Monetary and financial stability constitute prime objectives for central banks. Monetary policy decisions are taken on the basis of information concerning developments in a number of economic and financial indicators. It is important that these indicators are reliable at an early stage, i.e. that they have good real-time properties, so that they can provide relevant input for interest rate decisions. Norges Bank releases financial and monetary statistics on a monthly and quarterly basis. These statistics form part of the input upon which the Bank's monetary policy decisions are based (1). Among the statistics are quarterly financial accounts for households and nonprofit institutions serving households (NPISH), which are compiled in the database system FINSE (FINancial SEctor accounts).
In Norway, considerable attention has been paid to the discrepancy between net lending / net borrowing in the financial and non-financial accounts, respectively, which has grown in recent years. There has been an increasing lack of consistency in the derived relationship between the resources generated by disposable income and borrowing on the one hand, and the use of resources on consumption expenditures and the acquisition of non-financial and financial assets on the other. A supply of financial resources that exceeds use may lead to the question: where does the money go?
Transparency is important to enable users to achieve a better knowledge of financial accounts and to facilitate the use of the statistics. This paper is intended to describe the main concepts of the financial accounts compiled in Norges Bank. The discussion is based on the FINSE system and mainly addresses issues linked to the financial accounts for households and NPISH. The purpose of the analysis is to draw attention to the weaker points in the financial accounts with the aim of providing an explanation for the causes of the observed discrepancies between the non-financial and the financial accounts.
Statistics Norway (SN) has the overall responsibility for classification, methods and principles in the Norwegian statistical system. SN also compiles and releases statistics on non-financial accounts. Responsibility for financial accounts is shared between SN and Norges Bank (NB). NB has the main responsibility for compiling and releasing statistics on securities market and the financial corporations sector. This also implies compilation of indicators for financial aggregates (money supply and credit supply) and compilation of financial accounts, which takes place in the database system FINSE. NB releases quarterly financial accounts for households and NPISH (i.e. the household sector), while annual financial accounts for all institutional sectors are released when financial accounts data are transmitted to Eurostat. SN is responsible for compiling accounting statistics for insurance enterprises and pension funds and has released a set of annual financial balance sheets for the period 1993 to 1997 (main instruments and main sectors).
Framework and observed discrepancies
In the national accounts system we face several identities, which in principle should be fulfilled. In our context this is also the case for the relationship between non-financial and financial accounts. In theory, net lending derived from the non-financial accounts should be identical to net financial transactions (2) derived from the financial accounts. However, experience shows that significant discrepancies occur for the household sector.
To start with, it is essential to emphasise that discrepancies in data may be ascribed to flaws and shortcomings in both sets of accounts. In both the non-financial and the financial accounts the balancing items are calculated on the basis of large aggregates. Even relatively small errors in these aggregates may result in large fluctuations in balancing items like net lending and net financial transactions. There are also differences in input statistics. Therefore, we are faced with a major challenge with regard to harmonising principles, methodologies and data sources in order to reduce these discrepancies as much as possible. This should, however, enhance user confidence in both sets of accounts.
The tasks of quantifying the financial assets, liabilities and financial transactions of the household sector are particularly demanding, as the data to a large degree come from indirect sources. A very limited portion of the statistics is based on household surveys appropriate for compilation purposes. The sector's financial accounts are therefore mainly based on data from administrative sources or counterpart sector information. Accounting statistics with reconciled operational accounts and balance sheets, which may be used to check compilation results, do not exist for the household sector.
The relationship between non-financial and financial accounts is illustrated in Chart 1. Both net lending and net financial transactions are calculated as residual items. In non-financial accounts, net lending is calculated as the difference between all income items and all expenditure items including consumption expenditures and the acquisition of non-financial assets.
In the financial accounts, transactions in every financial instrument on the financial balance sheet of the household sector are summarised by net financial transactions. To a large degree, financial accounts are based on statistics on stocks of financial instruments. The most widely used method in FINSE is to quantify financial transactions as residuals, subtracting all other known changes in assets from changes in stocks in the same period. For some financial instruments, directly observed transactions are used in the compilations. In these cases, the consistency between stocks and flows is maintained by the holding gains and losses quantified residually.
Chart 2 shows the discrepancy between non-financial and financial accounts. The chart covers the years 1996 to 2003 and is based on the most recently released statistics. The chart shows that the discrepancy is largest at the beginning and at the end of the eight-year period. For the years 1998 to 2001 the discrepancy is moderate and the general picture is quite consistent. However, the discrepancy widens considerably over the last two years of the period. The...