4.6 Coordination of the national statistical system#
4.6.1 The how and why of coordination#
Regardless of the number of producers within a national statistical system, it is necessary to coordinate the activities of the system, for the following reasons:
to ensure comparability between the various outputs, also across subject areas, so that they can be meaningfully related to each other using common target universes, common classifications and unambiguous terminology of concepts;
to avoid duplication of efforts and undue burdening of respondents through collecting data through surveys when similar data exist as part of administrative records;
to ensure that official statistics cover all socially important subject areas;
to enhance the image of official statistics through branding and common release practices;
to ensure that the best possible data are forwarded to international organizations;
to ensure in federally organized countries that the information needs at sub-national level are incorporated efficiently into the national programme to reduce the need for additional statistical surveys to be carried out at the sub-national level.
The scope of coordination should at least cover the scope of the statistical programmes (see Chapter 4.4 – Annual and multi-annual planning and priority setting). In federally organized countries, an effort should be made by the NSO to look for coordination issues and possible efficiency gains also between federal and state-level activities. In some other cases, coordination has to go beyond the NSS activities, and include data from other origins, such as indicator sets required for national policies or by international organizations.
To make coordination work in practice is the responsibility of the chief statistician and the NSO (or the separate body in charge of coordinating the NSS). However, the other producers should not have reasons to perceive this function as an infringement of the NSO into “their” affairs, but as a mutually beneficial activity from which all producers benefit. For the NSO, a useful way of building a positive attitude by other producers is to couple coordination with advice and services like sampling, with sharing information about on-going activities at the international level, and with a participatory process of regular meetings with the other producers of the system to prepare the decisions to be made by the chief statistician.
This participatory process can take place either multilaterally in a process involving all producers if the issue is a cross-sectional one, or bilaterally with the producer directly affected by, e.g. a subject-oriented standard. For the multilateral aspects, the chief statistician may set up a coordination body, consisting of the heads of all producers, that meets regularly and advises the chief statistician. The flexibility of such a coordination body is essential, and therefore detail functional and governance modalities should not be ruled in a legal text.
Co-ordination is not only explicitly mentioned as an obligation in the UNFPOS. Recent decisions at the UN level also require NSOs to explicitly ensure coordination in the production and transmission to the UN of the SDG indicators, produced by various national actors[1]. A similar obligation for NSOs of Member States is included in the statistical law of the European Union[2].
The main mechanism for coordination is the preparation but also the implementation of the multi-annual and annual statistical programmes.
Specific mechanisms such as project groups or memoranda of understanding involving all actors could be set up to ensure that a particular activity is followed closely, and operational steps are laid down precisely. A proactive approach is much preferred to some of the traditional mechanisms of coordination, such as the approval of questionnaires since they become active only at a stage in the process that is too late to consider changes that improve outputs and avoid overlaps.
Other coordination instruments complement the preparation of statistical programmes. They have to find a basis in the statistical legislation, mostly at a lower level than the statistical law itself. Only the chief statistician’s authority, in consultation with other producers of official statistics, to endorse system-wide standards should have an explicit basis in the law itself. Some mechanisms are listed below:
Standards on the implementation of the UNFPOS;
Content-oriented standards;
Methodological standards;
Dissemination standards;
Standards on metadata, documentation, and archiving;
Standards on statistic-specific processes, notably quality management;
IT standards, including standards on data security;
Approval-based instruments outside the statistical programmes;
Staff-oriented instruments.
However, there are statistical laws that are silent about the system-wide authority of the chief statistician. This case, frequently found in developing countries, is equivalent to a purely voluntary approach to coordination, leaving the decision on respecting a system-internal standard to each producer. Even if the NSO is in charge of preparing statistical programmes, the chief statistician would not have the authority to demand changes in activities proposed by other producers for inclusion in the statistical programme. In this situation, bringing as many activities of official statistics as possible under the line authority of the chief statistician is beneficial, if the legal basis cannot be upgraded in terms of the system-wide authority of the chief statistician. Given the growing interrelationships between statistical activities, such as the same set of data can be an input for more than one statistics, the need for better coordination is essential.
