This project addresses the identification of consumption patterns, starting from a theoretical review of the particularities of the individual preferences underlying the Linear Expenditure System (SLE [1]). A method of estimating the Maximum Likelihood System proposed by Kao, Lee & Pitt (2001) is described, explained and implemented. To facilitate the estimation, the method contemplates the simulation of the part of the likelihood function that explicitly takes into account the absence of consumption of some goods by some households. To enrich the System, the method allows the inclusion of sociodemographic variables that are capable of determining consumption patterns. This method is applied to the study of consumption patterns in Bolivia, differentiated according to the geographical area: plains, highlands and valleys. In general, Bolivian households are found to prefer foods that have an intermediate degree of processing. It is found, among other important results, that this type of goods have a relatively high sensitivity to changes in their prices, household income and the prices of other food.
 
Published in Research & Development, No 9, 2009, pp.27 - 43​