The German Panel on Household Finances (PHF) is a panel survey on household finance and wealth in Germany, covering the balance sheet, pension, income, work life and other demographic characteristics of private households living in Germany. The panel survey is conducted by the Research Centre of the Deutsche Bundesbank.

The first wave of the PHF was carried out in 2010/2011, the second and third wave in 2014 and 2017, respectively. Wave 4 took place in 2021. All waves were conducted in cooperation with infas Institut für angewandte Sozialwissenschaften, Bonn. The following table shows the number of households interviewed in each wave. 

Wealthy households are oversampled on the basis of microgeographic indicators in order to better match the distribution of wealth across households and to shed light on the composition of wealth. A strong attempt is being made to select households from all economic strata. Participation is strictly voluntary.

The survey is designed to be a full panel, i.e. all households are re-contacted. The intended survey frequency is three years. Almost half of the 4,461 households in wave two took part for the second time. Two thirds of households who participated in wave three had already taken part in a previous wave. In wave four, the share of interviews with panel households in the total number of interviews reached 83%. Since 2010, a total of about 8,000 unique households participated in the PHF surveys, in some cases several times.

The results of all waves of our study were published in several Bundesbank monthly bulletin articles, reports and papers. The micro data from wave one to four are available for scientific research projects through the Bundesbank’s Research Data and Service Centre.

Aside from being an encompassing survey on household finance in Germany, PHF is an integral part of the Household Finance and Consumption Survey (HFCS). This system of wealth surveys collects ex ante harmonised micro data in every country of the euro area.

1. Background: A Euro Area Initiative

Households are characterised by a high degree of heterogeneity in their financial holdings and behaviour. Rich micro level data on household finances can be used for research and for analysing topics in household finance that are relevant to central banks. To achieve this goal, the euro-area national central banks and the European Central Bank (ECB), in collaboration with national statistical institutes, have set up a household survey that takes place in all euro area countries, the Household Finance and Consumption Survey (HFCS). The methodology and coordination of the survey is agreed by the Household Finance and Consumption Network (HFCN), which is composed of researchers and statisticians from the euro area National Central Banks, the European Central Bank and national statistical institutes.

In order to ensure comparability of results across countries, a set of harmonised output variables was agreed by the HFCN. A number of so-called 'core' output variables are to be provided by all participating countries. In addition, a set of ‘non-core’, non-compulsory, variables has been defined. Furthermore, a common blueprint questionnaire was developed by the HFCN. The blueprint served as the basis for the new surveys launched in several countries and is also the benchmark for existing surveys. Besides the core and non-core questions contained in the blueprint questionnaire, the questionnaire of each country may contain other country-specific questions and features.

2. Why do central banks care?

The "Household Finance and Consumption Survey" (HFCS), in general, and the German "Panel on Household Finances" (PHF), in particular, are central bank endeavours to collect micro-level information on household finance, as the available aggregate data are deemed insufficient. Measuring all relevant issues simultaneously at the individual level opens up the possibility of understanding structural relationships. Instead of being limited to observing the slow and jointly endogenous changes of averages and aggregates, micro data provide the cross-sectional perspective.

There are two major reasons for the central banks' efforts in collecting and making available detailed micro-level information. First, the financial conditions and financial behaviour of households have major implications for an economy’s development. And second, heterogeneity matters in household finance even more than in other areas. The "representative household" is a fiction which is often not helpful in understanding consumption, saving and the effects of monetary policy on them in much the same way as the concept of a "representative bank" is not helpful when dealing with financial stability issues.

Two examples: household indebtedness and the stock market puzzle

The informative value of aggregated data on household debt is constrained in a number of ways. The Bundesbank’s borrowers statistics show that the total debt of households stood at € 1,403 billion of the end of 2010. Dividing this figure by the number of households at that particular time yields an average household debt of € 34,813. However, such averages mask important information which can only be obtained using microdata. Figures from the PHF survey show that only 47% of German households are actually in debt at all. The households must consequently bear an average debt of approximately € 74,000. This is still insufficient, however. For financial stability, it is always the tail probabilities that matter, not the averages. It will be the heavily indebted households that take recourse to default. In order to gauge the risk, central banks have to assess how concentrated debt is and how much debt is borne by consumers with a debt service ratio that surpasses any given threshold. To make important concepts such as "value at risk" or "loss given default" really meaningful, central banks and other policymakers need distributional information.

