Bericht

In central banking decisions have to be made in real time with a measure of uncertainty. A sound statistical system helps to reduce the level of uncertainty and so lower the likelihood of suboptimal decisions.

The first decade of Eurosystem statistics has been very successful. First of all, the necessary statistical base for the common monetary policy has been set up in time and “fit for purpose” for the launch of the euro. Over the following years, much has been achieved in terms of euro area statistics in order to close data gaps and to cover new data needs. But the Eurosystem cannot rest on its laurels. New data needs are emerging as the financial sector develops very rapidly, generating new institutions, markets and instruments. At the same time, while the first decade was mainly characterized by the necessity to get things done in time, the second decade will put more emphasis on how to do things in the most efficient way. This will require – among other things – developing new and innovative forms of cooperation and divisions of tasks both within the Eurosystem as well as with statistical institutions outside the Eurosystem.

Modern central banks¹) are both extensive users and producers of statistics. Statistics are a core input for all policy decisions by central banks and are essential for successful central banking. At the same time, central banks have the expertise and the comparative advantage in producing statistics on the domestic as well as cross-border financial aspects of the economy – compiling and distributing statistics are core tasks of modern central banks. The same applies to the Eurosystem with its harmonized but decentralized model of statistics production.

The first decade of Eurosystem statistics has been very successful. First of all, the necessary statistical base for the common monetary policy has been set up in time and “fit for purpose” for the launch of the euro. Over the following years, much has been achieved in terms of euro area statistics in order to close data gaps and to cover new data needs. This has resulted in approximately a tripling of the data output with basically unchanged human resources. Data releases are a prime source of information, not only for policy makers but also for the market participants trying to understand and to anticipate Eurosystem policies, witness the more than 130 million hits per year on the statistical websites of the Eurosystem central banks. After all, statistics, including central bank statistics, are public goods that should be accessible to all. They are an important element of the public accountability of central banks and a solid base for the evidence based policy making that the public increasingly expects.

But the Eurosystem cannot rest on its laurels. The statistical challenges of the second decade need to be identified and faced. New data needs are emerging as the financial sector develops very rapidly, generating new institutions, markets and instruments. Especially the current financial turmoil has highlighted important statistical gaps with respect to new instruments and institutions. At the same time, the data requests face increasing opposition by the reporting entities and their representatives. Thus, while the first decade was mainly characterized by the necessity to get things done in time, the second decade will put more emphasis on how to do things in the most efficient way in order to free the necessary resources for fulfilling new data needs and to accommodate the statistical challenges of further enlargements of the euro area. This will require – among other things – developing new and innovative forms of cooperation and divisions of tasks both within the Eurosystem as well as with statistical institutions outside the Eurosystem. We will need to enter unchartered waters and create pioneer models for these and other tasks of the system.

At the same time, the last decades and especially the last few years saw a dramatic expansion in the size and importance of the financial sector. Financial assets are growing much faster than the real economy, both domestically as well as cross-border. Therefore, the relevance of financial market developments – both as drivers of growth and potential risk factors – has been rising dramatically. Policy makers both inside and outside of central banks increasingly have to deal with financial issues.

But financial issues are not just a matter for policy makers but rather for all parts of the economy and society. Households have been accumulating financial assets at an unprecedented speed, partly to compensate for the decline in expected future public pension entitlements. Growing financial assets and wealth will not only increase the importance of financial markets but also of asset prices, risks and risk mitigating instruments as people increasingly move into riskier assets.

All these developments require an adequate level of knowledge about financial markets, financial products and financial institutions. The basis – the conditio sine qua non – for this knowledge is reliable quantitative information, i.e. relevant statistics. Monetary as well as financial stability require good policies, good policies require good analyses, and good analyses require good statistics. But good statistics require a constant flow of reliable reporting by the market participants. And that is where increasing tensions arise. Central banks producing statistics see themselves confronted with another “incompatible triangle” of more, faster, and cheaper. The dramatic expansion of financial innovations results in rising information needs for policy; the speed of communication and the resulting need for fast policy response mean that data should be available almost immediately; and finally, those that have to supply the basic information, the respondents, see that as a cost factor and become increasingly “fatigued”. In this paper we try to develop some options for dealing with this incompatibility in order to reduce the tensions. We will first address some of the future information needs. We then look at the issue of respondents fatigue and finally outline some avenues for enhancing efficiency in the production of Eurosystem statistics both in the longer-term as well as the short- to medium-term.

