Likelihood-based Approaches for Inequality Restrictions Problems in Multivariate Linear Models with or without Incomplete Data

Abstract

Structure involving inequalities is often used in the statistical inference, e.g. genetic analysis, econometrics and clinical trial etc. In this setting, the parameter space is not open. Consequently, it is difficult to find the maximum likelihood estimates when the true parameters are lying on the boundary of the parameter space. Maximum likelihood estimation procedure under inequality restrictions A0?????? ???d a for complete or incomplete data are proposed in general linear model. The proposed procedure is illustrated with an example from income and consumption data during the years 1978-2002 in China.