4.3 Organization of national statistical systems#
4.3.1 Centralised vs. decentralised systems for producing official statistics#
The organization of official statistics varies considerably between countries. This depends on the institutional structure of the countries, on traditions and on how the official statistics have developed, in particular with respect to multi-national and international requirements and practices.
Thus, it is up to the countries to choose how many organizations in the national or federal administration should be defined as producers of official statistics; keeping in mind that the main criteria is for each of them the capacity and willingness to comply with the statistical legislation and the UNFPOS.
However, if independent statistical offices exist at the sub-national level, it is advisable to keep the production of official statistics at the federal level as centralised as possible.
The importance of separating the NSS clearly and credibly from the rest of the administration has been emphasized. As one of the key elements of institutional quality, Credibility is necessary to enable users, data providers, and respondents to establish and maintain trust in the producers, the processes, and the products of official statistics. Producers have to establish credibility with respect to the exclusive statistical use of individual data referring to natural and legal persons, as well as with respect to impartiality and professional independence of each producer of official statistics.
The organizational separation between official statistics and policy or administrative tasks is more intuitively credible, the higher the hierarchical level at which it is located. In many countries, the NSO is an organization that reports directly, i.e. without any intermediate hierarchical level, to the President, the Prime Minister or to a minister (see Chapter 4.3.3 – Chief statistician)[1]. At this level, the unique characteristics of the NSO’s status that differentiates it from administrative organizations at the same level are easier to implement and to convey to the outside, because it is an essential part not only of the legislation but of the common culture of all staff, given that official statistics is the core activity of all NSOs.
The situation is more complicated for a producer of official statistics that is a lower-level organizational department of a parent body whose core function is not official statistics, and where the UNFPOS are not part of the common culture of all staff of this parent body. This is especially the case when the statistical department is also in charge of data processing for administrative purposes. Even if this were not the case, the exclusive statistical use of individual data also implies that no such data would be shared for other than statistical use with the rest of the parent body.
This difference is one of the major advantages of centralisation of official statistics at the NSO. The greater the number of other producers, the greater the number of organization-internal borderlines (firewalls) between statistical departments and the rest of their parent bodies that would have to be established at a lower organizational level.
The NSS gains from having, alongside the NSO, a relatively small number of other producers of official statistics. Each of these producers should have a sufficient volume of regular statistical operations to raise their hierarchical level within the parent body and to justify the organizational complications of separating internally official statistical activities from administrative tasks. The statistical department of the Central Bank is such an example.
Another major advantage of a high degree of centralisation of official statistics is the recognition of the brand “official statistics”. The more it overlaps with the brand of an easily identifiable NSO, the easier it is for users to recognise products of official statistics as distinct from the rest of government and guarantee high-quality work based on professional standards. This branding is not easily extended to the output of other producers, which will in most cases be attributed to the parent body, a body that is likely to insist that its own brand is given priority over or at least equal status to the brand of official statistics[2]. The maximum overlap between official statistics and NSO requires that the many statistical products published by various administrative bodies without being official statistics are disseminated so that users immediately realise the difference compared to official statistics. This requirement is also valid for any statistical output of the NSO and of other producers within the system that does not constitute official statistics.
One argument frequently put forward in favour of centralisation is that coordination is easier when more subject areas are under the chief statistician’s authority. However, coordination in substantive terms requires the same input in centralised and decentralised systems, especially concerning the systematic gathering of information about user needs.
The flows of information for the preparation of statistical programmes may appear easier the more activities are concentrated within the NSO, but this may also work well in a decentralised system if all parts perceive a benefit in preparing and exchanging this information. However, monitoring the implementation of system-wide decisions is easier in a more centralised system. Whether coordination is working well or not in a decentralised system depends more on how well the producer in charge of the coordination, typically the NSO, can interact with both governmental users and other producers, and on the extent of the authority of the chief statistician, by law and by his or her professional quality, to set system-wide standards (see Chapter 4.6.2 – Coordination through standards). If this authority were to be absent, decentralisation is rather likely to result in uncoordinated activities of official statistics, as well as in a non-homogeneous interpretation of what are considered official statistics and what are not.
