Pseudo Integrated Least Squares Estimation for Single-Index Conditional Distribution Models


Ming-Yueh Huang
2010-06-04  12:30 - 14:30
Room 405, Mathematics Research Center Building (ori. New Math. Bldg.)

In this study, we considered a more flexible single-index conditional distribution model and proposed a pseudo integrated least squares estimation approach for the index coefficients. The existing methods will also be introduced and compared. In addition, the generalized cross-validation criteria were proposed for bandwidth selection and the bootstrap inferences were employed for the estimation of asymptotic variance and the construction of confidence intervals. With our defined residual process, a test rule is futher established to check the adequacy of the introduced model. For the sparsity of index coefficients, the adaptive LASSO for parametric models were extended to our model setting. Finally, a class of simulation scenarios is conducted to investigate the finite sample properties of the proposed estimators and assess the constructed inference procedures. (This is a joint work with Chin-Tsang Chiang)