The macro-prudential supervision of the financial system as a whole is being significantly enhanced as a result of the lessons drawn from the global financial crisis. We are now in the process of building up a coherent macro-prudential policy framework at the domestic, European and global levels. For its part, the European Central Bank (ECB) is, in particular, entrusted with the task of providing the Secretariat and analytical, statistical and logistical support for the European Systemic Risk Board (ESRB), which will start operating in January 2011.
The main new element for the conduct of macro-prudential supervision is the determination to link risk surveillance and risk assessment to concrete policy actions. In this context, the allocation of responsibilities to macro-prudential bodies, such as the ESRB, has emphasised the need for enhancements to the information base, not only for the purpose of risk analysis but also to prepare macro-prudential policy recommendations to address any systemic risks that are identified. The latter will require, for example, the analysis of the transmission channels of macro-prudential policy measures and their potential impact on the financial system. This is largely a new field of policy analysis for most central banks and may require specific additional data.
As has been highlighted by the speakers at this conference, the statistical function provides a key underpinning for the effective implementation of macro-prudential policies. In particular, a comprehensive and granular information base is required to facilitate the timely detection of the build-up of vulnerabilities, such as financial imbalances. Furthermore, the accuracy and reliability of data largely determines the quality of the systemic risk assessments that inform macro-policy decisions.
I am using the word “information” here in a broad sense to mean not only statistical data but also market-based information and information gathered via market intelligence efforts. Systemic risk analysis and the outlook for financial stability necessarily incorporate forward-looking elements included in current market prices of assets and market expectations as well as information on incipient trends, business practices and new financial instruments that emerges from market intelligence activities.
As illustrated by the topics discussed at this conference, the new statistics required to support macro-prudential analysis and oversight constitute a challenge shared by all authorities around the globe responsible for safeguarding financial stability. Given the global nature of the financial crisis, major initiatives have been launched at an international level, namely by the G20 supported by the IMF and the Financial Stability Board. These initiatives address the global agenda for improving financial statistics, in which the EU authorities and the ECB are playing an active part.
In addition, significant efforts are being invested in analysis and research with a view to enhancing and developing risk detection and risk assessment tools. They include indicators of emerging imbalances and early warning indicators, as well as tools for evaluating the severity of identified risks, including macro-stress-testing techniques, and models of contagion and spillover effects, both within the financial sector and between sectors of the economy. These analytical tools need suitable data of sufficient granularity in order to produce reliable and accurate results.
In my remarks today, however, I will focus mainly on statistical needs and the contribution that central banks can make in this regard.
Of course, macro-prudential analysis and oversight are not new tasks for central banks – or at least those with a responsibility for safeguarding financial stability. Central banks are in a favourable position to perform these tasks in view of, first, their analytical and statistical competencies, including their expertise in processing and managing data, and, second, the infrastructure that is already in place to facilitate the conduct of monetary policy and financial stability functions. Indeed, following the various regulatory reforms prompted by the lessons of the financial crisis, central banks (such as the Federal Reserve System and the Bank of England) have generally seen their role in macro-prudential supervision and policies reinforced. In addition to their key involvement in the financial sector and their analytical competencies, this enhanced role for central banks also reflects the essential features that enable them to ensure credibility in pursuing macro-prudential policies. These features comprise: (i) the fact that central banks are, by their very nature, anchors of stability; (ii) they have a medium term-oriented policy horizon; and (iii) they are independent.
The ECB’s extensive statistical know-how has been developed over the past 10 to 15 years predominantly to fulfil the needs of monetary policy analysis. This institutional expertise relates, in particular, to: (i) the development of harmonised requirements for quantitative statistical information derived from heterogeneous basic national sources; (ii) the codification of these requirements in legal acts and their implementation in reporting formats; and (iii) the production of large data sets in a timely fashion.
Accordingly, the expertise and infrastructure developed by the ECB and the European System of Central Banks (ESCB) can make a significant contribution to supplying the statistical information required for the macro-prudential functions of the ESRB, and can do so in the shortest possible time frame, while containing costs and limiting the reporting burden for respondents.
