R. P. Byron

Date of this Version

July 1990

Document Type

Discussion Paper

Publication Details

R.P. Byron (1990) Polynominal Approximations in Cross-Sectional Models

The author is indebted to the Department of Industry, Technology and Commerce, Canberra, for financial support and to Andrew Edquist for assistance with the housing data supplied by that department.

School of Business Discussion Paper ; No. 3, Jul. 1990

© Copyright R. P. Byron and the School of Business, Bond University



The Almon technique of parsimoneously estimating distributed lag coefficients by polynominal approximations can easily be applied in a cross-section context using Zellner's SUR estimator. The application is perfectly natural as in many cross-section time-series models the regression coefficients can be expected to change in a systematic way as, for example, the grouping variable changes. The evoluUon of such coefficients lends itself to polynominal aproximation. Two applications are given, a successful one based on age specific household formation equations and the second (unsuccessful) on Grunfeld's investment data.