Estimation of A Partial Linear Model with Instrumental Variable for the Longitudinal Data

Abstract A partial linear model with instrumental variables was developed for longitudinal data. In the partially model, explanatory variable is an endogenous variable, which correlated error term. The expressed by and item. estimated through least square method. B-spline regression combined QR decomposition used to approximate nonparametric function. For estimation of parametric, Quadratic inference function Secant method were applied. Under some conditions, estimator consistent asymptotic normality. Some simulation conducted prove finite sample behavior estimator.