LARROSA Juan Manuel Ceferino
congresos y reuniones científicas
Compositional Engel Curves. A Dynamic Panel Data Analysis
LARROSA, JUAN M.C.; MARTÍN-FERNÁNDEZ, JOSEP A.
Congreso; VIII Jornada de Difusión de Investigaciones; 2005
Instituto de Economía - Universidad Nacional del Sur
Engel curves remain as robust tools for testing consumer behavior in Economics. Linear (or almost linear) relationship between income and food share expenditure has been found across the World by different academic works. We extend previous research by analyzing compositional statistic transformations and by extending this very same analysis into food components using dynamic panel data estimates. Compositional data analysis (CODA) remains as the ideal technique for analyzing this particular constrained data. We use for the first time isometric log-ratio transformations on data previous to estimation. At the same time, transformations require in several cases of zero replacement strategies for dealing with several food components that show zero presence across a sample of countries.