Title: Identification and Estimation of Errors-in-Variables Using Nonnormality of the Unobserved Regressors
Abstract: This paper identifies and estimates the coefficients in a multivariate errors-in-variables linear model when the unobserved arbitrarily dependent regressors are not jointly normal and independent of errors. To identify the coefficients, we use variation in the second-order partial derivatives of the log characteristic function of the unobserved regressors; a property of only not jointly normal distributions. A root-n consistent and asymptotically normal extremum estimator performs well in simulations relative to third and fourth order moment estimators.
Publication Year: 2014
Publication Date: 2014-04-05
Language: en
Type: preprint
Access and Citation
AI Researcher Chatbot
Get quick answers to your questions about the article from our AI researcher chatbot