to substitute away from healthy towards unhealthy foods even if the magnitude
of the cross-price elasticity is not large. These changes appear to be consistent
in aggregate although they do not affect individuals equally and turn out to be
particularly important for poorer and less educated people. Thus, the results call
into question whether the government should intervene in the food market. Section
2 motivates our analysis by the descriptive analyses. It illustrates some descriptive
facts regarding the recent increase in overweight of the Italian population, matched
with patterns of relative healthy/unhealthy food prices. Section 3 presents the model
of interest and shows that an almost ideal demand system (AI) generally leads to
a consistent framework, even when it is extended to obtain long-run measures of
elasticities. Section 4 derives a long-run demand system from the dynamic system
approach and the identification issue concerning price elasticities is argued. Section 5
provides a range of cross-price elasticities and discusses policy implications. Section
6 concludes.
2 Basic facts
We begin this section by illustrating the patterns of the prices and quantities of
healthy and unhealthy foods purchased in Italy. The analysis serves to motivate a
theoretically consistent demand model, which follows in Section 3 and is based on
the classification (other than a residual food category) provided by Gelbach et al.
(2007) and adapted for features of Italian food consumption. We report the details of
this classification in Appendix A. This classification is built to highlight meaningful
substitution effects between food with high contents of fat or sugar (i.e., unhealthy
food) with respect to food with fewer calories (i.e., healthy food). It reflects, for
example, a strategy addressed to assess policy implications of taxes in some un-
healthy categories as their consumption grows when the prices of this category food
decrease. Although bread, pasta and olive oil are foods with high calorie contents,
we exclude these categories because they do not reveal a clear link with unhealthy
food. Rather, they are highly characteristic of the Mediterranean diet, in which
the habits and traditional aspects of Italian consumers preserve their consumption
behaviour. In addition, as discussed in Mazzocchi et al. (2008), medical studies
have shown that the balanced composition of these foods within the Mediterranean
diet leads to better conditions in terms of morbidity and general health.
As a benchmark of our model, we include these foods in a residual category,