Ahead of the banking crisis of 1988-1993, the household saving ratio fell sharply to negative levels (see Chart 1). When the economy turned, households reduced their debt and stepped up their saving. This led to reduced demand and weaker earnings and profitability in the enterprise sector. Losses on loans to enterprises increased, while losses on loans to households remained modest. In other words, for the banks, the indirect effects of reduced household consumption via increased losses on lending to enterprises were much stronger than the direct effects of losses on loans to households. (2) Analyses of household saving are therefore important in assessing the risk of financial instability.
Household saving consists of net fixed investment and net lending less net capital transfers (3) (see Chart 1). Buying a dwelling is a fixed investment for a household. If we adjust fixed investment for depreciation, we obtain net fixed investment. Households can change their financial wealth by buying or selling financial assets. They can also take out and repay loans. Household net lending is the difference between the change in financial wealth (adjusted for capital gains and losses) on the one hand, and the change in debt on the other. (4)
In the late 1980s, high levels of household borrowing caused household net lending to fall (see Chart 1). Overall saving decreased despite high fixed investment. Net lending picked up in the 1990s and was largely positive in the following years. In 2002-2005, households took out extraordinarily high dividends as a result of changes in the taxation of dividends. Some of these dividends were lent back to, or reinvested as shares in, the same company. Net lending grew during this period, but turned negative again in 2006. Adjusted for estimated reinvested dividend payments, household saving fell in the period 2002-2005. By historical standards, saving has been low since 2006. There might be a number of possible motives behind this low saving rate, including increased optimism among households and reduced uncertainty about economic developments. Whatever the motives, the result is that households are not building up financial buffers which can be used to counter unforeseen negative events. (5) In the light of historical experience, it is interesting to study which groups of households drive movements in saving when economic conditions change. In this article, we use micro data for households' financial wealth and debt to look at their net lending. Micro data offer an important alternative to macro data, especially when macro data give us limited information. Unlike macro data, micro data also make it possible to look at patterns in different groups of households, which can give us valuable insight into household saving behaviour.
[GRAPHIC 1 OMITTED]
In this article, we have grouped households on the basis of two characteristics: age and financial wealth. This choice of characteristics is, to some extent, affected by the calculation method. It is an advantage to have characteristics that are relatively constant over time in order to avoid excessive migration of households between groups. The chosen grouping is interesting in the light of changes in behaviour in different age groups. Changes in the taxation of dividends also make it relevant to look at household net lending broken down by financial wealth.
The article is organised as follows: Section 1 provides a brief description of the data. Section 2 looks at overall household net lending calculated using micro data and compares it with data from the national accounts and financial accounts. Section 3 examines whether any age groups have changed their behaviour in terms of net lending. We look at whether the decrease in overall net lending over the past 10-15 years is limited to specific age groups. Section 4 discusses whether specific groups of households had a particular impact on overall net lending in the period 2002-2005, when extraordinarily high dividends were taken out. Section 5 provides a summary.
Micro data for household financial wealth
Since the mid-1990s, Norges Bank has used micro data from Statistics Norway's income and wealth statistics to analyse household borrowing. In 2005, Norges Bank gained direct access to the detailed underlying data for household income and debt. Now, for the first time, Norges Bank has started to use the underlying data for household financial wealth broken down into individual items. In this article, we use data for household financial wealth from the income and wealth surveys for 1986-2003 (sample surveys) and from the full counts for 2004-2007.
The data are based on the items in tax returns. As tax regimes have undergone changes, most notably in 1992 and 1998, there are several breaks in these items. Some items have been removed, others have been added, or taxation has been extended to include more financial instruments. The first step in processing the data was to create time series for the various items for the period 1986-2007. One characteristic of the data is that they provide taxable values. For many financial wealth items, the taxable value is lower than the fair value, while debt is carried at fair value. To calculate net lending, the correct levels of both debt and financial wealth are required. The second step, therefore, was to calculate fair values for the various financial items. (6)
Overall household net lending
The estimates of household net lending at macro level are associated with a degree of uncertainty, both because different methods are used in the calculations and because information about households is not readily available. (7) Often, changes in household balance sheets need to be estimated on the basis of developments in other sectors. Net lending is also calculated on the basis of large aggregates. Small errors in these aggregates could result in major changes in the net lending calculated. (8) In some periods, there have been differences between net lending in the institutional national accounts (income accounts) and the financial accounts (see Chart 2).
Micro data are an alternative source for the calculation of net lending (see Chart 2). If we compare them first with the income accounts, we see that net lending from micro data follows movements in net lending in the income accounts to a certain extent, especially through to 2002.
[GRAPHIC 2 OMITTED]
There is not a direct accord between the definition of the household sector in the micro data and the income accounts. The income accounts include both households and non-profit organisations, whereas the micro data look exclusively at households.
To some extent, the financial accounts can correct for this difference. We can calculate net lending for households excluding non-profit organisations in the financial accounts, in any case since 1995. (9) In this calculation, we exclude financial investment in group insurance...