Frankfurt 22.4.2004 


I think we all agree that good macro-economic statistics are essential for the smooth functioning of the Economic and Monetary Union, working under a single monetary policy and decentralised but co-ordinated fiscal policies. At the same time it is clear that the institutional settings of the EMU impose immense challenges for the collection, compilation and dissemination of statistical data.

As for the conduct of monetary policy, a comprehensive set of timely, reliable, sufficiently-detailed and frequent aggregate-level statistics are crucial. Timely, reliable and frequent statistics are invaluable especially for a central bank like the ECB, operating in an environment where economic structures and the monetary policy transmission mechanism are not fully known and likely to undergo continuous change. As for the conduct of economic policies, national data may lose some of its significance in a monetary union, but nevertheless it tends to remain important, not only for national economic policies but also for their co-ordination.

Since the launch of the EMU, Europe has improved tremendously in terms of the coverage and quality of aggregate time series at the euro area and EU level. This is true in particular for the balance of payments statistics, which are now compiled and disseminated monthly for both the euro area and the whole EU. These improvements have been largely a product of the close cooperation between the European statistical system, i.e. Eurostat together with the national statistical institutes, and the European System of Central Banks, these two systems sharing the responsibility for the balance of payments statistics in Europe.

In spite of the substantial progress achieved since the launch of the EMU, there is still room for improvement. Aggregate level statistics have not yet achieved standards comparable to those of the other large currency areas, notably the United States. From the monetary policy perspective, action is needed in particular to improve the timeliness and quality of the euro area statistics. In the field of balance of payments statistics, the most urgent needs probably pertain to improvements in the quality of first monthly preliminary figures.

The institutional structure of the EMU naturally tends to make not only statistical requirements but also the data collection and processing procedures unwieldy. Compilation of the aggregates requires harmonised data with sufficient coverage, and therefore collection and compilation procedures necessarily are more complicated than in the case of a single country.

I do not know of any systematic comparative study on the efficiency of the production of statistics in major currency areas. However, even a casual comparison of Member States' national statistical systems gives some support to the notion that their efficiency varies quite considerably across Member States. This suggests that our statistical system, taken as a whole, cannot be fully efficient. To avoid misunderstandings, it is perhaps worth repeating that a comparison of aggregate level outputs and inputs between the euro area and other currency areas may be grossly unfair to the euro area, as data requirements and the conditions of the statistical system elsewhere may be very different from those in Europe.

There is a strong public good element to aggregate statistics, and their production has always been and probably will always be financed mainly from public sources. In view of the formidable budgetary challenges that most if not all Member States are facing today and even more in coming years, it would be wishful thinking from the part of European statistical authorities, irrespective of their institutional background, to expect significant increases in the total amount of budgetary resources for the production of statistics in general or balance of payments statistics in particular.

If we want to improve the timeliness and quality of balance of payments series, there seem to be no other options than to rebalance priorities within the statistical system. In my contribution, I would like to stress two aspects of this rebalancing of statistical priorities.

In the first place, it is necessary to focus demand for aggregate data only on what is really needed for policy use, also taking into account the cost of production of aggregate time series. The potential need for statistics is unlimited and cannot be satisfied with limited resources. Setting priorities for the demand for statistics used for monetary policy purposes is mainly the responsibility of the ECB but as far as cost assessment is concerned, also national authorities responsible for the collection and compilation of data are involved.

In the second place, it is absolutely vital to ensure, preferably in a co-ordinated manner, that statistics are collected and compiled as efficiently as possible, given the basic institutional constraints of the EMU. Setting priorities for the use of resources in the supply of statistics is mainly the responsibility of the national authorities, which are responsible for the collection and compilation of national data. However, the achievement of an efficiency frontier in the production of statistics may be impossible without substantial and constructive input from Eurostat and the ECB, which define the general conceptual framework as well as co-ordinate the collection and compilation of balance of payments data.

