Abstract
This study concerned with the estimation of demand systems for agricultural products in OECD countries. Three representatives demand systems with their extensions, namely the Rotterdam Model, An Almost Ideal Demand System (AIDS), and CBS model are used. These models are estimated by Seemingly Unrelated Regression (SUR) method. The procedures to estimate demand systems suggest significant empirical regularities for agricultural products in OECD countries. The study also applies a procedure for model selection. This procedure implies the superiority of AIDS and CBS models over the Rotterdam model. The main contribution of this study is to model demand for agricultural products over a wide array of items and across large number of countries.
Notes
1 For a wider discussion on the likelihood ratio test for model selection, see Amemiya (Citation1985).
2 For a wider discussion on estimation methods for demand systems, see Edgerton et al. (Citation1996).
3 The diagnostics (Breusch-Pagan LM test for heteroscedasticity and Durbin-Watson test of autocorrelation) show that the models are generally robust. The results of the diagnostics can be requested from the author.
4 All the parameter estimates not reported here will be available upon the request from the author.