4.2 Data and elasticities
The IHBS provides information about the socio-demographic characteristics and ex-
penditure levels of Italian households. Although this survey is monitored weekly and
published on a monthly basis, it does not provide any information about the quan-
tities purchased and the prices relative to the consumption of each good or service.
To obtain estimates of price (substitution) elasticities with a lack of survey informa-
tion, empirical works include household expenditures with aggregate national price
indexes. However, this approach requires a long span of cross sectional data to esti-
mate a demand system with sufficient price variation - features in which are almost
never available in empirical applications. Aggregate price indexes are highly corre-
lated due to the restricted number of categories of consumption normally analysed
in the demand system, leading to the rejection of theoretical restrictions and making
estimated elasticities highly uncertain. This problem was examined and discussed
recently by Coondoo et al. (2004).
For these reasons, surveys only gathering expenditure data have limited applica-
bility in modern demand and welfare analysis, unless if researchers are able to include
sufficient variability in prices. We believe that, by combining a regional price index
(RPI) - instead of a national price index (NPI) - with expenditure recorded in the
household surveys, we can respond to the issues discussed above satisfactorily, and
consistently estimate the demand system of interest made up of three aggregate cat-
egories: a) healthy foods; b) unhealthy foods; and c) other foods and nondurables.
The use of RPI is also important for identifying price elasticities. Above we iden-
tify the long AI model from a technical standpoint. However, a demand system also
requires that variations in prices should be a result of supply shifts. In this context,
it is assumed that regional price index differences reflect supply shifts rather than
movement along the demand curve. Following the line of argument of Gelbach et al.
(2007) and Zheng and Zhen (2008) we also assume for Italy that marketing healthy
foods like fruit and vegetables is more expensive than marketing unhealthy ones
like fats and oils, because the former entail higher costs in transport, refrigeration,
labour and packaging, and are much more prone to spoilage. But, in addition, the
use of a regional price index allows us to account for heterogenous differences in food
production and distribution across the Italian regions, and enable us to identify the
preference parameters for healthy food separately from unhealthy food.
Data on the consumer price index for the whole collectivity (NIC in ISTAT
14