I will begin by briefly focusing on the existing central bank statistics that can be used for macro-prudential analysis – what we can term the “existing supply”. I will then move on to the “new demand” for information in order to be able to effectively fulfil the new central banking responsibilities in the field of macro-prudential oversight. Finally, I will try to assess how large the gap is between the existing supply and the new demand for information. However, I would acknowledge at the outset that it is impossible to ever close this gap completely – particularly since financial innovation and structural developments will always create new demand for information. Consequently, flexibility – or, if you prefer, agility – should become a standard attribute of statistical processes in the fields of both macro and micro-data so that risk assessments can be produced in real time in order to capture innovation in financial markets.
Let me start with the existing supply of statistics. The ECB and the ESCB already have a good macro-financial statistical database. It is designed primarily to serve the monetary policy function but, to some extent, it also serves the needs of macro-prudential analysis and will be of benefit to the ESRB.
To be more specific, macro-prudential analysis covers both an analysis of the financial system – comprising financial intermediaries, markets and infrastructures – as well as an analysis of the financial system’s operating environment – namely the non-financial corporations, households and government sectors, as well as the global macroeconomic setting.
A wide range of existing macro-financial statistics on the operating environment that are compiled by central banks can be used for financial stability analysis. I am thinking in particular of the integrated euro area financial accounts, in which cross-sector balance sheet exposures allow the degree of interconnectedness among sectors in the economy to be gauged, and on the basis of which a framework for assessing the scope for contagion among economic sectors has already been developed. This framework combines the euro area accounts data with a market-based analysis of risks in a contingent claim flow-of-funds accounts model. 
Other examples of existing statistics relate to information of a more qualitative nature. In 2009 the ECB and the European Commission launched a survey on the access to finance of small and medium-sized enterprises in the euro area and the EU,  which will be helpful in assessing the vulnerabilities stemming from conditions in the non-financial corporate sector. To mention another initiative, the Eurosystem Household Finance and Consumption Survey will provide micro-data to complement the information available from the euro area accounts. It will enrich the analysis of risks to financial stability stemming from conditions in the household sector by providing information on distributions of household income, consumption and wealth.
Turning now to the financial system, a similarly wide range of existing macro-financial statistics, compiled by central banks, can be used for macro-prudential analysis. They include, for example, statistics on financial markets, money, banking and investment funds, as well as information on insurance companies and pension funds. However, statistics on financial sectors collected for financial stability purposes tend to require different formats and features than those collected for monetary policy purposes. For example, banking statistics for financial stability analysis should be compiled on a consolidated basis to reflect risks at the group level instead of on an individual bank basis as is the case for monetary policy analysis. In addition, the assessment of risks to financial stability stemming from the activities of investment funds (e.g. hedge funds) requires a coverage on the basis of where these funds operate, rather than where they are registered or domiciled. Over the years, with the help of national central banks and supervisory authorities in the EU, acting via the ESCB’s Banking Supervision Committee, consolidated banking statistics covering practically the whole of the EU banking sector have been developed and comprise the prime data source for conducting financial stability analysis.
While the financial crisis has clearly illustrated the need for micro-data, it should be recognised that during the crisis signals of imbalances did emerge from aggregated macro-financial data, although they failed to attract sufficient attention. I am referring, for example, to broad-based increases in financial leverage, both in the financial and non-financial sectors of the economy, and the steady growth of banking sector balance sheets, as well as growth in off-balance sheet items.
These sets of macro-financial statistics, combined with market-based information, e.g. prices of securities and related derivatives, market expectations derived from these prices, non-bank-based financing volumes and related maturities, and exposures to specific asset classes, as well as regular market intelligence, have proved to be essential elements in monitoring developments across sectors and markets, detecting price anomalies, and assessing the markets’ perception of the financial system’s ability to withstand shocks.
Overall, a significant amount of information is already being collected by central banks and official statistical bodies and provides a considerable share of the data needed for macro-prudential analysis.
Nevertheless, new data will be needed to support the conduct of reliable, effective and high quality macro-prudential analysis by the ESRB.
A useful way to organise our thinking about these data needs is by making a distinction between top-down and bottom-up analysis. A top-down analysis is mainly based on aggregate information, namely macro-financial data, and is performed in order to monitor vulnerabilities and assess conditions in specific sectors with possible implications for system-wide risks. It can be used to broadly identify areas of potential risk, indicating where further, more granular, drill-down analysis is needed. Bottom-up analysis, on the other hand, relies on firm-specific or micro-level data, either publicly available in individual financial statements or reported to supervisory authorities, that is then aggregated to produce sectoral or system-wide assessments.