4.6.2 Coordination through standards#
Coordination through setting standards that are valid for the whole statistical system is the second most important coordination tool. Although the chief statistician has the final authority to decide on such standards and their updating, all relevant producers must prepare these standards before they are formally adopted and generally made applicable through the participatory process mentioned in the previous section. Some countries’ legal tradition may require that such standards are also issued as lower-level legal acts, but this is better avoided. If a co-ordination issue involves only one producer in addition to the NSO, agreements or memoranda of understanding between the two organizations may be a good option.
Setting standards, especially content-oriented standards, is greatly facilitated by adopting the standards established at the international level to the maximum extent. Recently, the international statistical community has also developed international standards concerning metadata, data transmission, business architecture, or access to confidential data for research purposes, including in some cases national examples of good practices. The references to such international and national standards are given in the respective chapters of this handbook.
Not all UNFPOS principles need explicit standards to assist producers in respecting them. The principle that may be the first candidate for standards is the principle of confidentiality, notably for the transmission of confidential data between producers and the scientific community (see Chapter 4.5.5 – User access to confidential data for their own statistical purposes). Other elements for standards of this type are the suppression of small aggregates with a disclosure risk of information about a single statistical unit that could be identified indirectly, the handling of identifiers in the different forms of data collections, and access to and use of statistical registers. Equally important as setting standards is establishing a mechanism by which controversial issues related to one of the UNFPOS can be discussed and decided. Such a mechanism would apply both internally to the NSO and between producers. Over time this may allow a kind of collection of case law that serves as a model for treating similar cases. Cases that have a wider bearing for users and the public can also be brought before the SAC or the executive board.
The content-oriented standards comprise classifications, terminology, and definitions to be used across all official statistics areas. The purpose is to ensure that the same term is used for one concept and that different terms are assigned to different concepts. A corollary of this task for the NSO is to establish and maintain a glossary of terms and definitions used in official statistics and to work out conversion keys when important classifications undergo changes. In some cases, the national standard-setting consists mainly of fixing the terminology in national languages that corresponds best to the terminology used in an international standard.
Classifications such as the economic activity classifications are fundamental to ensure the coherence of statistics from various data sources in a subject area like economic statistics. When a national statistical system adopts international classifications, national specificities can be incorporated, but preferably in such a way that they do not form an obstacle to international comparisons. This can be a problem when these national specifications reflect legal or administrative categories used in non-statistical legislation, especially when these categories change frequently. Generally, it has to be avoided that classifications used in the NSS and crucial for international comparisons to become dependent on categories fixed in national non-statistical legislation. On the other hand, national users may have a legitimate need for statistics that reflect the nationally relevant categories. The considerations given in Chapter 4.5.3 – Interaction with user groups outside the statistical advisory council: capturing their information needs concerning the use of official statistics for administrative allocation purposes are applicable also in such cases to satisfy both types of user needs.
Methodological standards in the narrow sense refer to issues like sampling, treatment of non-response, imputations etc. It is more important that the methodological department of the NSO advises the rest of the NSO and other producers than to issue rigid standards. However, all sample surveys of official statistics should use, when applicable, the statistical registers managed by the NSO as sampling frames.
Dissemination standards are related both to standards about the implementation of the UNFPOS and those on processes. One example related to impartiality in dissemination is setting clear rules to ensure that all users, including those from government or parent bodies, have simultaneous access to results when released and ban any pre-release access to anybody outside the statistical system. Other examples for dissemination standards are about using the label of official statistics, a common internet platform for all results of official statistics, managed by the NSO but open to all producers, or the metadata that must accompany results (see below). A common dissemination platform for all results of official statistics from the entire NSS in the responsibility of the NSO is a very efficient and user-friendly tool to ensure harmonised dissemination by all producers, including metadata.