Based on their policy functions, their closeness to financial markets, their long-standing expertise and their credibility and – last but not least – the public good nature of such information – central banks are best placed to serve as the providers of such statistics about the financial economy. They are the so-called natural competence centers for financial statistics.

The relevant financial phenomena to be covered by statistics are by no means static – they are more like “moving targets”. New and more complex financial instruments are being developed at increasing frequency and speed. These instruments may either complement existing products or compete with or even partly or fully substitute them. In parallel, non-bank financial corporations (special purpose vehicles, insurance corporations, investment and pension funds) gain in importance. All these developments have the potential to change the transmission mechanism of monetary policy as well as the risk profiles for financial stability.

Therefore, the rapidly developing financial markets need to be covered by rapidly evolving statistics that correctly capture the policy relevant issues, including very complex areas, such as derivatives. As national borders lose relevance the cross-border supply of and demand for banking products rises dramatically, creating a growing need to capture these phenomena statistically both for monetary as well as financial stability analysis.

The recent financial turmoil has – among other things – highlighted an urgent need for more transparency in financial markets. Reliable and harmonized statistics are the key to transparency. Important gaps in the relevant statistical information available became apparent, gaps that need to be closed if we are to avoid a recurrence of the current problems. Especially, the correct measurement of the allocation of credit risk is incomplete or even completely lacking. The measurement of the extent and types of securitizations is also very incomplete or not comparable across countries and industries. Contingent credit exposures of banks, which turned out to be more important than previously thought, also deserve more statistical attention. The urgent need for harmonized data on housing markets became evident too.

The originate-and-distribute model of banking creates a need to gain further insight into credit transfers from banks to other financial institutions (financial intermediaries including hedge funds, insurance corporations, pension funds etc.) through credit derivatives and securitized loans. New markets in products like credit default swaps and collateralized debt obligations are spreading risk more widely and technology is enhancing trading opportunities in these instruments. The speed of financial innovation requires equally rapid reactions by statisticians in order to stay policy relevant. Therefore, the “time to market” of new statistics needs to be shortened in parallel with these financial innovations. At the same time, the continuation of statistics on financial products that lose importance needs to be re-evaluated.

One area where rapid action now appears warranted is the scope of the reporting population as specified in the ESCB’s basic statistical law (Council Regulation 2533/98 concerning the collection of statistical information by the European Central Bank). It needs to be extended to include insurance corporations and pension funds which are currently explicitly excluded, thus creating an important statistical gap. Information on these important financial institutions is required, both from a monetary stability and a financial stability perspective.

There is greater public demand for “evidence based policy making” and more interest in the underpinning evidence itself. Policy decisions will increasingly be based on and justified by relevant and reliable quantitative information, and policy results will be more closely tracked by data.

Increased uncertainty and unpredictability of the political, economic and financial environment requires more policy-adaptability in general and for central banks in particular. Policy-adaptability requires a broad (but also more frequently changing) range of (real-time) data and at the same time transparency and active policy communication. Information from financial markets thus becomes more important for central bank policy making.

Parallel to the identified needs for further statistical information we observe a trend towards increasing “reporting fatigue”. The concurrent developments of fierce European as well as world wide competition, the quest for shareholder value and the political drive in Europe towards a reduction of the administrative burden has led to greater attention for the costs of statistical reporting, even though statistics constitute only a very limited part of the overall administrative burden.²) In 2006 the ECOFIN Council came to the conclusion that “while high-quality statistics are crucial for policy making in Europe, enhanced efforts are needed to reduce the administrative burden caused by statistics in the EU.” This will increase pressures to reduce data demands, to maximize the use of existing data sources and to coordinate data collection and compilation both within and among NCBs and between NCBs and other statistical institutions. Policy makers will, however, have to take good care that this drive for simplification does not backfire and result in a major loss of crucial information.