In this contribution, my basic theme is that statistics, by their very nature, tend to be more or less incomplete and that the integration of markets due to globalisation and the single European markets has made them even more incomplete. My Danish college, Jens Thomsen, once put the very point very nicely - in the context of identifying balance of payment crises ex ante - by saying that when mistakes are analysed in retrospect, the relevant statistics – which could have prevented a wrong decision being made – always seem to be the ones which were missing at the time the decision was contemplated.

Because statistics are anyway incomplete, there is no need for actuarial precision in the design of the statistical system. I shall argue that we need a more focussed approach. In particular, in the case of the balance of payments statistics the adoption of more up-to-date collection systems with greater reliance on corporate survey is needed.

Challenges of globalisation for Balance of Payments statistics

While it is often claimed that balance of payments statistics are now less important than they used to be – say even twenty or thirty years ago – I fully agree with José Viñals' conclusion that, in spite of all the conceptual and operational challenges globalisation has put before the balance of payments statistics, balance of payments statistics continue to be very important, especially at the euro area level, for the monetary policy purposes. I would like to supplement his presentation, however, by arguing that statistical problems related to the globalisation process extend even further and deeper than he envisaged. I would like to start by drawing your attention to one well-known consequence globalisation has brought with it, namely the rise of very large global corporate units. A less well known but nevertheless evident fact is that these global companies are perfectly capable of growing and operating – or at least maintaining their head office – in a very small country. This development reflects the declining significance of national frontiers. It has taken place in Finland and probably also in some other smaller Member States, with the consequence that national statistical authorities are faced with the problem of recording the transactions of very large companies. These transactions may be extremely large, complex and numerous if compared with the transactions of “normal” enterprises operating entirely or mainly at national level.

Let me illustrate some of the challenges large companies pose for statistics by using the Finnish ITC company Nokia as a concrete example.

Nokia grew very rapidly in the 1990’s. Measured by its stock value, at the height of the IT boom in 2000, it was among the 10 largest enterprises in the world. Its market value was then nearly double the value of the Finnish GDP, reflecting of course high expectations of future growth of sales and profit opportunities.

Nokia's profits amount to several per cent of the Finnish GDP. After dividends and taxes, its undistributed profits are some 2% of our GDP. In national accounts, Nokia's undistributed profit is included in Finland’s national income, savings and current account surplus. This explains, in part, Finland's exceptionally high trade account surplus. However, some 90% of the company is owned by foreign investors. If undistributed profit to portfolio investment were allocated in national accounts and balance of payments in the same way as profit to direct investment, Finland’s national income, savings and current account surplus would have been annually some 2% lower than in published statistics. From an economic point of view, the current statistical figures can be regarded as misleading.

In international investment position (IIP) statistics, shares are valued at market prices. In 2000 Nokia’s high market valuation was reflected in Finland's IIP statistics with the consequence that, measured by its net international investment position, Finland was at the time one of the most indebted countries in the world. Of course, in spite of these statistics, neither the government nor any other body had any rational reason to worry about our international net position, as Nokia was mostly owned by foreigners, and almost all of its sales and presumably also company’s profits were made outside Finland. Finland’s high indebtedness in 2000 was nothing more than a statistical artefact. This observation was confirmed by subsequent developments. Finland’s net position diminished to about one tenth of its peak value as Nokia’s share price declined to some one forth from its peak value or even lower. Nevertheless, Finnish IIP figures have already confused those economists who do not know statistical recording practices and the structure of the Finnish economy– and are likely to continue doing so in the future, too.

The problems large global companies cause to national accounts and balance of payment statistics are fundamental. While in institutional sector accounts – households, public sector, non-financial corporations and financial institutions – it is relatively easy to classify activities by nation for the household and public sectors, this has become increasingly difficult and in some cases outright impossible for the nonfinancial corporations and financial institutions sectors. For the corporate sector, foreign trade statistics can already now be characterised as being merely indicative. Statistics have been contaminated not only by the cross-border mergers and acquisitions but also – and more importantly – by the so-called invisible transactions; foreign trade of services and capital income and the financial flows of large multinationals whose share of total value of transactions is often substantial.

As I already mentioned, the size distribution of relevant populations in economic and financial statistics, including the balance of payments, have become extremely skewed, as a consequence of the globalisation process. At the same time, many of the transactions made by these extremely large enterprises can be extremely complex.

