UNITED NATIONS ECONOMIC COMMISSION FOR EUROPE
CONFERENCE OF EUROPEAN STATISTICIANS
Workshop on Strategic Developments in Business Architecture in Statistics
( Geneva , Switzerland , 31 October - 1 November 2011)
FINAL REPORT OF THE WORKSHOP
1. The Workshop on Strategic Developments in Business Architecture in Statistics was held in Geneva , Switzerland , from 31 October to 1 November 2011. It was attended by representatives of the High-Level Group for Strategic Developments in Business Architecture in Statistics (HLG-BAS), and twenty-five expert groups whose work is related to business architecture in statistics.
2. Ms Lidia Bratanova, Director of the UNECE Statistical Division, opened the workshop and welcomed participants. She stressed the importance of addressing the new challenges facing the international statistical community. She introduced Mr Igor Kharitonov ( Russia ), who outlined proposals for a high-level seminar on modernisation of statistical production, to be held in St. Petersburg in October 2012.
3. The workshop was chaired by Mr Gosse van der Veen ( Netherlands ), and the agenda consisted of the following substantive topics:
(i) Overview of the HLG-BAS vision
(ii) From vision to strategy, and the mechanisms to implement it;
(iii) Concrete actions for different groups to implement the vision ;
(iv) Summary and way forward;
4. All background documents and presentations for the workshop are available on the website of the UNECE Statistical Division ( http://www.unece.org/stats/docu m ents/2011.10.hlgbas.html ).
Topic (i): Overview of the HLG-BAS vision
5. This session provided the background for the workshop. Gosse van der Veen ( Netherlands – Chair of the HLG-BAS) presented the HLG-BAS vision, endorsed by the Conference of European Statisticians in June 2011. He outlined some of the challenges facing national statistical organisations, and how they can be addressed by greater collaboration within a framework of agreed standards.
6. The second presentation was from Walter Radermacher (Eurostat) on ‘The ESS vision - from strategy to implementation’. He saw the two visions as complementary, and cited other examples where Eurostat has benefited from active involvement in global initiatives led by UNECE (e.g. population census recommendations), helping to ensure that the results could also be successfully implemented in the European Union.
7. The session continued with a short discussion, issues raised included:
The need for governance mechanisms to implement the HLG-BAS vision – moving to more supra-national governance of activities will be a challenge
The need for urgency, statistical organisations have to make the necessary changes quickly to avoid losing relevance, whilst also maintaining key outputs and time series. There is a risk that statistical organisations become producers of quality assured data for historians and archives, used to retrospectively benchmark "real time" data from other sources.
Product quality, defined at different levels, will be increasingly important.
The impact on the people working in statistical organisations must not be underestimated. Staff profiles and skills will have to change.
Sometimes standardisation work places too much emphasis on perfection, meaning that it takes too long, consumes too much resource, and may be obsolete by the time it is completed.
The support and active involvement of top managers is vital to implement change.
Statistical outputs need to better respond to real-life and emerging issues, to facilitate evidence-based policy decisions. This will increasingly require cross cutting output and analyses.
Process re-design should be seen as a way to improve products, and more dynamic, agile responses to new user needs.
Data are not the only products, methodologies and classifications are examples of other outputs, with different production processes.
We may need to change the way we use classifications, moving from ex-ante to ex-post classification, i.e. classifying based on the data, rather than trying to fit the data to previously established classifications
Communication between organisations and different categories of users needs to improve.
The goals of the vision need to be better communicated beyond the relatively small group of people involved in the different initiatives. Middle managers need to be sufficiently aware of, and committed to, the vision and strategy to communicate it effectively to implementers.
Topic (ii): From vision to strategy, and the mechanisms to implement it
8. This session focussed on how to turn the vision into a strategy, accompanied by clear outputs and goals. The first presentation was by Brian Pink, who proposed that a coordinated response between statistical organisations is needed to deal with the challenges. Standardisation will facilitate international collaboration and will provide larger markets for standard IT tools, both commercial and non-commercial. He presented a number of issues to be considered during the workshop, and to be addressed by the strategy.
9. Marton Vucsan ( Netherlands ) gave the second presentation, which presented a layered architecture for cooperation, and considered the practical aspects of improving international collaboration. Structures and mechanisms for more effective management of collaborative projects were proposed.
10. Issues raised during the discussion included:
The importance of defining the statistical industry and improving ability to recognize and respond to new demands were stressed.
More resources will be needed to accelerate the development and implementation of standards.
The cost of developing new systems and processes should be balanced against the cost of maintaining existing systems and processes, and the constraints they impose.
Common architectures will significantly reduce the cost of implementing common tools.
Many projects fail to make provision for the on-going implementation of their results. A longer-term view is needed. We can learn from the cooperation models developed by the successful collaborations such as PC-Axis and Blaise.
