Set Inferences and Sensitivity Analysis in Semiparametric Partially Identi ed Models
juan carlos escanciano,lin zhu
2013-01-01
Abstract:This paper provides inference tools for partially identi
ed semiparametric models. The main working assumption is that the
nite-dimensional parameter of interest and the in
nite-dimensional nuisance parameters are identi
ed conditionally on other nuisance parameters being known. This structure arises in numerous applications and permits the use of standard tools from empirical processes theory. We develop uniform convergence for a set of two-step GMM estimators and use the uniformity to establish set inference. We allow for the identi
ed set of nuisance parameters to be unknown and estimated, which in general requires to establish inference under local misspeci
cation. A global misspeci
cation analysis is useful in a formal development of sensitivity analysis. Inference is implemented with the assistance of bootstrap methods. Several examples illustrate the wide applicability of our results. Keywords: Partial Identi
cation; Semiparametric models; Sensitivity analysis; Willingness to pay; Empirical process theory. JEL classi
cation: C12, C13, C14 We would like to thank Stéphane Bonhomme, Kei Hirano, Arthur Lewbel, Enrique Sentana, Ellie Tamer, as well as seminar participants at UCL, ESEM Malaga, Oxford, CEMFI and Georgetown for helpful discussions. yDepartment of Economics, Indiana University, 105 Wylie Hall, 100 South Woodlawn Avenue, Bloomington, IN 47405 7104, USA. E-mail: jescanci@indiana.edu. Web Page: http://mypage.iu.edu/jescanci/. Research funded by the Spanish Plan Nacional de I+D+I, reference number SEJ2007-62908. zSchool of Economics and Management, Tsinghua University, Beijing 100084, China. E-mail: zhulin@sem.tsinghua.edu.cn.