Nonparametric instrumental variable derivative estimation

被引:6
|
作者
Florens, J. P. [1 ]
Racine, J. S. [2 ,3 ,4 ,5 ]
Centorrino, S. [6 ]
机构
[1] Univ Toulouse Capitole, Toulouse Sch Econ, Toulouse, France
[2] McMaster Univ, Dept Econ, Hamilton, ON, Canada
[3] McMaster Univ, Dept Math & Stat, Grad Program Stat, Hamilton, ON, Canada
[4] La Trobe Univ, Dept Econ & Finance, Bundoora, Vic, Australia
[5] Amer Univ, Rimini Ctr Econ Anal, Infometr Inst, Washington, DC 20016 USA
[6] SUNY Stony Brook, Econ Dept, Stony Brook, NY 11794 USA
基金
加拿大自然科学与工程研究理事会;
关键词
Nonparametric; derivatives; endogeneity; instrumental variables; ill-posed problem; regularisation; Landweber-Fridman; Engel curve; POSED INVERSE PROBLEMS; REGULARIZATION;
D O I
10.1080/10485252.2018.1428745
中图分类号
O21 [概率论与数理统计]; C8 [统计学];
学科分类号
020208 ; 070103 ; 0714 ;
摘要
The focus of this paper is the nonparametric estimation of the marginal effects (i.e. first partial derivatives) of an instrumental regression function phi defined by conditional moment restrictions that stem from a structural econometric model E[Y - phi(Z) vertical bar W] = 0, and involve endogenous variables Y and Z and instruments W. The derivative function phi' is the solution of an ill-posed inverse problem and we propose an estimation procedure based on Landweber-Fridman regularisation. We provide theoretical underpinnings of the proposed approach, examine finite-sample performance, and consider an illustrative application.
引用
收藏
页码:368 / 391
页数:24
相关论文
共 50 条