Crises et changements structurels dans la consommation alimentaire : un système structurel de demande

Abstract

Shocks and structural change in food consumption : structural demand system. This paper suggests a structural approach to demand modelling, which should help to better catch the modifications in consumer habits towards food and to identify sudden changes as those due to the release of information on food safety. In fact, traditional food demand models lose their reliability when significant structural changes occur over the time period under concern, especially when those changes are due to sudden food scares. Demand systems need to be modified in order to consider such changes. The structural approach suggested in this paper, once the hypothesis of parameter stability has been tested and rejected by appropriate tests, is based on the stochastic specification of some or all of the model coefficients. Maximum likelihood estimates of the parameters are obtained using the EM algorithm. An application on meat demand in Italy and the BSE crisis is proposed. The results show the relevance of the sudden BSE break of March 1996. However the changes in the expenditure share levels are accompanied by a sign inversion in the demand trend, which suggests that the initial shock tends to be reabsorbed. For the beef equation, the time-varying Marshallian own-price elasticity and the expenditure elasticity computed through the time-varying coefficients are analysed. The fall in the expenditure elasticity and the rise in the own-price elasticity suggest some effects due to the loss in consumer trust towards beef. Just after the food scare, prices tend to become a quality-proxy, so consumers avoid the low-price beef On the other hand, the demand for food safety is income-elastic and as beef is perceived as a less safe food, the expenditure elasticity decreases. This is just one interpretation to the 'lack of rationality' emerging from the data if the BSE crisis is not explicitly considered. In general, the possibility to have time-varying estimates opens the way to acquire information about the time patterns of food consumption habits, such as changes in trends or changes in price and expenditure elasticities. The limits of the suggested approach mainly lie in the large amount of data required by the estimation method, in order to avoid undesirable effects linked to the choice of the initial values.