Abstract
This article presents several specifications of a retail sales model to explain the differential sales of retail goods and services in large Standard Metropolitan Statistical Areas in 1954 and 1963. Stepwise regression is employed and some hitherto unused regressors such as local government expenditures and taxes are introduced to compare different forms of the model specification. Tests on the stability of the relevant elasticities obtained in the model over time and on the predictive accuracy of this model are performed. The results of these tests support the stable elasticities hypothesis and show very high predictive accuracy for the model.