ESTIMATING GENERALIZED METHOD OF MOMENTS BY USING MARQUADT-LEVENBERG ALGORITHM

• Main Author: Mukhlisah Abdal, Nurul
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/69636
• Institution: Institut Teknologi Bandung

Generalized Method of Moments is a method for estimating parameters using sample moments. GMM is used by the researcher specially in economics to determine econometrical models which their distribution function is hardly known. Not only for economics, but GMM also is useful for agriculture, transportation, health care, etc. Research methodology for this thesis is review of literature. There will be explanations about (1) Large Sample Properties for GMM, (2) GMM estimators are asymtotically normal, and (3) the other estimating parameter methods is special case for GMM. The point of this thesis is about combining GMM and Marquadt-Levenberg algorithm. The example for this thesis is estimating parameter for CIR model by using BI rate. The result of GMM-Marquadt-Levenberg algorithm estimators are compared to the result of GMM-Hybrid algorithm, combination of Marquadt-Levenberg algorithm and Genetic algorithm. The result shows that GMM estimator with Hybrid algorithm is better than Marquadt-Levenberg. It is because of the estimators have to go through two dierent algorithm which they have to satisfy both terms.