Towards a new Statistics Netherlands
blueprint for a process-oriented organisation structure
Statistics Netherlands is currently preparing for a fundamental reorganisation. This paper sketches the outlines of the new structure, which is due to be implemented autumn 2000. The new organisation offers better opportunities for standardisation and integration of statistical processes, and hence is expected to have positive effects on efficiency of processes and coherence of products.
Key words: Statistics Netherlands, organisation structure, statistical process, information technology
The current structure of Statistics Netherlands is characterised by rather isolated statistical processes, carried out in a number of more or less independently operating business units. Each of these units covers a particular statistical subject-matter area and is responsible for most of the stages in the statistical process cycle itself. This ranges from monitoring user needs, via data collection and processing, to publication and dissemination. Central steering is kept to a minimum.
Gradually, the insight grew that this model entails a number of artificial cuts which obviously disrupt "natural" coherence among statistical processes and among statistical products, and which hamper the full use of economies of scale. Moreover, there are drawbacks for respondents and users, the former bothered by irrelevant questionnaires sent from different places in the Bureau, and the latter forced to gather their data by shopping around, only to eventually find out that the data do not fit in with each other.
The drive to knock down the walls and to unite what logically belongs together was further strengthened by two external factors. First, the emergence of advanced IT tools encourages the streamlining, standardisation and integration of the numerous processes, and thus urges the shift from a subject matter to a process oriented structure. Secondly, Dutch government recently imposed annual budget cuts of 10-15% on Statistics Netherlands, to start in 2004 and with the added requirement that neither the quantity nor the quality of the output will suffer.
In the summer of 1999 this culminated in the decision to reorganise Statistics Netherlands. The outline of the new structure has now been established far enough to sketch a broad picture of the organisation, as due to be implemented in October 2000. Here this sketch is preceded by an elaboration of Statistics Netherlands’ view on the statistical process and the role of IT tools in this process. This view actually provides the conceptual foundation for the new organisation.
2. The Statistical process
The standard statistical process can be conceived as a cycle that is essentially similar to the production cycle of a composite industrial product. The following stages and steps apply:
A. Design stage
1. Exploration of user needs.
2. Exploration of production possibilities and cost.
After balancing 1 and 2:
3. Product design, resulting in a blueprint of the final product (i.e. the framework of a publication; a set of - empty - tabulations).
4. Specification of the production process:
- input: choice of administrative sources; set up of surveys; questionnaire and sample design; rules for straightforward/administrative editing;
- throughput: rules for advanced/statistical editing; methods for imputation, translation (bridging the gap between input and output concepts); estimation; integration.
B. Implementation Stage
- data collection / use of administrative registrations;
- administrative editing;
- matching administrative sources and surveys.
- statistical editing; imputation; translation; micro-integration;
3. The role of IT tools
New information technology provides tools for both for storage and processing of data. At present four main databases are being developed, each representing a particular moment and demarcating a particular stage in the statistical process:
1. The input micro-database BASELINE, holding all data as supplied by data sources, be they administrative registers or respondents from direct surveys. BASELINE eventually shows the data after straightforward/administrative editing and mapping on statistical units.
2. The output micro-database MICROBASE. Actually, there are two databases of this type; one for per-sons/households and one for enterprises/institutions. MICROBASE contains the data as they result from imputation, translation and micro-integration.
3. The output aggregate database STATBASE. This database contains the results after estimation for populations and sub-populations of statistical units, and eventually after macro-integration. It can be claimed that STATBASE holds all publishable data produced by the Bureau..
4. The publication data warehouse STATLINE. This can be seen as a set of views on STATBASE: it shows the total output of Statistics Netherlands as a structured set of multi-dimensional tables (data cubes), each covering an area of societal interest. The contents of STATLINE are published on cd-rom and via the Internet.
The four databases are ranked above according to their role in the implementation stage of the process. Notice that in the design stage the order is the other way around: agreed user needs are first represented as - empty - STATLINE cubes and specified in STATBASE in terms of metadata on statistical concepts and classifications. Subsequently, corresponding meta-data are stored in MICROBASE, and finally input concepts are derived and stored in BASELINE.
Figure 1 shows the joint design and implementation stage as a complete cycle:
4. The new structure: top level and staff services
The new organisation comprises three main levels: at the top level the executive board, at the second level four divisions and at the third level twenty departments. Appendix 1 shows the organisation chart, with staff numbers as foreseen after the reduction by 300, as imposed by the government.
The Executive Board consists of three people: the Director-General and directors for statistical policy and statistical processing respectively. The Board’s main tasks are strategic decision-making for the Bureau as a whole and day-to-day management. Corporate issues are dealt with in the Management Committee, i.e. the Executive Board plus the directors of the four divisions.
