should not be used without explanation for the same concept or data item, and that variations in methodology that might affect data values should not be made without explanation. Coherence has four important sub-dimensions: within a dataset, across datasets, over time, and across countries.

Room document 6

information about future policies have to be exchanged and discussed to identify the best “co-operative solutions”;

- best practices in designing products should be compared and evaluated, in order to offer recommendations for future work to statistical authorities;

- projects to implement common electronic dissemination solutions could be jointly developed by national and international organisations. Some national statistical offices (for example, the Group of Nordic European countries) have already established agreements in this field, while at the level of international organisations no initiatives have been taken until now.

Finally, co-ordination of dissemination policies is of high importance for the co-ordination of data exchanges amongst national statistical agencies and international organisations, as described in the paper on data collection.

41. There are currently several opportunities to discuss technical and political issues related to dissemination, in particular:

- United Nations Economic Commission for Europe (UN/ECE) meetings on “Statistical output for dissemination of information media”, where (mainly) political issues and good practices about dissemination to media are discussed;

- UN/ECE-Eurostat seminars on “Integrated Statistical Information Systems”, where, among other things, technical discussions about IT solutions for dissemination take place from time to time16;

- the Eurostat Dissemination Working Group (DSIS), the objective of which is to develop an European high quality Statistical Dissemination System;

- the round table of National Statistical Agencies called “International Marketing and Statistical Output Database Conference”, which aim is to explore experiences and current issues in developing and disseminating the contents of large and complex statistical output databases by National Statistical Offices. The U. S. Census Bureau will host the next conference in September 2002.

The constituency of these groups is quite heterogeneous and there is no forum where OECD countries can define common policies or technical projects. On the other hand, there are no opportunities where international organisations can discuss and agree to co-ordinate strategies and/or evaluate the opportunity to develop common tools (especially after the announced suppression of the ACC on statistics). One of the possibilities is to re-organise existing fora to address both technical and strategic issues, or to create a new group with a wider participation.

16 The OECD recently joined the steering group and the next seminar, scheduled for 2003, will be a joint UN/ECE-Eurostat-OECD seminar. The OECD will also host a web site to establish a collection of best practices in the field of IT for statistics and to promote the exchange of information in between ISIS meetings.

Room document 6

Table 1 – Issues and possible solutions identified by the “Ad hoc Group on Dissemination of statistics”

Problems Actions already taken New actions requiredWithin the “quality framework”

Outside the “quality framework”

1) Proposition on a potential product1.1) Lack of horizontal discussion about:- the content of the product;- the optimal technical solutions;- the optimal dissemination policy.

a) Compilation of the OECD statistical programme of work

a) Information to the Statistical Policy Group about new projects b) Creation of a Group to help statisticians in identifying optimal technical solutions

1.2) Lack of market analysis for new and existing products

Development of market analysis financed by the Fund for improving the quality of publications

2) Design of the product2.1) Development of different formats for different users

Guidelines on designing statistical products

2.2) Impossible the navigation across OECD internal databases to improve coherence with existing products

a) Creation of the gateway to OECD internal databasesb) Project on “reference series” to allow users the navigation across OECD databases

Implementation of the new OECD statistical information system

2.3) Unclear identification of the “best” instrument to produce products

a) A project has been lunched to identify the existing software for preparing statistical products b) Migration of some publications to Ventura8

Guidelines on designing statistical products

2.4) Lack of guidelines for designing the product Guidelines on designing statistical products

2.5) Lack of resources (time, skills, etc.) to implement optimal solutions

a) Statistical training programme Fund for improving the quality of publications

3) Promotion of the product3.1) Limited promotion to new users in countries a) Re-launch of the “Statistical Newsletter”

b) OECDdirect announces to about 20.000 registrants all new products

Fund for improving the quality of publications

Room document 6

3.2) More flexible pricing policy a) New publishing policy adopted by the Councilb) Policy for free dissemination on the Internet of a subset of available statistics

Definition of anew pricing policy

3.3) Emerging dissatisfaction about the data delivery to national data providers

Definition of a new policy

3.4) Lack of control on re-sellers about quality of data and metadata presentation

Guidelines on minimum standards for data and metadata presentation by re-sellers

Definition of a policy and its implementation in future legal agreements

3.5) Lack of resources (time, skills, etc.) Statistical training programme Fund for investing in publications

4) Production4.1) Delays in the preparation of CD and books Monitoring of production delays for STD products Improve efficiency of

production processes4.2) Unclear relationship between Olisnet, Internet and SourceOECD

Definition of a policy

4.3) Title management and citations poorly handled between different versions of the same product

Guidelines on designing statistical products

5) Customer support5.1) Users call authors instead of PAC for non-technical questions and vice-versa

a) Feedback system via SourceOECD b) centres as local feedback and customer service point for all paid publications and services

5.2) Organisation of ad hoc data extractions Definition of new rules5.3) Lack of a statistics catalogue A person will be hired during summer 2002 to

develop a comprehensive catalogue of paid-for and free statistical services

Preparation of a catalogue for statistical products

6) Feedback from users6.1) No information about the delivery dates New rules have to be

established6.2) Limited feedback to producers Preparation of a quarterly report on sales and

accesses to OECD databases6.3) Lack of systematic market analysis Development of market

analysis financed by the Fund for improving the quality of publications

16