Providing real time estimates of private consumption is of paramount importance for policy-making. Structural macroeconometric models provide only a partial solution to the problem. They systematically neglect the up-to-date information contained in early indicators of economic activity. Moreover, they generally do not incorporate any measure of the uncertainty characterizing agents' behaviour. This paper presents a quarterly model based on early indicators of economic activity which allows to forecast the consumption expenditure of Italian households one semester ahead of the last release of the National Accounts. In the search for leading indicators we heavily rely on anticipations variables derived from the Isco survey on households. This allows to have some direct measures of consumers' expectations and subjective uncertainty. Although the emphasis of the paper is on prediction, we follow a theory based approach in modelling our consumption equations, as long as data permit. In the last part of the paper we tentatively pursue a counterfactual experiment, in order to evaluate the impact of the recent confidence crises on the expenditure pattern of Italian households.
An earlier draft of this paper was presented at the Conference on "Modelli per l'analisi economica a breve termine", organized by the Italian Statistical Society jointly with Banca Toscana (Florence, June 15, 1994).
