the period 1997-2005.
Although we cannot directly retrieve the effects of food prices on weight outcomes,
we follow the schemes used by Chesher (1998), transforming the categories of food
consumption into energy (i.e., calories). By retrieving information from the Italian
Institute of Nutrition (INN, 1997), we can aggregate food categories, construct rela-
tive patterns, and associate then patterns of body weight. In particular, this source
provides data for over 1000 food items, which were first aggregated into 62 items -
the same source employed by ISTAT in its surveys - and further aggregated into the
categories of healthy and unhealthy foods considered in this study. The dotted line
of Figure 2, highlights the evolution of the energy changes (expressed in calories)
of unhealthy with respect to healthy foods. These patterns are close to those of
unhealthy versus healthy food consumption, indicating that changes in food prices
translate into changes in quantities of calories consumed by individuals, which may
contribute to explaining the increase in the share of overweight individuals, in line
with the argument of Drewnowski and Darmon (2005).
Although the patterns of the figures seem to be consistent with the working
hypothesis, our approach differs conceptually from that of works which use repre-
sentative time-series data to measure the elasticity of substitution (e.g., Zheng and
Zhen, 2008), assuming the existence of several effects of a unitary change in relative
prices on food budget share allocation across groups of consumers. For this reason,
the rest of this study focuses on measuring the elasticity of substitution by grouping
the household expenditures of the Italian surveys and estimating them according to
socio-economic and demographic characteristics.
At descriptive level we use the annual multipurpose surveys of Italian households
(2007) to discuss average changes in the prevalence of obesity and overweight for the
selected socio-economic and demographic groups. It is argued from this source that
the problem of overweight affects males specifically and increases with age, although
it should be noted that it is important at all ages. The rise in BMI disparities among
social groups is also reflected at educational level. In recent period, the only Italian
adult people who did not increase in the share of overweight and obesity were those
with higher education, whereas less educated adults showed a greater risk of obesity.
It is worth noting that, since the patterns of BMI shared by household income
or expenditure levels are not reported in the annual multipurpose survey of house-
holds, we can only assume the hypothesis of food substitution in these subsamples,