Four centralised staff units support the Executive Board, namely:
A small policy staff will support the Executive Board in the preparation of policy related issues, such as statistical programming, international relations, the secretariat of the Central Commission of Statistics and corporate communication.
Human Resource Management (HRM)
This unit develops and controls an HRM strategy in terms of career planning, management development programmes and in particular training and education. For the latter purpose an in-house School of Statistics will be set up to provide systematic statistical training facilities and programmes. All personnel staff belong functionally to the HRM unit, although in operational terms most will be assigned to Division staff units.
Planning and control
This unit is responsible for the preparation of the annual budget and all reporting requirements, both internally and externally. Its main tool is the recently introduced management information system BISNIS, in which working hours and expenditures are accounted for on a project basis. Transparency of principal-contractor relationships is crucial to the system.
All administrative systems of the Bureau, including financial matters, fall under the responsibility of this unit.
5. Divisions: the basics
Production versus R&D
A first distinction refers to production versus R&D. There is a separate division for the latter, which will also incorporate facilities such as domestic matters, printing and documentation in addition to information technology and statistical methodology. This division is called Technology and Facilities (TNF).
With respect to the traditional dilemma of the benefits of concentration of specialised knowledge on the one hand and the need for involvement of R&D staff in primary working processes on the other, a compromise solution has been adopted. Functionally, all IT and most methodology staff of the Bureau are assigned to TNF. In operational terms, staff involved in the development of tailor-made IT tools and in the day to day control are placed in the production divisions. For general methodological issues and general tool development, often carried out in large projects, there is a pool of methodologists and IT specialists, about half of whom are involved in the direct support of the production divisions.
Input versus throughput versus output; business statistics versus social statistics
Within the production sphere, similarity of processes is the leading criterion for distinguishing the three intended production divisions. The initial inclination was towards the well-known delineation between input, throughput and output, with the databases BASELINE and STATBASE as logical and physical demarcation points. However, further study revealed that the input and throughput processes for data on persons and for data on businesses have little in common, while on the other hand the input and throughput processes within the two populations proved much more coherent.
These findings formed the basis for the decision to set up separate divisions for Business Statistics (BES) and persons-oriented Social and Spatial Statistics (SRS). The spatial statistics were included in the latter division mainly for reasons of balance with respect to size.
As there were also stages in the process that cannot be assigned to either of these two divisions, a third production division was needed to accommodate cross-division issues, of which statistical programming, macro-integration and publication are the most obvious. Thus the division Macro- economic statistics and publication emerged.
6. The divisions and their departments
The four divisions break down into a total of twenty departments, each described briefly below. All the divisions have a Support and Development department, with high-skilled staff responsible for specific projects on statistical and automation issues and financial and administrative matters. Some of these staff are seconded from pools, which functionally belong to other units.
6.1 The Technology and Facilities Division (TNF)
The goal of this Division is to ensure the quality and effectiveness of statistical processes and products, both through innovation and by supporting the production divisions. In the light of the specific purposes and conditions of the current reorganisation, the focus is on integration of statistical methods in application tools, as well as on standardisation of these tools.
The division comprises four departments:
- Methodology and Informatics exercises the human resource management for three staff pools. The R&D pool consists of researchers and project managers, half of whom support the production divisions. All staff members from the pool for application development and management are assigned to the production divisions. Lastly, there is the trainee pool, consisting of recruits with academic or senior vocational qualificationss who are involved in a diversity of projects and thus become familiar with a variety of statistical skills.
- Information and Communication Technology is responsible for the organisation and maintenance of the technological infrastructure. There are sub-departments for front-office services, back-office and research.
- The Facilities department provides services with respect to the building, office space, printing and copying, documentation management and purchasing.
- Support and Development develops policies and implements projects with respect to corporate quality management.
6.2 The Business Statistics (BES) and Social and Spatial Statistics (SRS) Divisions
BES compiles statistical information on the activities of companies and institutions, or groups of these, as well as on business-related aspects such as technology, energy and the environment. SRS collects information on the demographics, activities and quality of life of persons and households, and on spatial issues. As their internal structure is essentially similar, the two divisions will be discussed simultaneously here.
In terms of the logical process description outlined above, the goals of these sister divisions can be phrased as follows:
1. fill STATBASE with publishable data that comply with concepts and classifications as developed in co-operation with division MSP. "Publishable" means comprehensive, reliable, consistent and non-confidential;
2. contribute to the compiling of publications, to be issued by MSP;
3. provide MSP with input for macro-integration.
Within the Divisions, the predominant distinction is that between data collection (from external sources) on the one hand and data analysis on the other, with the input database BASELINE as the demarcation point. The former demands skills that are primarily oriented towards the world of the data sources, whereas the latter is more subject-matter oriented. There is a further distinction with respect to data collection, i.e. according to type of source: existing registers versus own surveys.