In summary, centralising official statistics in the NSO offers the following advantages:
clearer organizational separation between official statistics and administrative or policy tasks, which is central with respect to the principles of confidentiality, impartiality and independence;
higher overlap of the branding of products and activities of official statistics with the easily recognisable NSO;
easier building (through training) and commitment to a common staff culture specifically based on the UNFPOS;
increased user-friendliness as more subject areas are covered in the dissemination platform of the NSO;
no need for replicating conceptual and methodological know-how in the coordination department of the NSO for areas of responsibility of other producers (see Chapter 4.6.1 – The how and why of coordination);
efficiency gains (up to a certain size);
possibility to resolve co-ordination issues by internal decisions and to monitor implementation by internal follow-up mechanisms;
easier monitoring of compliance with system-wide standards, especially quality management;
simpler relationship between federal and state producers of official statistics;
wider range of data-sets for methods such as data integration and data matching, unless the statistical legislation provides for the NSO to access data sets of other producers at individual level as recommended in the GLOS (see Chapter 4.3.4 – National statistical office);
comprehensive responsibility for international aspects of official statistics.
The three major advantages of a decentralised statistical system are:
higher capacity to provide in-house statistical services outside of official statistics that are relevant for the parent body;
proximity (organizational, physical, and intellectual) of statistical departments of ministries or similar administrative bodies to the subject-matter policy work;
reduced dependence on outputs of the NSO giving a greater autonomy in implementing the data collections that are necessary to produce the official statistics required by the parent body. In the case of the NSO being ineffective in delivering the output relevant for the ministry in time, a statistical department that is part of the NSS has a higher potential to act, at least, as a partial substitute.
The advantages of a centralized system have to be weighed against the advantages of decentralisation as listed before. However, opportunities to systematically reorganize a statistical system across the board will only occur at certain junctures, such as the preparation, adoption, and implementation of a new statistical law, in the case of a general restructuring of the administration, or after a crisis. The opportunities for smaller changes arise more frequently, such as when a parent body of another statistical producer plans a complete reorganization, or when the data collection methodology for a specific official statistics activity undergoes a significant change. In the context of the multi-annual programme preparation, the chief statistician can also trigger smaller changes in the assignment of responsibilities for specific activities to other producers (see Chapter 4.4.4 – The multi-annual statistical programme). A larger organizational change should be based on a professional global assessment of the system (see Chapter 4.2.5 – Relationship between national and international statistical systems), so as to identify the major shortcomings that have to be addressed. In most cases, it is not sufficient to simply reallocate activities between producers, but rather to revise the statistical legislation so as to establish a fully-fledged NSS with all characteristics as defined here.
Economies of scale are an important consideration. They are present in modern official statistics, mainly because of the higher extent of data integration of various sources as opposed to the traditional production in parallel stove-pipes, with one stove-pipe being responsible for one or more data collection activities and the subsequent processing and dissemination of results based almost exclusively on these inputs. However, economies of scale have to be weighed against organization and management difficulties that grow with increasing size of an organization. For this reason, larger countries tend to have functionally decentralised NSSs, but the reverse is not necessarily the case. Many small countries have not been able so far to establish an NSO in charge of producing official statistics in most subject areas, although centralisation in an NSO would not increase the size of the organization to a level where the above-mentioned management problems would show up. For a large group of countries, the history of how official statistics was taken up as a government function still appears to be the determining factor for the present degree of functional centralisation of official statistics, irrespective of the potential of efficiency gains through centralisation.
4.3.2 Governance of national statistical systems#
A national statistical system requires rules of governance of which at least a part is different from the general rules of decision-making in the rest of the administration, due to requirements for professional independence specific to official statistics. However, professional independence does not systematically coincide with managerial autonomy. The allocation of resources for official statistics as part of the overall budget appropriation follows the rules applicable to the entire national or federal administration. The same holds for general rules on accounting and procurement. In most countries, the general rules about the salary grid applicable for the central administration are also valid for staff working in the NSO or with other producers except the Central Bank.
Professional independence requires that all issues on definitions, classifications and methodology, as well as decisions concerning data collection, data processing and analysis, timing and content of all forms of dissemination, are made within the statistical system, without interference from any interest group or government organization. Such decisions should be based on consultations with users and follow international standards of official statistics. The statistical law has, therefore, to address two issues of governance:
Where exactly is the interface between purely professional decisions and decisions that are submitted to a body outside the statistical system such as the government, President, or Parliament?