Statistical organisations need to trust each other to benefit from shared developments.
A greater focus on statistical content standards is needed. Content harmonisation should be independent of technological standards, so that it is more sustainable.
Collaboration and communication are much better within layers of hierarchy than between them.
A suggestion to send groups to work together on specific issues for extended periods of time (several weeks) received widespread support though governance and costs could be an issue.
Topic (iii): Concrete actions for different groups to implement the vision
11. This session began with a presentation from Daniel Defays (Eurostat), which detailed responses received from the questionnaire sent out to the working groups to identify their priorities and the existing relationships between different groups. Much of the existing work of the groups seems to be already aligned with the basic goals of the HLG BAS vision.
12. Participants divided into four discussion groups based on the four corners of the common generic industrialised statistics model; GSBPM, GSIM, Methods and Technology.
13. The GSBPM group concluded that:
The groups had a primary focus on three main areas, coping with quality management, customer relationship management and human resources management.
There is a need to improve communication and interaction between expert groups.
A higher-level version of the GSBPM (including non-data processes and outputs) could give a better overview of the work of different groups
14. The GSIM group concluded that:
A simple explanation of the GSIM (and its relationship to the GSBPM) needs to be communicated to the wider statistical community.
GSIM development needs to be accelerated, and a plan needs to be created for its maintenance.
The model must integrate and connect with the business needs.
The model will need to develop iteratively.
15. The Methods group concluded that
When identifying and aligning the expert groups it was clear that some areas are not covered. There is a requirement to improve links between the groups and identify areas of need.
Different organisations have different needs and IT environments, which affect the implementation of plug and play solutions.
Methodology should be strengthened and shared allowing the combining of resources nationally and internationally.
More work is needed on standardising terminology and definitions.
There is a need to identify methodological priorities.
A clear governance model is needed, and should be reflected in the mandates of expert groups.
16. The Technology group concluded that:
The GSIM is urgently needed to facilitate the provision of unified solutions.
There is a need for an over-arching architecture and strategy to avoid patchy solutions.
The group could potentially provide environments for collaboration through technology.
There is a need for greater coordination between the groups and knowledge about them.
17. The second discussion group session was based around the issues identified in the draft HLG-BAS strategy paper, and the presentation by Brian Pink in topic (ii). The ideas reported by the discussion groups included:
Most of the issues are organisational and cultural rather than technical.
The lack of structure and differences in quality in many potential new data sources will be a challenge.
User segmentation can help to identify emerging needs.
Identification of good practice examples of implementing new architectures would be helpful.
The governance of the expert groups should be re-evaluated and goals should be prioritised in line with the HLG-BAS vision.
Users should be helped to produce new products based on statistical outputs.
Communication between groups and initiatives should be improved, and more information should be available to the statistical community.
The balance of data sources should take into account the need to reduce response burden.
The role of non-statistical organisations needs to be clarified in a new definition of the industry in which statistical organisations operate. A wider definition of the industry may require the definition of different quality levels.
In some statistical organisations the understanding of materials in English is limited.
Process re-engineering will require the development of new skills such as business architecture.
There is a need to define the “global statistical system”.
Top management support is essential. Internal management and governance structures will need to be reviewed.
It may be useful to create a separate organisational and IT environment as a laboratory for developing new products
18. Points raised in the plenary discussion included:
The need to clarify terminology and create clear links between the documents presented to the workshop. They should be posted on the HLG-BAS wiki for comments.
The HLG-BAS should consider the scope of the vision and strategy – do they apply to national statistical organisations or the whole statistical system?
Increased use of social networks could help to build trust between organisations.
Standards and quality measures are necessary for new types of data sources, particularly private sector sources.
How to manage data sources that cross national boundaries?
Governance models for collaboration between organisations – should they follow a traditional hierarchical approach or a more organic network approach?
Some synergies exist between expert groups (e.g. between the Statistical Network and METIS regarding the GSIM), but more synergies are needed between different regions. All countries should know what is happening, and have the chance to participate.
Topic (iv): Summary and way forward
19. The final presentation of the workshop, by Gosse van der Veen, summarised the conclusions developed during the workshop and discussed some of the proposed plans for furthering the vision of the HLG BAS. Following a discussion, the workshop participants agreed on the following summary conclusions and actions:
Further develop the HLG-BAS strategy and associated actions
Prioritize the main goals according to HLG-BAS vision
Organise the work of expert groups according to the priorities from the HLG-BAS vision, and improve interaction between groups
Develop the Generic Statistical Business Process Model (GSBPM) or a related model to cover other layers of the statistical architecture
Develop the Generic Statistical Information Model (GSIM) quickly
Develop models for the governance of projects to implement the HLG-BAS vision, and associated business cases. Governance models to be considered include:
o Classic (hierarchical reporting structures)
o Frontrunners with cost recovery
Improve communication to all stakeholders