文章基本信息

标题：How Much Do Starting Values Really Matter? An Empirical Comparison of Genetic Algorithm and Traditional Approaches
本地全文：下载
作者：Tonsor, Glynn T. ; Kastens, Terry L.
期刊名称：Journal of Food Distribution Research
印刷版ISSN：0047-245X
出版年度：2006
卷号：37
期号：SUPPL
出版社：Food Distribution Research Society
摘要：This research evaluates the impact of using different starting conditions in estimating meat demand systems. Results suggest that as the econometric task becomes increasingly nonlinear, specification of starting conditions becomes increasingly important. This work demonstrates implications of failing to use the best available starting value conditions and how these implications vary with the complexity of the underlying econometric model of interest. Furthermore, this piece proposes a universal approach to be used by all applied econometric practioners to developing appropriate starting values for use in subsequent model estimation.