What is the extent of autonomous decision-making on professional issues for the other national producers, or in other words, what kind of professional decisions require national producers to obtain some kind of approval from the chief statistician?
National practices on these two issues differ quite substantially. Concerning the first issue, the greatest extent of autonomy can be found in countries where the general decision making of government and Parliament only defines the total budget appropriation for official statistics, leaving the allocation of funds for official statistics entirely to the statistical system. This approach can be found in some countries, but it generally applies only to an NSO with a high degree of functional centralisation, especially in countries where the NSO is an autonomous agency outside the main administration (see Chapter 4.3.3 – Chief statistician). It does not really apply to a statistical system composed of an NSO and several other national producers. It shows a high degree of confidence in the chief statistician, not only concerning professional competence but also in terms of priority setting among subject areas. In such countries, the government has no role in the process of adopting statistical programmes.
However, some countries provide at least for some role of decision-making bodies outside the NSS, in most cases the government, basically because of three considerations:
ensuring a balanced priority-setting among subject areas;
providing a higher degree of legitimacy for the necessary data collections for statistical purposes from households, businesses, and other respondents outside the government, especially when such an activity is associated with a response obligation;
seeking to ensure that the information needs and data collection are balanced against the need to limit the response burden.
The exact involvement at the government level in such decisions can take two forms, which may be combined:
either through the adoption of the various statistical programmes for the whole national statistical system (see Chapter 4.4.5 - Authority to take decisions on the programmes), or
by adopting legal acts at government level, based on the statistical (and not any other) law, about a set of statistical surveys or activities, or exceptionally, an individual major survey with response obligation.
These two forms can be combined. The first form addresses the issue of balance and strategic development, and the second form the two other considerations mentioned above. However, when adopting a statistical programme or deciding on data collection activities through the adoption of lower-level legal acts, the government should not modify any methodological and terminological issues that fall under the statistical system’s professional independence. In countries with the NSO as an autonomous agency, most of the objects of decision-making by the government are assigned to the executive board of this agency, thus bringing such decisions to the inside of the NSS.
Concerning the second aspect of governance, the decision making on professional issues within the NSS, it should be noted that producers that together form the NSS are not completely independent of each other. Other national producers, although not subject to the authority of the chief statistician, have not only to respect the full set of provisions contained in the statistical legislation, but also the final system-wide authority of the chief statistician on professional issues, provided this competence is explicitly foreseen in the statistical law. They have to respect and implement in their activities of official statistics any standards that the chief statistician has decided to be valid for the system as a whole (see Chapter 4.6.2 – Coordination through standards).
The dual role of sub-national statistical offices as members of two parallel statistical systems is also reflected in differences of system-internal autonomous decision making on professional issues. When carrying out activities for the federal statistical system, their autonomy is even more reduced than in the case of other national producers, because they must comply with all specifications of the specific activity to which they contribute, and which are fixed by the responsible producer at the federal level (preferably after a participatory process with the sub-national statistical offices). However, concerning their activities as members of the sub-national statistical system, their professional autonomy is defined by the sub-national statistical law and, legally speaking, not constrained by any provision of the legislation at the federal level (see Chapter 4.2.4 – Relationship between national and sub-national statistical systems).
4.3.3 Chief statistician#
Appointment, protection, and reporting
In the great majority of countries, the chief executive of the NSO is also assigned by law the responsibility of chief statistician, i.e. as the leader and coordinator of the whole NSS[3]. In particular, at the international level, he or she represents not only the NSO but the entire national statistical system.
In addition to the managerial responsibilities as chief executive officer of the NSO (further discussed in Chapter 5 - The National Statistical Office, the chief statistician should be made, through a provision in the statistical law, responsible for the professional integrity of the whole system, for securing trust in the statistical system, and not just the NSO. Because of the professional independence of the NSO, the superior of the chief statistician cannot be held politically responsible for any professional errors that may occur within the statistical system. Therefore, a chief statistician does require not only sound experience and professional know-how about statistics from a broad producer or a broad user perspective, but also management and communication skills as it is the case with similar top-level jobs of the civil service. Therefore, this post does not lend itself to be part of a mobility scheme for senior staff in the civil service, through which incumbents would be replaced regularly after a relatively short period. This is one of the reasons why the GLOS recommends that the minimum term of office for a chief statistician should be at least 4 years.