Some central banks have already set up systems that keep the respondents’ burden to a minimum. In Austria, for instance, the new balance of payments compilation system was designed to make reporting as light as possible. Since the quality of the data has to be ensured nevertheless, the central bank statisticians have to shoulder a larger share of the compilation burden. In the production of statistics there is “no free lunch”.

One additional avenue that might help to reduce the “perceived, subjective burden” might be to invest in the extended return of useful information to the reporting agents (“customized feedback”) with the potential consequence that the overall ‘net burden’ (the cost of reporting minus the usefulness of the data received back) is perceived as more tolerable.

Statistics compilers might also be helped by advances in technology. The power of the tools is steadily increasing, allowing for the processing and storage of huge amounts of data at faster speed and lower cost. Moreover, as most modern economic activity generates flows of information through informational networks, new technical opportunities to capture relevant data from those networks could be developed. Collection systems have to be very technologically advanced and flexible in order to grasp the most relevant information, generally micro data from which information relevant at some more aggregated level may be derived. For instance, collection systems could be connected to the information systems of the declaring agents (e.g. the companies’ ledgers). This will require important efforts by the Eurosystem to keep up with those advances in technology and maintain and develop appropriate skills and tools in empirical economics, statistics and IT. It also calls for proactive cooperation with the reporting agents. Confidentiality obviously needs to be guaranteed.

All these efforts to address the challenge of reducing the respondents’ burden might have repercussions on the quality of the statistical output/products. Only statistics that are considered as relevant and reliable and are not challenged can support the credibility of the respective institution. For the Eurosystem this applies not just to the common European aggregates but also to their individual/national parts. Perceived or real quality problems in one part of the system can create collateral reputational damage for the whole system. Such risks need to be avoided in the interest of credibility. Therefore, efficiency cannot come at the price of major reductions in quality. Eurosystem statistics have to remain “fit for purpose”.

Wikipedia defines efficiency as “… the idea that a system proceeds with the minimum amount of waste. Efficiency is improved if the amount of “waste” or “friction” is reduced.” Moreover, “this takes place when production of one good is achieved at the lowest cost possible, given the production of the other good(s).”

But the concept of efficiency is related to that of effectiveness. We use here a broad concept of efficiency that includes the idea of “doing the right things” as well as “doing the things right”. To achieve this we have to define what are the right things, i.e. the statistics needed for the policies of the ESCB, namely monetary policy as well as safeguarding financial stability, but also for ensuring payments systems stability. We need to have information “fit for purpose”. This means measuring the relevant economic phenomena, providing data with the necessary detail, quality and timeliness. In addition, the data might – ideally – serve several policy purposes simultaneously. All these aspects have potentially large repercussions on the costs of producing  statistics.

Then there is the issue of using appropriate statistical methods and processes to achieve the necessary results. Do we need a census or can we work with sample surveys? What kind of estimation techniques can we use to achieve the needed results? Can we reuse existing information?

Organizational aspects are also important for the efficiency of the statistical production function. How is the work flow organized? What is the division of labor between respondents and compilers? What is the division of labor between different statistical compilers, nationally, within the Eurosystem, as well as internationally?

And finally, efficiency needs to be ensured inallthe different stages of the statistical production process, namely the input phase, the compilation phase as well as the output phase. Indeed, the drive for efficiency must go beyond the statistical production process and take in the entire process, starting from the advent of the information need and the formulation of the respective user need. Achieving maximum efficiency depends on evaluating and optimizing all stages of the process.

An additional challenge is to gain insight into interdependencies between different solutions. Some might be complementary, others – however – might result in incompatibilities, i.e. efficiency gains in one dimension might prohibit or offset efficiency gains in other dimensions. Achieving overall efficiency gains in the production of central bank statistics is no easy task. Nevertheless, it is a challenge we have to face and to master.

However, we are not starting from scratch. Efficiency in statistics production is not a new topic for Eurosystem statisticians. Much has been achieved already during the first decade of the Eurosystem. Otherwise, we would not have met all the challenging deadlines or tripled the statistical output with more or less unchanged human resources, as has been achieved with – what E. Domingo Solans once characterized as – a “silent revolution”.