These considerations result in a basic structure for each of the two divisions as illustated in Figure 2.
Both divisions have a Registers department, with the following tasks:
- to map the data derived from a variety of registers on the relevant statistical units (per-sons/households/dwellings for social data; establishments/companies for economic data), and to enter units and data in BASELINE;
- to check the data for straightforward administrative errors;
- to supply sampling frames for data collection (i.e. for the Surveys Department).
There is one Surveys department in SRS, and two in BES. Their main tasks are:
- the development and execution of surveys, according to designs as provided by the Statistical Analysis Departments;
- editing. The scope of this may vary with the type of statistics concerned. For specific subjects (e.g. the environment) the editing is limited to obvious errors, while for general subjects (e.g. turnover) editing includes relational checks;
- supplying the data to the Registers Department, which maps them on the statistical units in BASELINE.
To satisfy the one-counter principle, the departments are structured in such a way that respondents are contacted by one sub-department only, whatever the type of survey. For the 250 largest companies, this includes the profiling of statistical units and their attributes.
Statistical Analysis Department
Each of the two divisions has two Analysis Departments. They are the linking pins with the MSP division, both in the design and implementation stage of the process:
- in the design stage, they put the working programme as established by MSP and as embodied in the meta-compartment of STATBASE into practice;
- in the implementation stage, they supply MSP with:
- inputs for publications, by filling the data compartment of STATBASE, and by contributing to publication texts;
- inputs for macro-integration.
The Analysis Departments bridges the gap between incomplete and inconsistent data on the input concepts of BASELINE and the complete, consistent, reliable and non-confidential, thus publishable data on the output concepts of STATBASE. They do this by statistical editing, imputation, translation, micro-integration, entering in MICROBASE, assigning weights, and lastly estimation.
6.3 Division Macro-economic Statistics and Publications (MSP)
In general, MSP has the responsibility for stages in the statistical process that surpass the scope of the Divisions SRS and BES. The division embodies both the alpha (statistical programming) and the omega (publication) of the statistical process cycle:
Department Prices, Short term indicators and Programming
This department actually initiates the statistical process cycle. Based on user needs, among which those of strategic relations have a paramount position, the department prepares Statistics Netherlands’ statistical programme, which is subsequently discussed in the Programme Council (in which all divisions are represented) and the Advisory Committees and is eventually authorised by the Central Commission for Statistics. Lastly, the department puts the programme into practice in terms of entering metadata in STATBASE. As such, STATBASE entails the specification of the intended output, to be produced by Divisions BES and SRS.
The concentration of all output-metadata in one central database provides promising opportunities for the long cherished goal of standardisation and harmonisation of concepts and classifications. Statistical co-ordination is therefore an explicit task of this department.
Besides programming and co-ordination, the department compiles price indices and short term indicators
Department Information services
This department is responsible for the publication and dissemination of all statistical information, wherever it is produced at Statistics Netherlands. It comprises a publishing unit, an editorial unit, a communication unit and an information service unit.
Obviously, the output data warehouse STATLINE is the hub of the wheel of core activities of the department. STATLINE is one publication, composed of a number of data cubes, each describing a certain theme or sub-theme, together covering the whole statistical programme. All STATLINE tabulations are based on STATBASE; thematic cubes are actually structured views on STATBASE. Themes are "owned" by editors, who are responsible for cube design and delineation, as well as for all other publications in their area, including relevant parts of the Internet site. The actual preparation of cubes and publications is the common responsibility of editors and data suppliers from the Analysis departments in the BES and SRS divisions.
As STATLINE is to comprise all published data in a structured and easily accessible form, it is also the proper data source for the central information service. Only in case of very specific questions, will clients be referred to subject-matter specialists.
Department of Financial Institutions and Government
The macro-economic character of these statistics and the fact that they serve similar user groups, led to the decision to place these statistics in the neighbourhood of the national accounts.
National Accounts Department
This department produces a variety of cross-division integrated statistics, most of which are based on data entered in STATBASE within BES and SRS. National Accounts will benefit greatly from the impact STATBASE is expected to have on overall standardisation of concepts and consistency of data.
At the time of writing, the design of the new organisation structure has not yet been finalised. Some of the issues presented in the paper will be discussed and worked out further. Therefore, the views expressed here are preliminary and subject to change. This is particularly the case for:
- the delineation of tasks between the Surveys and Analysis departments in the BES and SRS divisions with respect to editing and micro-integration;
- the extent to which the one-counter principle can be actually adhered to, in particular with respect to data collection and editing of business surveys;
- the extent to which statistical processes for different subject matters can be actually integrated;
- the degree of centralisation of the publication and dissemination function in the Division MSP.