The stock market participation puzzle is another case in point. Aggregate data, say, from financial accounts, provide information on how much public equity capital households hold in total. These holdings may be less than expected or warranted, leading to excess premiums. Micro data are needed even to perform the very first step, recognising that market participation is an issue: most households do not own any stocks at all. Micro data do far more than this, however: by providing multivariate distributions, micro data allow researchers to see who it is that holds shares, how many and what the concomitant characteristics are in terms of income, wealth, age, job security and financial education. This sort of information is necessary for solving the stock market participation puzzle. Careful calibration of macro models will be of no assistance.

3. The scientific focus of the PHF

The German PHF is an integral part of the Euro Area HFCS. Nevertheless, it has important special features that set it apart from the prototype HFCN survey and make it a major research endeavour in its own right. First, PHF places special emphasis on two key topics in German economic policy: savings and old age provision. Second, by establishing a full panel, PHF takes a life cycle perspective. In terms of methodology and content, there are two major precursors of PHF in Germany: the approach to household savings was pioneered by SAVE (Sparen und Altersvorsorge in Deutschland), a university-based study organised by MEA in Munich, and the self-refreshing panel structure is borrowed from SOEP (Sozio-oekonomisches Panel), conducted by the DIW (Deutsches Institut für Wirtschaftsforschung) in Berlin. The Bundesbank’s PHF is designed to be compatible with both, such that the three sets of survey data can be used complementarily.

Measuring Saving

Measuring household saving is a major conceptual problem for all wealth surveys. Conceptually, PHF can start from the complete asset side of a household balance sheet, as this is part of the core survey programme anyway. This opens up the possibility of measuring savings in two ways:

1. In the cross section, the questionnaire loops over all assets and asks for saving contracts. We use the fact that most savings in Germany are based on long-term contracts, the conditions of which we ask the household to specify. This yields gross flows into the most important saving vehicles in Germany: Riester pension accounts, private annuities, life insurances, building society contracts, savings accounts, mortgages, etc. In order to arrive at net saving, the survey also asks for discontinuous saving and the dissolving of savings. This approach was pioneered by SAVE.
2. Important complementary information will come from the panel dimension, comparing asset holdings over time.

Based on these two measuring concepts, the survey will be able to provide a very accurate picture of the saving dynamics in the medium and long run.

Old age provision

The saving decision is intimately linked with the old age provision system. The resident population in Germany is ageing rapidly. At the same time, the country is moving from a full pay-as-you-go system to a funded retirement system. This subjects active labour market participants to a double pressure. The basic problems of ageing societies are almost universal in rich countries, but details are very specific to each system. Thus, for meaningful analyses, country-specific data are required.

PHFmirrors the German pension system in the phase of transition, together with detailed information on wealth, savings, and work life. The survey will allow research to describe the dynamics of income, saving and wealth.

Wealth dynamics and family dynamics

Data on wealth distribution is scant in Germany and elsewhere. Beyond the statics of the wealth and income distribution, the PHF will permit us to observe their evolution. The PHF panel component is set up similarly to the SOEP and PSID approach. All original sample members are followed over time and interviews are conducted with all household members they live in.

By following all original sample members the PHF will give inside into the wealth accumulation process of households and persons over their life-time. The panel component fits very well with the two special topics mentioned above, as it will provide an additional measure of household saving by differencing the wealth levels from several waves and follows individuals as they enter retirement. Observing saving, the asset portfolio, transfers and inheritances in a self-refreshing, long-run panel will give us the Markov chain that governs the distribution of income and wealth. The survey will open the way to fascinating research on the interaction of family dynamics and finances.