The existing balance of payments statistics already witness the quality implications of these phenomena; the most apparent case being the first monthly preliminary figures for the current year retained profits that are highly unreliable. In fact, proper data is available only a year or more after the first estimates. Preliminary figures are often subject to delayed major revisions. At the disaggregated level, these problems may hamper their use for monetary policy purposes.

While the problems of extremely large corporations – in relation to the size of the country – and complex foreign trade transactions may be exceptionally acute in Finland, there cannot be any doubt that they are highly prevalent also in other Member States, at least in the smaller ones. Furthermore, these problems are likely to increase and become more intensive with the advancement of global integration. As many of the problems in the balance of payments statistics are universal, solutions to them should preferably be sought, in the first place, at international level. Some of the old typologies and classifications may have become outdated. It is a challenge for the international community to drop outdated dimensions from international standards.

The IMF has already started preparations on the 6th Balance of Payments Manual. It is expected to be ready around 2008. Topics already under discussion include the increased consistency and integrity of national accounts and balance of payments statistics, the increased role of foreign asset positions, and the relation of statistical standards to international accounting standards. Attention has also been paid to balance of payments problems in monetary unions. In this context, I would particularly like to mention the need to harmonise further national concepts in national accounts and balance of payments statistics. This need is demonstrated by the large “asymmetries" seen in the balance of payments statistics for the major currency areas and the whole world. Another example I would like to mention are institutional sector figures in national accounts, where different treatment of depreciation allowances by different authorities make economically interesting comparisons very difficult.

While revisions in international standards are important, the process whereby they are established is likely to be very slow. It may be too much to expect substantial progress already in the next balance of payments manual. In any case, some restructuring of statistics must be carried out in advance of the revision of the international manuals. Without prompt response at the European or Member State level, problems may emerge, as statistical concepts no longer are in line with relevant economic or business concepts. Unfortunately, these cases have not been rare in recent years in practice.

Emerging single European markets: challenges for Balance of Payments Statistics

While market integration has in the global context been largely a market driven process, in the European context it is an explicit policy goal of the highest priority. In many markets, integration at the European level has already proceeded significantly towards truly single markets. However, many European mar-kets, including those for financial and other services, are still highly segmented along national borders. Considerable policy effort, both in the context of the Lisbon strategy and elsewhere, has been devoted to removing the remaining legal and other factors which act as frictional elements in the creation of integrated European-wide markets. The establishment of fully integrated European-wide markets creates challenges for statisticians. Moreover, in some cases statistical issues may also be included in those frictional elements, giving rise to a consideration of different priorities.

I would like to mention a couple of examples of difficulties related to balance of payments that affect re-cent policy issues in the EU more directly.

The Regulation (2560/2001) on cross-border EU payments effectively made the cost of cross-border payments within the EU equivalent to the cost of domestic payments. At the same time, it relieved banks from all reporting requirements on cross-border euro payments below a threshold of EUR 12,500. In the framework of the regulation, the Commission will report no later than July 2004 on the advisability of increasing this threshold to EUR 50,000 as from January 2006. This rise in the threshold has been resisted on the grounds that it could pose a risk to the quality and detail in balance of payments. This clearly offers an illustration of a conflict between technical statistical requirements and the achievement of the broader economic and financial policy goals of the Community.

Rather than resisting the rise of the threshold, Member States should be asked to accelerate their plans to modernise their data collection systems for balance of payments statistics. This is all the more relevant because the bias in errors and omissions in the euro area balance of payments continues to raise concerns. Much of the “asymmetries” in European aggregate balance of payments statistics are due to the deficiencies of national data collection and compilation systems in Member States.

In many countries work has already started on the improvement of existing data collection systems. In particular, Member States are introducing corporate surveys and building more accurate registers. Furthermore, they have committed themselves to compiling more reliable portfolio investment statistics, to building more accurate registers, to recording mergers and acquisitions, reinvested earnings and income more thoroughly, the latter being a major reason for existing asymmetries. These examples are only a sample of the existing quality projects. This work is very important both in principle and in practice. However, my personal feeling is that it would be more useful to focus directly on external transactions in order to reduce asymmetries in major currency areas' balance of payments. This is because it is practically impossible to remove all asymmetries from different national statistics, at least at reasonable cost.