For top-down macro-prudential analysis, efforts need to be made towards: (i) harmonising; (ii) increasing the frequency; and (iii) extending the coverage to the whole EU of the statistics for the banking sector and other financial sectors, such as insurance. Harmonisation efforts largely depend on the ongoing efforts to establish unified financial supervisory reporting in the EU. For the banking sector, where progress is more advanced, a considerable amount of work needs to be done in respect of consolidated banking data. Improvements are under way, but the new templates for supervisory reporting in the EU (which are based on new reporting standards) are still under discussion. The collection of similar statistics for the EU insurance sector faces the same challenges and preparations are currently ongoing at full speed.
Another critical area for improvement in the top-down analysis is the effective coverage of the non-regulated financial sectors, or the so-called shadow banking sector. Although there is no formal definition, this sector comprises institutions, vehicles, instruments and markets whose businesses largely replicate core elements of traditional banking: i.e. credit and maturity transformation. Major components of this shadow banking sector include certain money market funds, hedge funds, structured investment vehicles, off-balance sheet vehicles (reliant on banks’ credit lines), and securities lenders. They tend to be overly dependent on the liquidity of some markets and, just like banks, are vulnerable to runs. Importantly, a complicating factor in the collection of such data is that the shadow banking sector is not confined to a specific institutional group or type of business, but is spread across entities. It relates more to the way in which entities operate in certain financial markets and their use of financial instruments. Critical markets are, for example, the asset-backed securities market, the repo markets and the securities lending markets, where key players can be unregulated or non-banking regulated entities that are not captured on the macro-prudential supervision “radar screen”. We need to be agile in our statistical efforts if we are to capture the bulk of the financial intermediation activity being channelled outside regulated sectors, even though such activity appears to be subdued in the aftermath of the crisis.
Without attempting to be exhaustive, let me also mention the much needed improvements to the integrated euro area financial accounts that I referred to before. There is a need to expand the coverage to the EU level, to have more granularity in balance sheet exposures and to produce additional breakdowns by types of financial instrument. These would be essential inputs for evaluating propagation effects using contagion and spillover models, for example. This type of analysis can be used to evaluate the impact of the failure of specific components of the financial system, by assessing the transmission of instability among financial intermediaries and markets.
Overall, satisfying the above-mentioned data requirements will be crucial for developing analytical tools and methodologies for systemic risk analysis, including financial stability indicators, early warning indicators and stress-testing models, in addition to contagion analysis. The quality of the analysis will depend not only on the sophistication of the methodologies, but, to a large extent, on the quality of the data. The ability to detect financial imbalances at an early stage, to distinguish them from potential structural developments, and to deliver risk assessments in real time will depend on the quality and flexibility of the information base.
Moving now to the field of micro-data, it has been widely acknowledged that institution-specific information is needed for what I called bottom-up analysis. Viewing the financial system as a network, the focus of this analysis shifts to the key nodes in the network and, notably, the concentrations and linkages between them that could represent risks and vulnerabilities affecting the stability of the entire network. There is a need for information on interlinkages between the major financial system players, including counterparty credit exposures in various forms, and funding exposures of individual financial firms, as well as detailed information on their maturity mismatches and leverage. This is because vulnerabilities can stem from interlinkages such as those emerging from interbank lending, securities lending, repurchase agreements, funding interdependencies, positions in credit default swap markets and exposures in other derivatives markets, and ownership links. They can also relate to common exposures (to particular economic sectors or regions) or to holdings of assets subject to contagion via asset markets (e.g. asset fire sales). This detailed information should allow linking the exposures of the regulated financial sector to non-regulated entities, such as structured investment vehicles, conduits and hedge funds, via the monitoring of exposures to specific financial markets and specific asset classes (e.g. structured credit products). In this context, the development of a Global Risk Map – as a unified database picturing the network of mutual exposures that exist among (i) large and complex financial institutions (LCFIs); and (ii) between the LCFIs and major counterparties, e.g. insurance firms, hedge funds, major corporations and central banks – should be supported, as proposed last year by the committee chaired by Otmar Issing,  and for which Stephen Cecchetti proposed a practical way forward yesterday. 
