Collective Decision-Making and Heterogeneity in Tastes
提出一种随机品味参数化方法，将家庭层面的PIGLOG需求系统扩展为二次形式，并利用加拿大横截面数据检验异方差性，发现允许异方差后理论性质与实证证据一致。
This article begins by proposing a random taste parameterization of a quadratic extension of the PIGLOG demand system at the household level, which is consistent with exact aggregation. This variation in tastes is a random function of household characteristics. The econometric implication is that a well-defined heteroscedastic error enters the demand system. This heteroscedasticity can be handled by a GMM technique. Using a large Canadian cross-sectional household data set and the Quadratic Almost Ideal Demand System, homoscedasticity is rejected; however, when heteroscedasticity is allowed for, there are no longer inconsistencies between the theoretical and empirical evidence, particularly with respect to the collective decision-making properties of homogeneity, SR1 symmetry, and distribution factor proportionality and linearity.