Another example with a significant European dimension I would like to take up here concerns the statute on the European Company, which is due to enter into force in October this year. This statute will remove some of the legal and regulatory impediments which thus far have prevented EU firms from fully exploiting the benefits of cross-border corporate restructuring. The European Company form would facilitate not only the merger of complicated group structures into a single legal entity but also the transfer of the company's head quarter from one Member State to another. Financial groups with pan-European or regional cross-border activities are also likely to be interested in the new company form. The new company form has important implications for financial stability, prudential supervision and crisis management.

The statute may have significant implications also for money and banking statistics, specifically for balance sheet statistics relating to monetary financial institutions (MFI) but also to other financial intermediaries. In principle, a change in legal status will have no direct statistical impact, since for statistical purposes foreign branches and subsidiaries are treated identically, regardless of legal status. However, in practice there may be several indirect effects. There is also a risk that economic transactions pertaining to national areas would no longer correspond to existing statistical reporting and that the reorganisation of business activities could significantly affect the information content and usefulness of national statistics.

An MFI operating as a European Company would be able to restructure its business activities on an ongoing basis and in a smoother fashion than is the case today. This could for example involve possibly recurrent reorganisations of branch activities across the European Company that might entail considerable implications on the statistical reporting of the various national branches. As an example of such a reorganisation, the administration of all derivatives business might be concentrated in one national branch. While the branches in the other countries may still continue to conduct derivative transactions with their customers, this activity made on behalf of the designated branch would no longer show up in the national statistics. That is, while the economic transactions still pertain to the various national areas, they may no longer correspond to the statistical reporting. In addition to derivatives, it is possible to envisage other examples of reorganisation, such as the concentration of the loan portfolio or securities holdings in the group’s head office. It may even be envisaged that a cross-border banking group may choose to record its deposit accounts in specific locations, for example in order to exploit differences in deposit guarantee systems. Such re-bookings could imply a substantial reduction in key items of the national monetary statis-tics of the countries hosting the group’s branches.

These have become current issues in Sweden, Finland, Denmark and Norway, with the Nordic banking group Nordea Bank announcing its intention of becoming a European Company and already taking steps to implement this legal restructuring. As a European Company the Nordea Group would be a single legal entity under Swedish law, and its present subsidiaries outside Sweden would become its foreign branches. Its legal restructuring and the consequent adoption of a branch structure would therefore be a major challenge for statistical authorities in Sweden, Finland, Denmark and Norway as well as in the Baltic countries.

Even though the implications for national monetary statistics may in some cases be substantial, the effect on euro area aggregates would be limited to situations where activities are transferred from or to euro area MFIs to or from non-euro area MFIs. However, this is precisely what is anticipated in the Nordea case, where the bank is active both inside and outside the euro area.

The handling of Finnish clients directly from the head office would mean that under normal procedures different deposit, credit and other items would be recorded only in the balance of payments, not in Finnish or euro area monetary or credit aggregates. Nobody can say how predominant this kind of customer relationship would be, but at most, some 30-40% of the amounts currently recorded in Finnish MFI statistics could be transferred to Swedish statistics. Such reclassifications would show up clearly even in euro area monetary aggregates.

As a further implication of the move to a single corporate structure, it is to be expected that cross-border banking groups will start to compile a single set of business accounts and be subject to the rulebook of a single supervisor. This is likely to generate a reorganisation of internal information management systems, as the “home country” reporting system becomes predominant.

I would like to add a few remarks on this discussion.

The imposition of extraordinary information requirements for a European company is expensive, and could run against the basic aims of the European Company statute. Therefore, reporting requirements to be imposed on European Companies need careful consideration.

