Survey of Household Finances (EFF): description, methods, and preliminary results

This article presents the Spanish Survey of Household Finances (EFF), a new survey conducted by the Banco de España that is included in the National Statistics Plan, recently approved by the Council of Ministers. The main aim of the EFF, whose methodology is described in Bover (2004), is to obtain detailed information on Spanish households’ wealth and financial decisions. In recent years, the modernisation of financial systems and the progress in most industrialised economies in consolidating a stable macroeconomic environment have wrought substantial changes in households’ saving patterns, financing and investment. In the Spanish economy, in particular, low interest rates, strong job creation, and the increased flexibility and competitiveness of the credit market have been conducive to an increase in household indebtedness. Similarly, broader investment opportunities, stemming from greater openness and financial innovation, have led to significant changes in the composition of household assets: the share of marketable instruments has risen, while that of traditional bank products has lost ground. These changes in the size and composition of households’ balance sheets have enabled them to take advantage of the developments in the financial and macroeconomic environment to improve their welfare. At the same time, households have become more exposed to certain macroeconomic or idiosyncratic shocks, such as those caused by changes in interest rates, asset prices or household incomes. This increased exposure has a direct impact on the determinants of household consumption and investment (see Malo de Molina and Restoy, 2004), which account for almost three-quarters of economic activity. In turn, the rise in debt taken on by households has increased the exposure of credit institutions and, consequently, of the financial system as a whole to changes in the factors governing the soundness of households’ wealth position. In its analyses of the economic situation and outlook, and in its reports on financial stability, the Banco de España has thus increasingly focused on the financial decisions and wealth of households. To date, these analyses have mainly been based on the Financial Accounts of the Spanish Economy (FASE), compiled by the Bank itself. The FASE contain a wealth of statistics, within the framework of National Accounts, on the aggregate balance sheet of the household sector and on its financial flows with the rest of the economy’s institutional sectors. However, they do not enable us to conduct disaggregated analyses that could prove useful for assessing all the implications of the position and financial decisions of households. Indeed, from a financial and a macroeconomic standpoint, it would be useful to obtain not only the position of the sector as a whole, but also the distribution in the population of the different assets and debts positions. For example, when estimating the potential impact of a rise in interest rates, it is important to know whether debt is proportional to households´ income or wealth or if it is concentrated among the least wealthy households. Likewise, the consequences of potential sharp corrections in asset prices depend on the capacity of the households owning these assets to absorb this shock and, accordingly, on their income and wealth. Factors such as age or labour market status of the members of the household also come into play. For this reason, the Banco de España decided in 2001 to launch the EFF, following in the footsteps of other countries that have been compiling this type of survey for some years. The Spanish survey is specifi cally based on that of the Banca d´Italia (Indagine sui Balanci delle Familie, IBF) [see Banca d’Italia (2000)] and above all on that of the US Federal Reserve’s Survey of Consumer Finances (SCF) [see Aizcorbe, Kennickell and Moore (2003)]. The EFF questionnaire contains a wide range of questions on assets, debt, incomes, spending and socioeconomic variables relating to the household sector. The availability of information on these variables for each reporting household unit will help refine financial and macroeconomic analyses of the Spanish economy. At the same time, the disaggregated information will be most useful for extending our knowledge of specific key questions, such as the relationship between the distribution of income and wealth in Spain, property ownership rates, the types of households with pension fund holdings and the financial and socioeconomic determinants of saving. Despite the fact that, in view of the complexity of the questionnaire, the number of responses (just over 5,000) is lower than in other official surveys, quality comparisons made drawing on data in other national statistics and on similar surveys in other countries yield satisfactory results. This is largely due to the sophisticated and innovative sample design, involving a blind system of collaboration between the National Statistics Office (INE) and the Tax Authorities. Naturally, we are also extremely grateful to the households that contributed to this survey. This article describes the main methodological features of the EFF and some of the preliminary results obtained in the first wave, whose reference period is 2002 Q4. In Section 2, we will briefly outline some relevant aspects of the making of the survey, a more detailed description of which is given in Bover (2004). The results are then presented in subsequent sections. Specifically, in Section 3, results on households’ incomes and net wealth are provided. In Sections 4 and 5, households’ holdings of real and financial assets are respectively analysed. Section 6 gives information on households’ debt levels and debt service burden. Section 7 focuses on aspects relating to household spending. And, lastly, Section 8 offers some concluding remarks. The main tables of results, which describe asset and debt holdings (Tables 3, 5 and 7), give the percentage of the different types of household holding each asset or type of debt, along with the median value of each instrument for households where this value is significant. Given that for most of these variables the distributions show very high values for a relatively small number of households, the median is a better approximation than the mean to the typical values of the distribution.