In view of the principle of impartiality, it is furthermore important to avoid that the chief statistician is perceived as directly or indirectly linked to one or the other side of the political spectrum.
This requires a transparent, non-political process of hiring the chief statistician to make the final appointment on purely professional grounds. This applies although the appointing body often is a political one, in most cases, the government. To further strengthen the non-political and impartial character of the post, it is recommended that the term of office does not coincide with the terms of office of the appointing body and that the term of office for the chief statistician should not be affected by changes in the government or come automatically to an end when national elections take place or government changes. Provisions of this nature may not apply to other posts in the administration at the same level[4]. For this reason, it is crucial to include them in the statistical law as an exception to the general rules laid down in some other legislation.
The impartiality and professional independence of the chief statistician also has to be protected to discourage threats of dismissal by superior or other influential bodies. Therefore, it is recommended to include another special provision in the statistical law, which exhaustively lists the possibilities by which a term of office can exceptionally be terminated before expiry. The statistical law should also regulate the number of times the term of office for an incumbent can be renewed. The GLOS, article 6, contains a few specific proposals on this and other aspects of the status of the chief statistician, together with a summary list of responsibilities as mentioned in various parts of this and other chapters.
Because of the importance of having a clear borderline of the statistical system to the rest of the administration, (see Chapter 4.2.1 – Delimitation of the national statistical system), it is crucial that the chief statistician reports directly to the government. In some countries, the chief statistician operates at the ministerial level and reports to the government (prime minister as head of the government or President of the country). However, because of the non-political nature of the post, it is recommended that the chief statistician is not seen as a regular member of the government and only takes part in meetings at government level only when issues of official statistics are on the table. The same prudence is advisable for the participation of the chief statistician in policy committees[5].
Various countries have set up different solutions as to whom should a chief statistician report. The following conclusions can be drawn:
The reporting line should be a direct one to the government level, without intermediate superiors between the chief statistician and a minister;
The reporting line should be a stable one and not redefined every time a new government is formed;
The line of reporting chosen should minimise the perception of political influence and proximity to the general communication function of the government (which in some countries may exclude direct reporting to the Prime minister);
It is also important to avoid the perception that the superior of the chief statistician biases the activity pattern of official statistics in favour of the narrow purposes of his or her ministry.
Whether the chief statistician in such set-ups can participate in government meetings when issues of official statistics are on the agenda, depends on whether the laws regulating the government functions allow such extensions in special cases, not only for official statistics. In many countries, this is legally impossible. If the chief statistician cannot participate personally in government meetings when issues of official statistics are discussed or decided, his or her government minister has to act as spokesperson for official statistics and for the NSO. This is the case for statistical programmes, statistical legislation, and especially for decisions on financial and staff resources as part of the general budget procedure.
Countries have in the past tried solutions other than a member of the government being the superior of the chief statistician. As mentioned before, the first is the government as a whole, but without the presence of the chief statistician at government meetings. Defacto, this has sometimes resulted in reporting to a deputy prime minister, whose status and/or interest in statistics is likely to be limited. Another solution is the President of the country. He or she may have some role in appointing the chief statistician, in order to buttress the perception of impartiality. This may not be optimal for the regular reporting line unless the President is involved regularly in government meetings. In countries where the government is considered highly politicised and partisan, there have been attempts to have the chief statistician report directly to the Parliament or a parliamentary committee. In this case, there is a risk that no designated person feels responsible enough to act as a spokesperson for the cause of official statistics on a continuous basis. The GLOS recommends reporting to the Prime minister or the President if the latter is an operational part of the executive branch. However, many countries have chosen a member of the government other than the Prime minister or President with an important cross-sectional portfolio like the minister of finance.
