Estimating a Third-Order Translog Using Canadian Cross-Sectional Micro-data.

A third-order translog is estimated using Canadian cross-sectional micro-data, stratified by four family sizes. Third-order terms and family size are found to be statistically important explanatory variables. Homogeneity and symmetry restrictions are also rejected for the third-order translog, the translog, and the "almost ideal demand system." Directions for future research implied by these results include conditioning the third-order translog on more demographic variables, or relaxing some aggregation conditions imposed in the model used here.