Page 319 - Special Topic Session (STS) - Volume 1

STS441 Andrea N.
sector (including nonprofit institutions serving households). FA do not
exclusively focus on the household sector but rather describe relations
between all institutional sectors. The definitions of instruments, sectors and
concepts such as valuation are given by the ESA 2010 and are mandatory in
all EU countries.
The two sources of information present differences in the definitions of
the household sector, in the periodicity, the timeliness, the reference periods
and in the valuation criteria.
In theory, once such notional differences are taken into account the two
sources of information should provide consistent information. Unfortunately,
this is not often the case.
Two major reasons for this discrepancy relate survey data. The first one is
unit nonresponse that happens when some groups of the selected households
refuse to participate in the survey. This is usually the case for very rich
households (in most cases they are even difficult to contact to negotiate the
interview). Unfortunately, rich households tend to concentrate a large share of
wealth in their hands and therefore such missing wealth will not show up in
survey data. Second, since wealth surveys include both complex and sensitive
items, respondents are not always able or even willing to report the correct
value of an item. The underreporting behavior generally differs across the
groups of the population or across wealth components. For instance, rich
households may be prone to underreport their wealth because of social
desirability bias. Moreover, financial assets tend to be more underreported,
whereas rich households tend to have larger portfolio shares of these assets.
The combination of nonresponse and underreporting leads to a general
missing wealth problem in surveys. Figure 1 shows the survey estimates of
total financial assets as a percentage of the national accounts. For most
countries, there is considerable discrepancy between the two sources of
information.

3. The experience of Banca d'Italia
Banca d’Italia has a long standing interest in the comparison of micro and
macro statistics relating household wealth. This interest was initially motivated
by the need to assess the quality of the survey results by comparing them with
external benchmarks.
Cannari and D'Alessio (1990) compared SHIW estimates of real estate
wealth with those derived from the Census. They found that the number of
residential properties was quite well estimated in the SHIW, but that the
number of rented and vacation homes was severely underestimated. In the
same paper the authors proposed a method for correcting the survey estimate
of the number of dwellings owned.

308 | I S I W S C 2 0 1 9