The chief statistician as head of an autonomous agency
The GLOS mentions this agency model as an option. The main difference with an agency set-up is a decision making and oversight board as the highest level of governance, either specifically for the NSO or the NSS. The exact title of this executive board varies. The statistical law would have to explicitly provide for this solution with all necessary organizational provisions, such as the appointment and office terms for the executive board members. The executive board is meant to take over most of the government’s decision-making powers in matters of official statistics other than the provision of resources, e.g. the adoption of statistical programmes and decisions on the strategic developments of official statistics. In some countries, the executive board also plays a role in the process leading to the appointment of the chief statistician by the government or President of the country.
In this agency model, the chief statistician no longer reports directly to a member of the government (except for general issues such as accounting) but reports to the executive board.
It is recommended that the chief statistician acts as chairperson, or at least as deputy chairperson, of such a board. If a country opts for the solution of the NSO as an autonomous agency with an executive board, the statistical law is very likely to be a law on this agency only, without due regard to statistics produced elsewhere.
If the executive board’s composition is balanced and credible from the point of view of impartiality and independence, such a solution is a solid signal to underline the special status of the NSO in terms of professional independence. However, it implies a very high degree of centralising activities of official statistics in the NSO.
Being outside the main administration under the government, either as an autonomous agency, as part of the presidential administration, or under the Parliament, may also lead to difficulties in establishing and maintaining regular networks with those many parts of the main government administration that are regular users of official statistics or providers of data. This difficulty has to be weighed against the advantages of an agency model. It may also make access to administrative data for the NSO more difficult, in legal, organizational, and practical terms.
4.3.4 National statistical office#
As the major producer of official statistics, the NSO is the backbone of official statistics in a country. Official statistics are the core business for this organization, with notably the following responsibilities:
ensuring continuity of production and dissemination of all statistics under its responsibility;
providing an effective performance of the system-wide functions such as setting standards, compiling multi-annual and annual plans, and coordinating the NSS, as well as providing services for other producers (e.g. sampling);
ensuring quality and efficiency of the production and dissemination processes, based on using the best-suited methods and equipment, and carrying out research to enhance quality and effectiveness;
developing and maintaining networks with the various user groups to assess the relevance and capture new and changing information needs sufficiently early;
ensuring trust of users, respondents, and data providers in its institutional quality and in the integrity of the whole NSS through regular monitoring of adherence to the UNFPOS or equivalent principles across the whole national statistical system;
participating in international activities of official statistics.
The division of work between the NSO and other national producers varies across countries. To avoid duplication of efforts by various producers of the national statistical system, notably concerning data collection, data should be shared among the producers of official statistics, including individual data of statistical units that may be subject to confidentiality. The statistical law should contain explicit enabling provisions to facilitate this system-internal exchange of data, however within the strict limits of the principle of exclusive use for statistical purposes of such data. In countries where the NSS is not yet considered to be established strongly enough to allow the exchange of all confidential data between all producers, the statistical law may allow the exchange of confidential data with identifiers of the statistical units only in one direction, from another producer to the NSO, but not the other way round.[8] In such cases, it is also recommended that one method frequently used in the production of official statistics, the matching at the level of statistical units of data sets from different sources by using identifiers, is the exclusive responsibility of the NSO[9], with the necessity to follow strict protocols.[10]
An important systemic function of the NSO is establishing and maintaining statistical registers, i.e., lists of respondents/statistical units that combine direct identifiers and selected content characteristics for relevant target universes of official statistics.
Other producers should not engage in keeping parallel registers, even if they were to cover only a subset of units. If they need samples of statistical units for their own production, the NSO has the obligation to provide such services. Statistical registers are an exception to the standard practice where direct identifiers should be removed from sets of individual data generated for statistical purposes as early in the production process as possible. Statistical registers use many sources for updating, both administrative and statistical. For this reason, they may differ from related registers, managed for administrative purposes by other parts of the administration, which are based on legal requirements which are not equivalent in content or coverage to the concepts of official statistics. Censuses are a major source for the establishment and a comprehensive update of statistical registers, but other sources should be identified for being used for more frequent updates of at least part of the statistical units between censuses.