If the same data collection procedures are adopted as before, Nordea would not report at all to the Finnish authorities on its direct transactions from Finland to Sweden for the balance of payments statistics. In this case, the Finnish authorities would need to collect data for the balance of payments statistics from Finnish bank customers, as is standard practice with Finnish residents dealing with non-resident banks. However, in practice, this would be difficult if not impossible, at least for households. On the other hand, it is not inconceivable that Nordea could provide supplementary information voluntarily. This practice would not be in line with current practices in balance of payments and monetary statistics, and would be possible only as an unofficial supplementary system. Unofficial national figures could perhaps be published, but one can foresee that confusion and problems could arise, for example with respect to the confidentiality of the figures. Furthermore, while in the case of Nordea there is every reason to expect good cooperation between it and statistical authorities, the risk of non-cooperative companies in future cases cannot be excluded a priori.

Nevertheless, in the future the use of complementary statistics is likely to emerge as an established procedure if the branch office structure becomes popular for banks operating across national borders. In any case, if the branch office structures and direct links between customers and head offices become prevalent, reliable national accounts and balance of payments statistics apparently require good cooperation between national authorities and possibly also multinational reporting systems. In order to prevent information requirements from becoming excessive, such systems should not try to cover all aspects and should avoid trying to include every detail.

As long as there are no such complementary data, complementary data for banking statistics could in principle be collected by utilising balance of payments data. Under standard statistical procedures, this procedure would be very difficult, because the correspondence between balance of payments and monetary and financial statistics is still rather weak. Furthermore, balance of payments statistics do not include information on currency denominations, to say nothing of information on interest rates etc.

Concluding remarks

When European statisticians have encountered difficulties in the classification and compilation of balance of payments data for the euro area, their prevailing – almost instinctive – reaction seems to have been to increase the collection of more detailed information. This strategy of increasing details and developing almost real time registers has, however, obscured the difference between fast indicators and structural statistics. The costs of compilation are exploding and in spite of all the computations, fragmented data does not lend itself easily to fast and effective quality control methods. The outcome is that the data will be overly fragmented and difficult to handle. This development may lead to a vicious circle of growing IT and personnel costs, combined with explosive growth in the amount of information.

Another approach could be to differentiate between the compilation of fast aggregate information and the compilation of structural statistics. As I have mentioned before, the respondent populations in financial statistics including the balance of payments are highly skewed. However, it is possible to identify reasonably nicely behaved subpopulations in view to the distributional properties. These subpopulations cover a relatively high proportion of the total value of the main aggregates facilitating low-cost collection and make possible the compilation of timely, nonbiased preliminaries of aggregates. These aggregates can be very useful, provided that their indicative nature is properly understood. On the other hand, the whole respondent population should be sufficiently covered for the more structural national statistics, which need not be as frequent and as timely as indicator statistics.

In the EMU, as probably elsewhere, the backbone of the collection and compilation of statistical data is provided by basic national statistical information systems. These systems provide plenty of useful information, also for monetary policy purposes. However they tend to be expensive, rigid and are not easily adapted to rapid changes in the economic environment brought about by policy decisions and globalisation. By their very nature, statistics are incomplete. Furthermore, one consequence of global and Euro-pean integration seems to be that national statistics will become increasingly uncertain. This conclusion certainly concerns also balance of payments and international investment position statistics. This does not necessarily imply that the significance of balance of payments statistics has diminished. The implication is rather that the responsibility of the user of statistics has increased: he or she should be able to interpret time series properly and cautiously.

If we want to maximise the support balance of payments statistics give to monetary policy uses, we should aim at developing good rather than perfect statistics. This implies that we should be selective when deciding what statistical items to produce; that the limited aggregate data set should be timely and reliable; and that data for it should be collected and processed as cost-effectively as possible, given the institutional constraints of the EMU. In order to improve timeliness, quality and cost effectiveness of balance of payments statistics, we should consider, in particular, a careful review of the sampling practices implemented at European and national levels.

What we need for policy purposes is a focused approach. In the first place, in order to maximise the support statistics give to the ECB's monetary policy we should be extremely selective in deciding what statistics should be produced for policy purposes. Secondly, this focused set of statistics should be produced as efficiently as possible, while respecting existing institutional constraints in the production of statistics.