Estimasi parameter model regresi data panel random effect dengan metode Generalized Least Squares (GLS)
| Main Author: | Rizki, Novi Aulia |
|---|---|
| Format: | Thesis NonPeerReviewed Book |
| Bahasa: | ind |
| Terbitan: |
, 2011
|
| Online Access: |
http://etheses.uin-malang.ac.id/6740/1/07610064.pdf http://etheses.uin-malang.ac.id/6740/ |
Daftar Isi:
- INDONESIA: Data empiris dalam suatu penelitian terdiri dari berbagai macam tipe, yaitu time series, cross-section, dan data panel, yang merupakan gabungan antara time series dan cross-section. Model regresi yang dibentuk dari data panel disebut model regresi panel. Dalam regresi panel terdapat tiga model regresi, yaitu model Common Effect, Fixed Effect, dan Random Effect. Model random effect digunakan untuk mengatasi kelemahan model fixed effect yang menggunakan peubah semu, sehingga model mengalami ketidakpastian. Model random effect menggunakan residual, yang dianggap memiliki hubungan antar time series dan cross-section. Oleh karena itu, estimasi perlu dilakukan dengan model komponen error atau model random effect. Karena data yang digunakan adalah cross-section, sehingga terjadi heteroskedastisitas, maka dilakukan estimasi melalui kuadrat terkecil yang diberlakukan secara umum atau disebut Generalized Least Squares (GLS). Dari hasil análisis, diperoleh estimasi β GLS = (X’WX)_-1(X’WY - Xμ) dan model regresi data panel random effect pada pengaruh kurs terhadap harga saham adalah Ýu = 131667.72 - 3727.581 - 2343.391 - 5464.264 - 1730.128 - 2.007681X ENGLISH: Empirical data in a study consists of various types, namely time series, cross- section and panel data, which is a combination of time series and cross-section. Regression models formed from panel data regression model called the panel. In a panel regression, there are three regression models, namely the Common Effect model, Fixed Effect and Random Effect. Random effect model is used to overcome the weaknesses of fixed effect models that use pseudo-variables, so the models have uncertainties. Random effect model using the residual, which is considered to have the relationship between time series and cross-section. Therefore, the estimate needs to be done with an error component model or random effect model. Because the data used are cross-section, resulting in heteroskedastisitas, then carried through a least squares estimation is applied in general or the so-called Generalized Least Squares (GLS). From the analysis, obtained estimates β GLS = (X’WX)_-1(X’WY - Xμ) regression model and random effect panel data on exchange rate effects on stock prices is Ýu = 131667.72 - 3727.581 - 2343.391 - 5464.264 - 1730.128 - 2.007681X