Agenda

26 Gen 2022 17:30

Tao Wang - Endogeneity in Modal Regression

Online

Tao Wang (University of California, Riverside (UCR)) - Endogeneity in Modal Regression 

Abstract:
In this paper, we propose a control function approach to account for endogeneity in a parametric linear triangular simultaneous equations model for modal regression, where the conditional mode of the unobservable error term on explanatory variables is nonzero. We adjust endogeneity with the residuals from the conditional mode decomposition of the endogenous variable as controls in the structural equation, and develop a computationally attractive two-step estimation procedure with the conditional mode independence restriction. The proposed estimators could be conveniently solved by virtue of a modified modal expectation-maximization (MEM) algorithm. The consistency and asymptotic properties of the estimators for both parametric and nonparametric parts are rigorously established under generic regularity conditions, and the estimation of the nonparametric component is oracle. Monte Carlo simulations are conducted to evaluate the finite sample performance of the proposed estimation procedure. To motivate the proposed control function method, we introduce a dynamic model of rational behavior under uncertainty, in which the agent maximizes the present discounted value of the stream of future modal utilities, and develop a modal Euler equation derived from the maximization model that the agent must satisfy in equilibrium. We estimate the modal elasticity of intertemporal substitution (EIS) directly from the stochastic Euler equation. Two other applications to the real datasets of Return to Schooling and Colonial Origins of Comparative Development are presented to further illustrate the proposed estimators in practice. We in the end construct an adaptive least absolute shrinkage and selection operator (LASSO) technique for selecting instrumental variables and demonstrate the oracle property of the proposed penalized modal regression model. Since mode is identical to mean with symmetric data, several remarks on modalbased control function mean estimation are also addressed for the sake of thoroughness.

The seminar can be attended connecting to ZOOM: https://unive.zoom.us/j/81421353937

ID riunione: 814 2135 3937

Lingua

L'evento si terrà in inglese

Organizzatore

JMSeminars

Cerca in agenda