Comment by Athanasios Orphanides, Governor of the Central Bank of Cyprus, at the Fourth ECB Conference on Statistics: A Strategic Vision For Statistics: Challenges For The Next 10 Years
Frankfurt, 24 April 2008
Considerable wisdom accumulated over the past decade is summarised in this paper. As you might imagine, this makes it hard to fulfil the role of a critical discussant. Instead, I will present a selective summary of the ideas in the paper I find most important - expressing my broad agreement. I will then raise some related questions that remain open.
It is widely accepted that good quality statistics are an invaluable tool for policy making. As noted by Governor Liebscher, rapid financial innovation and globalisation pose several new challenges to the statistical function. The speed of innovation requires new statistics, while recurrent financial stress situations increase the demand for relevant and timely statistics that serve as early warning indicators. Indeed, the recent financial turmoil has highlighted potential gaps in the statistical framework, and particularly the lack of harmonised data on credit risk transfer through structured financial instruments. The need of the central banks to monitor market developments in a timely fashion, to identify potential market disruptions and, hence improve monetary policy and macro-financial stability analysis, has encouraged initiatives for a coordinated approach, possibly even a global initiative. A dynamic world economy demands a dynamic statistical system and central banks, such as the Eurosystem, are the natural competence centres for providing financial statistics.
While the demand for high quality statistics will keep increasing, the drive to reduce the administrative burden put on the economy has led to a high priority given to limiting the response burden. The trade-off between the ever increasing statistical needs and the “fatigue” of the respondents needs to be successfully tackled.
As mentioned in the paper, the Eurosystem statistical function has succeeded in tripling statistical output with basically constant resources, and has generated a wide range of harmonised and high-quality euro area statistics. However, efficiency can be further improved, thereby saving resources and creating room to address the new statistical challenges. I read this to be the key message in the paper.
The authors outline possible avenues to increase efficiency in all stages of the statistical production process - input, compilation and output - based on coordination and collaboration. Some could be implemented in the near future and others in the longer term. Implementation of any potential solution requires further in-depth analysis and investigation, while the interdependencies between them pose additional challenges. The paper covers many interesting proposals. I will highlight some key elements which I find to be of particular relevance for our discussion.
According to the authors, reaping efficiency gains to the full extent would require, among other things, developing new and innovative forms of cooperation and division of tasks both within the Eurosystem as well as with partners outside the Eurosystem, such as national statistical institutes (NSIs) and financial supervisors. The goal should be to collect data from reporting agents only once and share them among different uses to the maximum extent possible. I believe this principle can be universally endorsed.
As regards the cooperation with external partners such as financial supervisors (extra-Eurosystem collaboration), national collaboration models among Eurosystem central banks seem to differ from country to country. Several factors influence the collaboration strategy, including the national legal framework (defining NCB competences). Naturally, those NCBs that have supervisory and financial stability competences are at an advantageous position. In this respect, integration at the euro area level may be time-consuming and may reduce the flexibility required by the users. These potential inconveniences need to be balanced with the benefits.
With regard to the cooperation between NCBs and NSIs, the exact modalities of the division of labour will depend, at least in the medium term, on national realities. (For instance, which institution currently compiles the current account of the balance of payments or the financial accounts for all sectors of the economy). Nevertheless, for the purpose of eliminating duplication in data collection or compilation by the two institutions, it may be necessary to review the division of tasks, based on comparative advantages, at the national level and possibly at the European level through the Committee of Monetary, Financial and Balance of Payments Statistics (CMFB). Most importantly, enhanced cooperation would improve consistency between the different statistics and ensure their comparability and usefulness.
Another important idea stressed by the authors is that the re-use of data is most efficient if information is shared at the micro data level. We all recognise that the collection of aggregated statistics is becoming increasingly burdensome with lead time for design and implementation. Is this sustainable in an era of faster financial innovation with a short “time-to-market” being almost a precondition to provide statistics “fit for use”? Increased use and sharing of micro-data seems especially promising. As the reporting burden is increasing with each breakdown required, pioneer models where the information reported is simple and remains unchanged with new or amended statistical requirements could provide the solution.
Let me only add that I find particularly illuminating the example of the Centralised Securities Database (CSDB) for demonstrating the potential in this direction. It is pioneering in technical, statistical as well as organisational terms and its success may pave the way for similar approaches to be adopted in the future.
Gains in efficiency may also result from the introduction of euro-area-wide criteria for selecting the reporting population (statistical techniques, such as sampling or cutting off the tail and so on). Criteria at the euro area level could lead to a reduced number of institutions that are obliged to report and to reduced cost for the producers of statistics. The resulting cost saving may be substantial in small countries like Cyprus.
However, there may still be different statistical needs across countries as financial integration has not yet been achieved in all fields, especially in retail banking. Also, if we take into account that data may also be needed for national statistical products (e.g. for the national accounts) or where MFI statistical and supervisory reporting systems are integrated, the benefits of such techniques may be limited. The full gains from these approaches can be reaped if the results only need to be representative of the euro area as a whole. As suggested by the authors, if representative national results are also required, the system should be flexible enough so as to allow each country to collect additional information. However, even in this case there should be some scope for reducing the burden.
Further IT harmonisation and sharing of tools is another direction towards greater efficiency. It is clear that advances in technology create new opportunities for developing sophisticated and more efficient and flexible systems; at the same time the collection, production and publication of statistics rely heavily on IT tools.
It has been estimated that a substantial amount of resources, around 30% of total resources in statistics, is devoted to statistical development and infrastructure. This suggests that there is a substantial potential for efficiency gains in this area. Also, in view of the significant level of resources required as a minimum for statistics in any NCB (particularly small NCBs), a pure country-by-country development of the collection and compilation systems for all Eurosystem statistics is unnecessarily costly. This may be mitigated by taking advantage of common tools.
An example of progress in this area is the Statistical Data Warehouse (SDW) which was created by the ECB. The SDW is a comprehensive output database that is now been used by the ECB, the NCBs as well as by external users. For the smaller NCBs that may currently not have an elaborate national database, this central data warehouse offers a potential important benefit, i.e. saving the costs of building and maintaining one. Moreover, some NCBs, including the Central Bank of Cyprus, are already contemplating relying on the SDW if the common database is supplemented with additional and locally needed data. This will become easier as further automated feeding into analytical applications is facilitated.
All in all, I am broadly in agreement with the suggestions in the paper. But not all issues are completely resolved. So I would like to conclude with six questions:
These are some questions where one could imagine the answers are not necessarily clearcut. The precise best approach to pursue increases in efficiency may depend on the answers. Regardless, best effort towards raising efficiency should continue. For efficiency gains could well be the philosopher's stone.