There are a few cases where an NSO is made responsible by law for managing administrative records, such as processing individual data for administrative use (by other administrative organizations). Sometimes, data processing for statistical and administrative uses are mixed in operations like the management of registers assigned to the NSO. In both cases, the principle of exclusively statistical use is violated, and therefore should such an activity not take place within the NSS. An NSO in charge of such activities would have to concentrate them in one department that would not, like the rest of the NSO, be part of the national statistical system. The NSO would, in this case, have the same problem of a credible organization-internal borderline or firewall as the other national producers inside their parent bodies. This combination is a potential reputational risk, and it is therefore not recommended, even if it may offer some efficiency advantages. Where such an arrangement is present, ways should be found for the NSO to shift the responsibility of the administrative part of the activities to an administrative body, but to maintain regular access to this data source for the purposes of official statistics, especially as a main source for keeping up to date the statistical registers for which the NSO is responsible.[11]
In most countries, the statistical activities related to the functioning of the system as a whole, as opposed to the production and dissemination of statistics in subject areas, have been assigned to the NSO for which official statistics are the core function. This includes strategic and multi-annual programmes, legislation, coordination, setting of professional standards for the whole system, including setting standards for quality management and monitoring the adherence to the UNFPOS.
However, this may involve a risk of putting the production and dissemination interests of the NSO ahead of being a non-partisan mediator between the different producers of the system; or of the NSO being too weak to exert sufficient authority over other producers of official statistics in terms of quality and adherence to the UNFPOS. For this reason, in a few larger countries, some or all of these functions have been assigned to an organization that is separate from the NSO and has no production function of its own. The earliest example of this model is the US (though for slightly other reasons), but it has been recently followed by the UK and France.
Such separate entities with programmatic and coordination tasks, but without production functions, are the fourth type of members that a national statistical system may include, in addition to the NSO, the other national producers, and the sub-national statistical offices in federally organized countries. Such an extension would have to be explicitly foreseen in the statistical law.
However, NSO staff may perceive the creation of a separate body which, among other tasks, has the role of monitoring their adherence to the UNFPOS as a loss of professional independence and a lack of trust in their professional competence. A less radical approach is to concentrate these activities in a special department within the NSO and assign to it some prerogatives, the essence of which may also be incorporated into the statistical law.
4.3.5 Other producers of official statistics#
Other producers of official statistics at the national and sub-national level must be professionally independent organizational entities, and exclusively or primarily focused on statistical work. The statistical activities can comprise a substantial volume of statistical work outside official statistics, but to qualify as producers within the national statistical system, regular production of official statistics in a given subject area must be part of their responsibility. The NSO may wish to sign bilateral agreements with other producers or parent bodies to increase their personal commitments to see the UNFPOS followed by the statistical department with respect to all its activities of official statistics.
National producers of official statistics should be included in the annual statistical programme when they have demonstrated their capability and willingness to comply with all UNFPOS principles and national statistical system standards. They may also be included at an earlier stage as part of the effort to ensure that they comply with the UNFPOS.
Belonging to the NSS can bring the following benefits:
strengthening the institutional quality, notably the professional independence, of each producer of statistics in line with the statistical law and strengthening the common staff culture within the NSS;
supporting quality improvements and development of statistics;
facilitating professional exchange within the NSS;
increasing awareness and use of common tools and standards as a major element of the common culture of official statistics;
providing a strong mandate for surveys and ensuring access to the necessary administrative data;
providing a higher status as a producer of official statistics in the country with the subsequent labelling of products;
enabling the regulated exchange of individual data for statistical purposes within the NSS;
keeping producers better informed of the latest international developments in statistics.
The GLOS recommends that the heads of other national producers are hired through a non-political process similar to the one recommended for the chief statistician. It may be desirable that the chief statistician is involved to a certain degree in such recruitment processes. Some NSSs may require the chief statistician’s explicit consent for the appointment of heads of other national producers.
4.3.6 Overview of the main types of national statistical systems#
The organization of NSSs can be grouped into the following categories:
A functionally centralised NSO as the only member, including its executive board if placed as an autonomous agency outside the main administration. The Central Bank’s statistical department will also produce official statistics but may not formally be part of a system that comprises all official statistics. In federally organized countries, the NSO acts in this set-up as truly national statistical office and not only as a federal statistical office; there are no independent sub-national statistical offices.
An NSO as the core producer, in most cases located within the national government administration, and a limited number of other national producers, with various degrees of functional centralisation, and with coordination responsibilities for the chief statistician. The Central Bank’s statistical department may be a part of such an NSS in which case this will typically be authorised in the law of the Bank. There are no independent sub-national statistical offices.
