• consumption
• forecasting
• joint density
• microdata
• projections
• surveydata

The aim of this work is to provide projections for the consumption of meat and other animal source foods in the United States, using representative microdata from National surveys. Following a data-driven approach, we firstly estimate a semi-parametric model for the consumption of meat in the United States, using the age structure and the income distribution of the population as the main explicative regressors, and controlling for other socio-demographic characteristics of the individuals. The main body of the thesis then deals with the hurdles of obtaining projections of the individual consumption from cross-section data for different years rather than from a (more advisable but not available) panel-data structure.