Some of the most important elements that are already in place and play an important role in ensuring that the right things are done in the right way deserve mention here:

There is the formal so-called “merits-and-costs procedure” for all new or expanded statistical demands. Potential users requesting data are asked to define their policy needs and the measure of detail and timeliness they require, while the reporting institutions are asked to calculate the costs such new data reporting would entail (with different options). The statisticians then match the reported merits against the costs and try to develop the most efficient package.

Statistical representativeness of the aggregate data does not (always) require census reporting. Since the start of the Eurosystem, statistical methods have been used to reduce the reporting burden especially for small institutions. “Cutting off the tail” and “sampling” are methods already in use. Both have helped to reduce the number of reporters for the respective statistics in a major way. Also the reuse of mone- tary data for administrative purposes, namely for calculating the minimum reserve obligations, is a tested procedure to alleviate reporting obligations. This is much appreciated by the reporting agents as it prevents a double reporting necessity. Survey approaches are also already applied successfully by several Eurosystem countries, for example for external statistics.

But several member states have already achieved important efficiency gains through the allocation of tasks. Specifically, via formal and permanent cooperation with the respective national statistical institutes (NSIs). In addition, statistical cooperation with banking supervisors is already being success-fully practiced – albeit to different degrees – in several countries.

There are also some ongoing projects that may yield important efficiency gains, like the common Eurosystem effort to establish a Centralized Securities Database (CSDB). This CSDB, including the reference micro data on all individual securities issued, is expected to accommodate a multitude of statistical products and thus to become a major step towards more efficiency. On the output side, the Statistical Data Warehouse (SDW) that was established by the ECB and can be used by NCBs might play a central role for an efficient distribution of Eurosystem statistics to a very wide user community.

The quest for efficiency is therefore nothing new for statisticians but already well established, maybe more so than in other fields of central banking. Nevertheless, the developments we mentioned above – like financial innovation, respondents’ fatigue, the desire to reduce administrative burdens as well as increased public sector accountability – make it imperative that additional efficiency gains are identified and implemented where feasible.

The initial question to be tackled is which type of statistics we want to produce more efficiently. Do we want to optimize only the production of those data required for Eurosystem aggregates or do we plan to optimize the whole statistical production of our central banks, i.e. including all the national data requirements. The answer to this question will have important repercussions on the optimal solution(s) available. We need to avoid optimizing the compilation of euro area statistics at the expense of creating inefficiencies on a national level. Ideally, the two should go hand in hand, but we have to be aware of potential conflicts of interest. Coordinating on a national level with different statistics producers might hamper coordination on a Eurosystem level if arrangements differ from one country to the other. The same might apply if the link between monetary and supervisory data production is organized in a very country specific way. Cooperation in new models of production therefore has to remain voluntary, respecting national specificities and preferences. However, over time a convergence of the models might be possible and even desirable. In economics we define the long-run as the time horizon when everything is flexible. Legal and institutional arrangements can also be changed over time. Learning from best practices in other countries might also be an important driver towards efficiency gains.

For a long-term vision for the most efficient statistical production process in the Eurosystem we have to ask ourselves, how would a (comparable) multinational enterprise with 15 or more national subsidiaries approach such a task. The long-term vision must then be to work like one organic system that optimizes the multinational production process. Given the existing differences in national financial markets, languages and traditions as well as the expediency of taking advantage of the closeness to the reporters, an optimal system for the Eurosystem would not be centralized. Under these circumstances the most efficient solution seems to be a network, with harmonized technical standards, a clear division of tasks, no duplicity, common infrastructures, maybe even centers of excellence for specific tasks and decentralized contacts with reporting agents and with data users.

The Eurosystem’s statisticians therefore face the longer-term challenge of developing such a model, or several options for such a system, by building on the respective comparative advantages. Ideally, it should optimize the production of euro area as well as national data at the same time. The model would have to encompass all the statistical, organizational, legal, technical, as well as all governance issues that need to be resolved for such a cross-country production network. But in view of the large diversity in euro area statistics and across reporting institutions, financial markets and institutions to be covered, different solutions for different statistics may be conceivable. One model might not fit all statistics.³) Examples in the financial industry show that such solutions are both technically feasible and highly efficient. As a first step, it may be most appropriate to apply such a consolidation for new statistics that have yet to be developed.