A crucial part of the standards in official statistics is related to metadata and documentation. For both system-internal purposes and access by users, documentation about sources, methods, and definitions is essential to honour the principle of transparency. The NSO, in cooperation with the other producers, should establish guidance or rules in this respect, both for accompanying sets of individual data from which results of official statistics are generated and accompanying results that are disseminated. The NSO should also be available for advice on how to apply such rules in particular circumstances. Together with the national archive, the NSO should also work out standards for archiving data sets and products from official statistics. However, data sets should be available within the statistical system for some time to facilitate processing any request that would come up at a later stage.
Another important set of standards refers to processes common to the production and dissemination of many official statistics activities. Examples of such standards are:
collection of information about user needs as an input into draft statistical programmes;
quality management, and definition of quality criteria for results to be released as official statistics;
revisions of existing series, e.g. due to new benchmark data such as censuses, or the implementation of a revised international standard also for past periods;
correction of errors in published results;
testing of new or modified data collection vehicles, notably questionnaires;
transmission of data to international organizations.
Concerning data transmission to international organizations from national producers, the NSO has to define a process by which it is ensured that the most authoritative national data are transmitted, together with the necessary metadata. This applies particularly when the data is not simply a subset of a regular production process within official statistics, or when the transmission involves the selection and transmission of data across different areas of responsibility within the NSS.
4.6.3 Operational coordination by the chief statistician#
The implementation of internal NSO activities follows the management rules of the NSO, which foresee periodic reporting to and approvals by different levels of the hierarchy at appropriate junctures. These are questions of the internal organization of the NSO addressed in more detail in Chapter 5 - The National Statistical Office.
It is the producers of official statistics’ responsibility to implement the activities and deliver the statistical results assigned to them in the statistical programmes. Where interdependencies exist with activities of the NSO, or any other producer, regular cooperation, collaboration and exchange of information should occur. But formal interventions or explicit approvals by the Chief Statistician are not recommended generally for operational activities conducted by other producers. This could undermine the sense of responsibility of individual producers and cause unnecessary delays. An exception could be the adoption of new terminology and definitions for statistical results that the chief statistician has not endorsed. Consistent terminology and definitions should be used across the NSS. However, the operational coordination through the chief statistician’s approval should not delay, more than needed, the overall production process and eventually, the dissemination of the results.
A traditional approach of coordination is the approval of forms for all data collection for statistical purposes by the NSO. In the modern way of producing official statistics, where many sources are combined for producing results, such a formal approach is unlikely to produce value-added. Piloting a new or modified survey with a small set of respondents would probably contribute more to the improvement of questionnaires than a formal approval by the chief statistician.
4.6.4 Coordination through staff policies#
The set of shared standards, rules of procedures, and guidelines constitute a key element of what is referred to as a common culture of official statistics in a country. They transform the intentions of the statistical legislation into operational rules that respond to issues most staff would be confronted with in their respective duties. However, staff will not automatically use such standards if they are not aware of them, or if they remain too abstract to serve them in their daily work. Therefore, training staff on when and how to use the various standards, rules of procedures, and guidelines is essential and can be considered the third important coordination mechanism. Most NSOs organize such training for their staff, but staff from other producers at the national and sub-national levels should be encouraged to participate in these training events for official statistics. In particular, in decentralised NSSs, the statistical departments of other national producers of official statistics are confronted with the cultural environment of their respective MDAs and such training courses mingling, statisticians from the NSO and the other producers of official statistics may contribute to the development of a common official statistics culture within the NSS.
Some countries with a substantial number of other national producers use a system of exchange or rotation of staff between producers to ensure that the culture of official statistics is effectively spread to all members of the national statistical system.
If this takes place at a sufficiently high level, it is an effective way of strengthening the notion of a system of official statistics as a single-family. However, this would not replace other coordination instruments necessary at a more operational level.