Pemodelan Simultan Spasial Data Panel Dinamis Dengan Pendekatan Instrumental Variabel Generalized Method Of Moment Pada Kasus Adanya Heteroskedastisitas

Kusrini, Dwi Endah (2022) Pemodelan Simultan Spasial Data Panel Dinamis Dengan Pendekatan Instrumental Variabel Generalized Method Of Moment Pada Kasus Adanya Heteroskedastisitas. Doctoral thesis, Institut Teknologi Sepuluh Nopember.

Text 06211460010007-Dissertation.pdf - Accepted Version Restricted to Repository staff only until 1 April 2025. Download (3MB) | Request a copy

Abstract

Penelitian ini bertujuan untuk membentuk model persamaan simultan spasial data panel dinamis (SSDPD) dengan kasus heteroskedastisitas. Model yang dibentuk adalah model efek tetap, dengan efek individu tidak dihilangkan tetapi ditempatkan dalam model kesalahan untuk mengakomodasi kasus
heteroskedasticity dalam model. Pengembangan estimasi Generalized Method of Moment (GMM) dipilih sebagai metode estimasi model SSDPD karena dapat menangani kasus heterogenitas pada model, dengan metode two stage least
square dynamic (GMM-2SLSD) dengan pendekatan single equation dan metode three stage least square dynamic (GMM-3SLSD) untuk pendekatan system equation. Efektivitas perkiraan dilihat berdasarkan nilai RMSE (Root Mean Square Error). Hasil simulasi Monte Carlo sebanyak 100 kali pengulangan dan
dengan pasangan parameter yang berbeda serta pasangan N dan T yang berbeda pula dapat disimpulkan bahwa perubahan skenario parameter tidak memberikan banyak efek pada nilai RMSE, tetapi perubahan N dan T memiliki pengaruh pada
nilai RMSE, dimana nilai RMSE dari penaksiran GMM-2SLSD mempunyai nilai lebih besar dibandingkan dengan RMSE dari penaksiran GMM-3SLSD. Nilai RMSE model error mempunyai kecenderungan akan naik nilainya jika nilai N
bertambah, hal tersebut disebabkan di dalam model error terdapat efek individu yang sangat dipengaruhi oleh pertambahan niai N. Perubahan nilai T dan skenario
parameter tidak terlalu berpengaruh terhadap nilai RMSE dengan menggunakan metode estimasi GMM-3SLSD dan RMSE model error dari pendekatan system, namun sangat berpengaruh terhadap nilai RMSE dengan metode estimasi GMM�2SLSD yang cenderung membesar dan tidak stabil ketika ada perubahan skenario parameter. Penerapan model SSDPD pada model LogFDI dan LogGDP/kapita memberikan hasil bahwa efek sinifikansi bobot spasial menunjukkan bahwa pengaruh LogFDI suatu negara juga mempengaruhi negara tetangganya, untuk model LogGDP/kapita dimana pengaruh bobot spasial variabel endogen dan endogen lag cukup besar, sehingga dapat diartikan LogGDP/kapita suatu negara akan mempengaruhi LogGDP/kapita negara yang punya hubungan perdagangan internasional dengannya, dan sangat dipengaruhi oleh LogGDP/kapita waktu
sebelumnya baik dari negara itu sendiri maupun negara tetangga. Sedangkan pengaruh variabel eksogen Inflasi dan REER tidak terlalu besar pengaruhnya saat kondsi yang lain Cateris Paribus. ========================================================================================================================
This study aims to form a model of spatial simultaneous equations of dynamic panel data (SSDPD) with cases of heteroskedasticity. The formed model is a fixed
effect model, with individual effects not eliminated but placed in an error model to accommodate cases of heteroskesdasticity in the model. The development of
the Generalized Method of Moment (GMM) estimation was chosen as the SSDPD model estimation method because it can handle heterogeneity cases in models, with the two stage least square dynamic (GMM-2SLSD) method with a single
equation approach and the three stage least square dynamic (GMM-3SLSD) method for the system equation approach. The effectiveness of the estimate is viewed based on the RMSE (Root Mean Square Error) value. Monte Carlo simulation results of 100 repetitions and with different parameter pairs and different N and T pairs it can also be concluded that the change in the parameter scenario did not have much effect on the RMSE value, but the change in N and T
has an influence on the RMSE value, whereas the RMSE value of the GMM�2SLSD estimation has a greater value compared to the RMSE of the GMM�3SLSD estimation. The RMSE value of the error model has a tendency to increase in value if the value of N increases, this is because in the error model there is an individual effect that is strongly influenced by the increase in the value of N. Changes in the T value and parameter scenarios do not have much effect on the RMSE value by using the GMM-3SLSD estimation method and the RMSE error model from the system approach, but it greatly affects the RMSE value with the GMM-2SLSD estimation method which tends to be enlarged and does not tend to grow and does not tend to grow and does not affect the GMM-2SLSD estimation
method which tends to grow and does not tend to grow and does not affect the stable when there is a parameter scenario change. The application of the SSDPD model to the LogFDI and LogGDP/capita models gave the result that the siniffic effect of spatial weights suggests that the influence of a country's LogFDI also affects its neighboring countries, for the LogGDP/capita model where the influence of the spatial weight of endogenous variables and endogenous lag is large enough, so it can be interpreted that the LogGDP/capita of a country will affect the LogGDP/capita of the country that has international trade relations with
it, and is strongly influenced by the previous LogGDP/capita time both from the country itself and neighboring countries. Meanwhile, the influence of exogenous variables Inflation and REER is not too big an effect when other conditions
Cateris Paribus.

Item Type:	Thesis (Doctoral)
Uncontrolled Keywords:	GMM, Simultaneous Spatial, Data Panel, Dinamis, FDI, ASEAN
Subjects:	H Social Sciences > HA Statistics > HA30.6 Spatial analysis
Divisions:	Faculty of Science and Data Analytics (SCIENTICS) > Statistics > 49001-(S3) PhD Thesis
Depositing User:	- Davi Wah
Date Deposited:	02 Dec 2022 06:13
Last Modified:	02 Dec 2022 06:13
URI:	http://repository.its.ac.id/id/eprint/95132

Actions (login required)

View Item