These are all issues to be studied thoroughly during the next few years by the respective experts, mainly statisticians, IT-experts, organization ex-perts but also legal experts. This is a very challenging task but one worth pursuing with determination. In addition, it could serve as a model for other areas.

As this is a rather revolutionary longer term vision, its feasibility should first be evaluated within a more limited framework. The CSDB is a small model of such a network. A common databank for foreign direct investments would also be a logical candidate for evaluating the feasibility and the preconditions for such euro area wide production networks.

In the three stages of the statistical production process mentioned before – input, compilation and output – potential initiatives for reaping short- to medium-term efficiency gains are possible and should be evaluated with a view to implementation.

On theinputside, the goal should be that all data are collected only once. Any form of double reporting or redundant collection should be discontinued or avoided. Data already available – for whatever reason – should be reused if found useful for (other) statistical purposes, while strictly safeguarding their confidentiality and ensuring that the sharing is legally allowed or explicitly agreed by the reporting agents. Likely candidates are existing credit registers as well as balance sheet data from balance sheet offices or supervisory information. Experience in some central banks has shown that the use of such data for statistical purposes can lead to a very significant reduction in the response burden, higher data quality and lower costs to society.

On a national level, a formalized exchange of administrative data with institutions outside the central bank like the NSIs or the tax authorities would also help to reduce reporting costs. An important precondition and help would be the maintenance of common company registers with the NSI. Extending this idea across national borders one could think of common international data bases – allowing for the exchange of (confidential) micro data e.g. on significant cross-border mergers and acquisitions that need to be recorded symmetrically in the respective statistics of both affected countries.

The integration within each central bank of overlapping parts of the statistical production will avoid double reporting and enhance data consistency while improving overall efficiency. This integration could even take place at the micro data level, i.e. through a fully integrated data collection by NCBs and the definition of aggregation procedures tailored to different statistical purposes. Such an approach is conceptually appealing, but requires considerable coordination efforts within the central bank and could mean a major change for many Eurosystem NCBs. An intermediate solution is the sharing of aggregate data (building blocks) between different statistics, which could be facilitated through the application of common taxonomies. Finally, experience in the Eurosystem has shown that an efficient integration of all statistics, or at least those serving the Eurosystem tasks, is greatly facilitated if their production is concentrated in one functional area at the central bank, headed by a chief statistician.

While supervision is not an ECB task, several of the euro area NCBs that are nationally responsible for supervision have already integrated or linked their supervisory data collection with the statistical MFI or investment fund reporting framework. The initial focus of alignment will be on potentially overlapping requirements for the supervisory data collected by an NCB, but this may be extended, where appropriate, to collaboration agreements with (outside) financial supervisors. More far-reaching collaboration steps at European level may also hinge upon the speed and intensity of further harmonization of supervisory reporting across the EU. A first prerequisite may be a reference to the possibility of collecting data for financial stability purposes in the basic legislation governing ESCB Statistics (EU Regulation 2533/98).

Another substantial first step would be the joint development of basic concepts and definitions by statisticians and supervisors. This may also include an investigation into the use of common reporting formats and taxonomies (e.g. XBRL) for more integrated reporting.

As far as the statisticalcompilationphase is concerned, efficiency might result from the use of modern statistical techniques. Currently, most Eurosystem statistics are collected and compiled in a way that ensures country as well as euro area representativeness. Where only euro area data are required, the use of advanced statistical techniques might reduce the number of (credit) institutions that are obliged to report. This would require the selection of reporting agents in line with euro area based criteria. The system should allow each country to collect additional information within the same collection process if it also required representative national results. While the reporting burden would thus remain unchanged for the latter countries, it would be reduced in the countries that do not require such (detailed) fully representative national data and also for the euro area as a whole.

However, statistical techniques that reduce the reporting population may not be fully applied where the data collection also serves supervisory or administrative purposes that require a full census (e.g. for minimum reserves purposes).⁴) The enhanced use of common Eurosystem statistical techniques in, for instance, quality control, extrapolation, back data estimation, and seasonal and working day adjustment also offers possibilities for cost savings and more harmonized, better quality statistics. The wider application of statistical techniques requires, however, the availability of well trained statisticians at the central banks.

