Micro-Demand Systems Analysis of Non-Alcoholic Beverages in the United States: An Application of Econometric Techniques Dealing With Censoring

A censored Almost Ideal Demand System (AIDS) and a Quadratic Almost Ideal Demand System (QUAIDS) were estimated in modeling non-alcoholic beverages. Five estimation techniques were used, including the conventional Iterated Seemingly Unrelated Regression (ITSUR), two-stage methods such as the Heien and Wessells (1990) and the Shonkwiler and Yen (1999) approaches, the generalized maximum entropy method and the Amemiya-Tobin framework of Dong, Gould and Kaiser (2004). Our results based on various specifications and estimation techniques are quantitatively similar and indicate that price elasticity estimates have a greater variability in more highly censored non-alcoholic beverage items such as tea, coffee and bottled water as opposed to less censored non-alcoholic beverage items such as carbonated softdrinks, milk and fruit juices.