The Effect of Expenditure Aggregation on Hypothesis Tests in Consumer Demand Systems
用聚类分析识别随机希克斯加总商品，发现不当的支出加总会影响假设检验结果，建议在估计或检验需求模型前先用该方法确定合适的加总商品。
In this paper, cluster analysis is used to identify stochastic Hicksian (SHA). Such aggregates comprise goods for which relative prices are constant, up to a random element. The need to identify such aggregates arises since inappropriate aggregation could influence subsequent estimation and tests results. Evidence is found that inappropriate aggregation of expenditures influences the outcome of hypothesis tests. Furthermore, cluster analysis appears to identify goods which can be treated as SHA. It would thus appear that these methods should be used to identify the appropriate goods to aggregate, prior to the estimation or testing of demand models.