半參數與參數估計法之比較與檢定:以荷爾蒙生長激素之採用為例
Semi-Parametric and Parametric Estimations and Tests of Adoption: An Application to Bovine Somatotropin Technology
美國康乃爾大學應用經濟與管理學系(Department of Applied Economics and Management Cornell University Ithaca, U.S.A);美國康乃爾大學應用經濟與管理學系(Department of Applied Economics and Management Cornell University Ithaca, U.S.A)
- 卷期
- 35期
- 出版
- 2005年12月
- 頁碼
- 001–026
- 編號
- NTUEJ00010696
你目前的連線不在校內範圍,所以無法開啟全文。書目與摘要不受限制;回到校內網路後重新整理本頁,全文就會直接顯示。
摘要
在許多實證領域中,參數估計法已經普遍地應用到二維選擇模型之探討,但是研究者往往忽略檢定此模型背後之基本假設。本文以紐約州酪農之生長賀爾蒙激素採用為例,來說明檢定參數二維選擇模型之重要性,並對傳統模型與限制較少的半參數估計法之估計結果加以比較。本文顯示,傳統的Probit 模型不但高估了實際的賀爾蒙生長激素採用率,並且高估了此賀爾蒙生長激素對牛奶產量之成
效。
Abstract
Although the parametric model is widely used for analyzing binary choice behavior, that approach is sensitive to model misspecification. We use the
rBST adoption behavior of New York State dairy farms as an application to illustrate the impact of this misspecification. In our application the conventional Probit model over predicts the adoption rate, and over estimates the impact of rBST on milk production per cow; almost twice the impact compared to the semi-parametric estimation.