In federally organized countries, an NSO with a high degree of functional centralisation acting as a federal statistical office, a very limited number of other producers at the federal level, and independent sub-national statistical offices in most or all territories insofar as they are involved in the production of federal statistics. The chief statistician has the mandate to coordinate the system.
These are the main types found in practice, but there are many national variations of the three basic set-ups, either towards more complexity or, because of a relatively old statistical legislation, through some missing elements. For the second and third type, the Central Bank’s statistical department should be a member of the national statistical system.
Many developing countries face a range of organizational and institutional difficulties to set up a comprehensive and effective NSS. In addition to a lack of resources, they often suffer from the following:
each department and ministry have started to produce statistics in an uncoordinated way, mainly for in-house information needs;
users in ministries have difficulties in specifying their information needs;
it is not clear which data sources are exclusively for statistical use;
there are no or only weak dedicated organizational units for official statistics in these departments and ministries;
statistical staff, both in ministries and the NSO, may not have a statistical background and adequate training;
statistical standards that are valid for all producers of statistics may be lacking, notably on quality management;
neither data nor intermediate results are shared between producers before they are disseminated, and therefore the use of techniques like data integration or data matching is rare;
dissemination to the public is irregular;
there is no clear conceptual basis like the UNFPOS, and statistical laws are either non-existent or old and insufficient.
In such situations, there is a need to plan for enhancing the institutional and organizational frameworks of official statistics (see Chapter 4.4.4 – The multi-annual statistical programme). A development plan should have, as a strategic component, the establishment of an NSS based on a statistical law that contains the clear conceptual basis of the UNFPOS. It should also include an institutional framework for all producers, including the NSO, which is able to strengthen the system and shield the producers from undue interference into professional issues., At the same time, the plan should give a long-term perspective that might attract international donors for supporting the building of all types of capacities and skills necessary for a well-functioning NSS. This is also the opportunity to increase the functional centralisation of official statistics activities at the NSO and concentrate the staff with the required skills in the NSO, especially in small countries.
4.3.7 Including actors from outside the national statistical system#
There may be a need to include in the NSS actors that normally wouldn’t be part of the system for specific tasks and predetermined periods. In such cases, it is necessary to ensure that these actors are incorporated temporarily in the NSS are bound by the country’s statistical legislation and the UNFPOS. This applies in particular to the confidentiality requirements concerning the exclusive use of data for statistical purposes. Such conditions arise typically when an external actor, public or private, is assigned by law or subcontracted for specific tasks related to official statistics, including processing confidential data. An example of the first case is the use of staff from local municipalities in traditional population censuses. Examples of the second case are the sub-contracting of telephone interviews for a given survey to a specialised private company, or the use of an IT centre dependent either on an administrative body or a private company.
In most cases of this type, the rules have to be even more restrictive than simply respecting all relevant provisions of the statistical legislation, by:
including only a limited number of staff and not an organizational unit in its entirety;
limiting the authorised use of data for statistical purposes to the narrow part of a statistical operation contained in the legal act or in the contract;
prohibiting any forwarding of confidential data to other receivers than those prescribed;
excluding any access to such data by non-authorised persons;
handing over all data to the responsible producer at the latest at the end of the mandate.
The legal act or the contract would have to spell out these obligations. The producer listed as responsible in the statistical programme has the obligation to make sure that the involved persons respect these limits. However, the producer still bears the full professional responsibility for the process irrespective of any sub-contracting or delegation based on arrangements of this type. Such restrictions are also applicable to statistical departments of state administrations in federally organized countries that do not qualify as members of the national statistical system if they are involved in the production of a specific activity of federal official statistics.
With regard to IT centres, it is essential that at least the NSO as the main producer has full managerial responsibility for its IT used for the production and dissemination of official statistics. Similarly, the NSO should not be dependent on an IT centre under the responsibility of another part of the national administration for any processing that involves complete sets of confidential data protected by the statistical law[12]. On the other hand, other national producers will depend on the IT of the parent body in most cases. Hence, the responsible producer must ensure that the staff from these non-statistical organizations and their superiors are aware of and respect the obligations that accompany their work on official statistics.