An Analysis of Consumer Demand in Switzerland : Estimation of a Quadratic Almost Ideal Demand System with Censored Alcohol Consumption

In this thesis, I estimate a Quadratic Almost Ideal Demand System using Swiss household expenditure data from 2006-2017. To control for the censored budget share variable of alcohol, I implemented a two-step Heckman-type model for consistent estimation of the demand system. Furthermore, I computed Stone-Lewbel-like prices to increase price variation, and controlled for expenditure endogeneity with an augmented regression approach. For the estimation, I partitioned the sample into consumers and abstainers of alcohol and enhanced each demand systems with the appropriate inverse Mills ratio. The two models were then estimated with an iterated linear least squares estimator. Firstly, I find that income elasticities are in the interval of ca. 0.3 (Food) to 1.7 (Recreation). Secondly, uncompensated own-price elasticities range from -1.3 (Others) to -0.2 (Alcohol). All own-price elasticities are of a slightly lower magnitude in the compensated case due to the income effect and the fact that all commodity groups are normal goods. Lastly, theoretical restrictions are empirically tested. The results suggest that homogeneity and symmetry are rejected, while approximately one-third of households in the sample satisfy negativity