Substantial savings, although probably mainly in the medium- to long-term, would be generated if central banks shared own-developed statistical tools (i.e. software). A further step could be the common development of such software in specific central banks, which would require an enhanced governance structure and coordination, but would also save development time as well as server and maintenance costs.

Several statistics producers operate at the national level. Cooperation between these institutions could generate significant synergies.⁵) An allocation of tasks according to the respective comparative advantages – central banks for financial statistics, statistical institutes for non-financial statistics – would offer the twofold advantage of better statistics as well as avoidance of duplication. The Austrian example of six years of such close cooperation could be followed by other countries. Specifically, the joint development and maintenance of a single business register, including fully harmonized identification and classification of the reporting units by NSIs, supervisory authorities and statistical departments of central banks could lead to efficiency gains, a substantial reduction in the burden for reporting agents and quality improvements in statistics. In the longer term, these registers could then even be linked into a single European business register.

In order to avoid duplication while providing consistent data, all agencies involved in the production of statistics should, where possible, apply the same concepts, definitions, classifications, reporting forms and reporting formats (e.g. XBRL) at national level. Again, exploiting comparative advantages may in most cases entail that the NCBs collect data from the financial corporations, while the NSIs mainly deal with the other sectors. The application of harmonized approaches at national level will also facilitate a further integration at Eurosystem level.

Another aspect of efficiency is the time it takes to deliver useful information on newly emerging phenomena. Rapidly changing financial markets require fast adaptation of statistical concepts and collection systems. A short ‘time-to-market’ is almost a precondition for providing ‘fit for use’ statistics in such a rapidly changing environment. This requires the early identification and broad assessment of such new phenomena that have a (potentially) relevant effect on policy. However, before the statisticians are asked to set up new systems we need a quick understanding and initial analysis of the policy relevance of the new phenomena by the respective analysts in cooperation with the statistical experts. New approaches for such cases should be developed that might not require a full scale collection system and a new legal instrument.

The dissemination of statistics – theoutput– is an integral part of the statistical production process. Reaching more users with the same amount of data is one aspect of increasing the efficiency of the statistical production process, thus reducing the ‘cost per user’. Useful data feedback is also an important aspect for securing reliable reporting by the respective reporting agents. At the same time, the statistical data available in the Eurosystem provides the system with a wide array of possibilities for providing services to the general public as well as to different subgroups and so to link up with the public as both reporters and “consumers” of data. In addition, packaging Eurosystem statistics to serve the needs of specific target groups (including the reporting agents themselves) may enhance public support for the Eurosystem as a whole and may also reduce the ‘perceived’ reporting burden.⁶) The system, however, needs to take a strategic decision to proactively use statistics for its communication and to accept a role as a “statistical competence centre for the euro area”. By reason of its global economic importance, the Eurosystem has also a responsibility to provide access to its statistics and the respective metadata for economic and financial research.

Many of the potential efficiency gains identified above are only fully achievable if the statistics producers in the euro area cooperate to the maximum extent. The current legal framework – especially concerning the exchange of confidential data between different statistics producers – is an obstacle to such an efficient solution. Exchanges of confidential data between statisticians within the ESCB/Eurosystem as well as with those of the European Statistical System (ESS)⁷) for statistical purposes could offer significant efficiency gains via a reduction in the respondents’ reporting burden and an improvement in quality. It would be a true win-win situation. The respective basic statistical legal acts of the ESS as well as the ESCB should include such provisions. The ongoing revisions of these legal acts form a perfect opportunity for including such provisions and so considerably reducing the administrative burden.

A further precondition for some of the identified synergies and efficiency gains is dialogue, and coordination of information needs, between those responsible for price stability as well as macro financial stability and those in charge of supervision.

