Page 116 - Invited Paper Session (IPS) - Volume 1

IPS102 Ilja K. K. et al.
wealth distribution between households. This appears in the collection of data.
The FA data are often reported by counterparts, e.g. banks, as the HFCS data
are typically surveyed directly from the households. This can also lead to
different interpretation of economic concepts. The populations in the two
1
statistics differ slightly, i.e. persons living in institutions are excluded from the
survey population. Concerning the periodicity, timeliness and reference
period, the FA are quarterly statistics which are available within four months
of the reference period (last day of the quarter for balance sheet items). The
HFCS is conducted every three years in most countries and there is typically a
long lag between the data collection and data availability. Although the HFCS
has gradually synchronised its data collection, the fieldwork periods are still
varying from country to country. Finally, concerning the valuation, the FA
follow in principle market valuation, or a proxy of market valuation where there
is no active market and therefore no easily measurable market price for some
assets. Unlisted shares and other equity can be mentioned as an example of
these types of assets. The valuation in the HFCS is based on self-assessment
of households. This is supposed to be broadly similar to market valuation but
particularly in the case of less liquid assets households may not be able to
report market prices.
The work of the EG LMM aims at minimising the differences, i.e. to adjust
the data when it is possible. Adjustments are made for the definition of
household by using source-specific population figures of the two statistics in
the calculation of per capita values, i.e. assuming that the portfolio of
households living in institutions correspond with the average portfolio of the
whole population. Regarding adjustments for periodicity, the closest quarter
of the FA to the HFCS reference period – varying from country to country –
was chosen for the analyses.
Additionally, related HFCS and FA specific issues and potential errors have
been analysed. However, these are typically issues which cannot be corrected
in the short-term. In the case of the HFCS – and households surveys overall –
these are predominantly related to reporting and sampling bias. Particularly,
the sampling biases vary from country to country, as different countries are
dealing with those by means of different oversampling strategies, largely
depending on the availability of auxiliary data sources applicable for
oversampling. These different strategies affect the comparability of the results
between the countries. Additionally, the way of collecting data vary from
country to country. The majority of countries collect most of their data with
traditional surveys via CAPI (Computer Assisted Personal Interviews), but the
Netherlands collects their data through a web-survey and Finland uses a

1 For instance: hospitals, old people’s homes, residential homes, prisons, military barracks,
religious institutions, boarding houses and workers’ hostels, etc
105 | I S I W S C 2 0 1 9