الفهرس 
Quantile regression for dynamic panel data models
SummaryOnly 14 pages are availabe for public view
 |  | from | 167 |  |  |
| | | from | 167 | | |
Abstract
There is a single main approach for handling the problem of endogeneity in quantile regression (QR) for dynamic panel data. The widely used approach considers the inclusion of reliable instrumental variable(s) to minimize the biasedness problem. On the other hand, this thesis proposes another different approach that is concerned with the implication of two-stage quantile regression for dynamic panel data (TSQRDPD). Three different versions of TSQRDPD are newly introduced. The first version of TSQRDPD approach is concerned with estimating both stages by quantile regression with the same rank of quantile. The first stage is related to estimating the lagged dependent variable by one lagged value, and then include its fitted values in the second stage to estimate the parameters. The derivation of asymptotic distribution for this version of TSQRDPD approach estimator is derived. In addition, the general form of variance-covariance matrix is obtained and its consistency is proved. In addition, the properties of the new estimator in terms of equivariance, invariance to monotonic transformations and robustness in the frame of quantile regression are provided