Another important way of achieving efficiency gains is a frank assessment of the further need fornationalfinancial data and a commitment to action in this area. This has important repercussions for crafting statistical production processes in the most efficient way, both for monetary stability as well as for macro financial stability purposes. In light of the common single monetary policy, national data of publishable quality might appear to be no longer necessary. However, financial integration is still incomplete and therefore the transmission mechanism of the common monetary policy might be nationally quite diverse, thus prolonging the need for national monetary information. In addition, national economic policy making requires national data, like national balances of payments. Also for financial stability purposes national data – like financial accounts or retail interest rates – remain crucial. However, different levels of detail might be required at euro area level than at national level. The size of the respective country might also be relevant where only euro area aggregates are required.

Finally, efficiency gains by means of new forms of cooperation hinge on an openness for embarking on new forms of collaboration. Collaboration models replace autonomy with mutual dependencies. Potential risks related to these dependencies need to be mitigated by crafting adequate governance rules, though some remaining risks need to be accepted.

The Eurosystem/ESCB is and will be an undisputed global player in the area of monetary, exchange rate and financial stability policy. Being a global player brings international responsibilities, including a proper dissemination and analysis of financial data as well as the education of data users. The Eurosystem will have to provide statistics on the euro area as an international public good. At the same time, the Eurosystem has a comparative advantage in providing financial data for the euro area, and in this context, can help to foster and strengthen a euro area/ European identity (“make Europe tangible”). It also needs to make financial information more widely accessible, and more relevant to a wider audience. This will include the need to reassess the data/information needs from a global perspective.

In central banking decisions have to be made in real time with a measure of uncertainty. A sound statistical system helps to reduce the level of uncertainty and so lower the likelihood of suboptimal decisions. It is thus an investment for the stability of the monetary and financial system. As this crucial task of “lifting the fog” comes at a cost, both for central banks as producers of statistics as well as for society as suppliers of data, it is a continuous challenge to ensure that the merits outweigh the respective costs. In evaluating this balance, we should consider the costs of unfounded decisions or decisions resting on an incomplete or even faulty basis. One important element for a positive balance is to set up compilation systems that are efficient. Efficiency, however, is not a static concept but calls for continuous monitoring and adaptation of processes. Especially with the advent of the Eurosystem not only the statistical demands of the users have changed radically but also the challenges of producing euro area statistics.

The information needs of the policy makers are changing rapidly as the underlying developments in the economy and especially in the financial markets evolve with increasing speed. In light of the rather limited resources available in central banks for their statistical sections and the growing demands for limiting the reporting burden of respondents, the statistical production processes need to be as efficient as possible. Efficiency gains are mainly needed to accommodate the new data demands – both at the national as well as on the European policy level. This applies to all stages of those processes, starting from the identification of a data need by policy makers and analysts up to the dissemination of final results. Optimizing these processes cannot be done by the statisticians alone but requires the active cooperation of users, reporting agents, legal experts as well as those in charge of the information technologies used.

There is a need for a strong and continuous dialogue and cooperation between the statisticians, the “masters of the data”, and monetary and financial stability analysts. The data needs of the monetary and the financial stability analysts might be rather similar. Early coordination might thus avoid double reporting. Different demands might be justified due to different analytical needs, but that needs to be clarified and verified before the central bank statisticians approach the respondents.

Efficiency gains are achievable but they are by no means a “philosopher’s stone”. This “lapis philosophorum” is a legendary substance, supposedly capable of turning inexpensive metals into gold. Statisticians are not alchemists. But let us show that by cooperation and collaboration within the Eurosystem as well as with other statistical producers we can achieve important efficiency gains and reduce the respondents’ burden. This should enable us to free resources for the new challenges coming from financial innovations. Let us also stop viewing central bank statistics as a burden or pure cost factor. As Bill Poole, President of the St. Louis FED observed recently “of all things on which we spend money in the Federal Reserve, surely the return on our data services is among the highest.” Thus, let us consider them as what they really are, namely a valuable central bank asset, almost like gold.

Irving Fisher Committee on Central Bank Statistics. 2008.

Cooperation between Central Banks and National Statistical Institutes: The Cases of Australia, Canada, and the Netherlands. IFC Working Papers 1. Basel.

Mayerlen, F., P. Sanders and C. Willeke. 2005.

Monetary Analysis in the Euro Area: Statistical Issues at Stake. Frankfurt am Main: ECB (unpublished).