This micro-information is essential in order to carry out analyses of propagation channels of systemic risk – via direct contagion or spillover effects – as well as to enhance the quality of macro stress-testing exercises, in the period ahead. In this respect, efforts are already under way at the ECB to enhance the coverage of firm-level information and distribution indicators for large financial institutions, including information on interconnections and common exposures, building primarily on publicly or commercially available sources.
Indeed, macro-prudential analysis assesses the collective behaviour of financial institutions and the way in which it may pose risks to the overall system. It investigates the scope for negative externalities caused by key players in the financial system, since each financial intermediary rationally manages its own risk, without considering the potential impact on others and the system as a whole. This contrasts with micro-prudential analysis which looks at institutions in isolation and produces assessments at the individual firm level.
Recent initiatives in central bank statistics attempt to address some of the issues raised by the interlinkages among financial intermediaries – for example by means of what we call “from-whom-to-whom” information on deposits and loans of financial corporations. Work is also under way on securities holdings statistics with a view to creating who-to-whom data for sectors of the euro area economy. Furthermore, the use of dedicated ad hoc surveys is also envisaged in the near term, as a way to gather information on interlinkages and thereby enhance the granularity and quality of risk analysis.
In order to obtain the relevant micro-data, not only at the sectoral level but also at the individual firm level, that is required to perform bottom-up analysis, there is a clear need for close cooperation between the ESRB and micro-prudential supervisors, and, in particular, the European Supervisory Authorities (ESAs), which are likely to collect information through supervisory reporting and may collect systematically relevant firm-specific information upon reasoned request.
To ensure that cooperation between the ESRB and the ESAs is successful, a number of practical arrangements are needed, particularly in order to develop a clear understanding of the division of tasks between the macro and micro-prudential authorities. A clear understanding is also needed as regards the quality, comparability, timeliness, punctuality, frequency and transmission formats of the data, as well as confidentiality and the legal provisions covering them. The ECB and the ESCB, in their supporting role vis-à-vis the ESRB, is already collaborating with the ESAs to ensure that data collection is both effective and efficient, preventing duplication of work.
The new data requirements stemming from the need to conduct macro-prudential analyses are very challenging. Some commentators have underlined the role of the global financial crisis in identifying the new data required for macro-prudential analysis and the measurement of systemic risk.
However, we should not overlook the fact that considerable efforts have already been made with a view to improving the quality and coverage of the information. Moreover, there is a reasonable information base for the conduct of macro-prudential oversight at the EU level, which should allow the ESRB to start operating smoothly in January. At the same time, we should not be complacent: we need to acknowledge that there are considerable gaps between the existing supply and the new demand for data. Narrowing these gaps will take time and there is a need for strict prioritisation of the development work. Efforts will be further complicated by the need to ensure flexibility in statistical processes so as to more effectively capture financial innovation and other forms of structural development in the future.
The ESRB, together with the ECB and the ESCB in their supporting role, are not alone in facing challenges as regards the availability of statistics for macro-prudential analysis. The global scale of the financial crisis has clearly illustrated that close cooperation is needed at the global level. In this respect, the ECB is committed to making a significant contribution to the major global initiatives under way to narrow information gaps, led in particular by the Financial Stability Board and the IMF.
The ECB and the ESCB have recorded remarkable achievements in developing and collecting statistics which have made the common monetary policy for the euro area possible and a success. I am therefore confident that the existing knowledge and expertise, as well as the technical infrastructures of the ESCB statistics community, will ensure that the new challenging tasks that lie ahead as regards the new data for macro-prudential analysis will be met. In this context, I would reiterate that progress can also only be ensured if effective cooperation between the ESCB, the ESRB, the ESAs and national authorities is established.
 Castrén, O. and Kavonius, I.K., “Balance sheet interlinkages and macro-financial risk analysis in the euro area”, Working Paper Series, No 1124, ECB, Frankfurt am Main, December 2009.
ECB and European Commission, “Survey on the access to finance of small and medium-sized enterprises in the euro area”. The survey is carried out every semester (results available for 2009).
Issing et al., “New Financial Order: Recommendations by the Issing Committee Part II”, White Paper No II, Center for Financial Studies, Frankfurt am Main, March 2009.
Cecchetti, S.G., Fender, I. and McGuire, P., “Toward a global risk map”, paper presented to the fifth ECB Conference on Statistics entitled “Central bank statistics: what did the financial crisis change?”, October 2010.
Reproduction is permitted provided that the source is acknowledged.