Heterogeneity biases, distributional effects, and aggregate consumption: An empirical analysis using stratified microdata
利用加拿大FAMEX调查的分层微观数据，检验了不同阶层边际消费倾向的异质性是否阻碍了加总过程。发现尽管边际反应存在显著异质性，但若分层需求在均值附近近似线性且商品非奢侈品，加总预测偏差很小；同时，混合分层数据的截面估计会受到交叉矩的污染。
Abstract Using stratified microdata from the Canadian FAMEX (78–86) surveys, this paper investigates whether observed heterogeneity in marginal propensities to consume across strata actually hinders the aggregation process. Despite significant heterogeneity in marginal responses, the divergences between aggregate predicted consumption and the predictions from a model that uses average strata responses are found to be small, whenever the strata demands are approximatively linear at the mean and the commodity group considered is not a luxury good. On the other hand, some cross‐sectional estimates obtained by pooling the strata are shown to be contaminated by unwanted cross‐moments. Further, the analysis reconciles the fact that while there exists significant heterogeneity in consumer demands, the related distributional effects in the aggregate equation have not been found to be important.