Proyeksi Kurva Imbal Hasil Obligasi Pemerintah Amerika Serikat Dengan Metode Hybrid Dynamic Nelson Siegel-Long Short Term Memory (LSTM)

Shafira, Alya (2025) Proyeksi Kurva Imbal Hasil Obligasi Pemerintah Amerika Serikat Dengan Metode Hybrid Dynamic Nelson Siegel-Long Short Term Memory (LSTM). Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5003211040_Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (4MB) | Request a copy

Abstract

Cadangan devisa memiliki peran penting dalam menjaga stabilitas ekonomi makro, khususnya di tengah ketidakpastian global. Bank Indonesia sebagai otoritas moneter bertanggung jawab dalam pengelolaan cadangan devisa, termasuk pemanfaatan instrumen seperti U.S. Treasury (UST), yang dikenal sebagai aset safe haven dan acuan pasar keuangan global. Kurva imbal hasil (yield curve) UST menjadi indikator utama dalam mengidentifikasi ekspektasi pasar dan risiko ekonomi, termasuk kemungkinan resesi. Penelitian ini bertujuan membangun model proyeksi yield curve dengan pendekatan hybrid Dynamic Nelson-Siegel (DNS) dan Long Short-Term Memory (LSTM). Estimasi parameter peluruhan DNS dilakukan dengan metode Newton-Raphson menghasilkan nilai 0,59776, diikuti estimasi parameter lainnya menggunakan Ordinary Least Square (OLS). Parameter DNS tersebut kemudian digunakan sebagai input dalam model LSTM dengan berbagai konfigurasi hyperparameter. Hasil evaluasi menggunakan Root Mean Squared Error (RMSE) menunjukkan bahwa model DNS-LSTM menghasilkan akurasi lebih tinggi dibandingkan pendekatan DNS-ARIMA pada seluruh tenor. Temuan ini mengindikasikan bahwa model hybrid ini lebih adaptif dalam menangkap dinamika kompleks yield curve, khususnya pada periode pascapandemi. Penelitian ini diharapkan memberikan pendekatan yang lebih adaptif dan berbasis data untuk mendukung pengelolaan cadangan devisa, memperkuat stabilitas nilai tukar, mengoptimalkan investasi, serta memastikan ketahanan ekonomi Indonesia dalam menghadapi tantangan global yang semakin kompleks.
====================================================================================================================================
Foreign exchange reserves play a vital role in maintaining macroeconomic stability, particularly amid global uncertainties. As the monetary authority, Bank Indonesia is responsible for managing these reserves, including the use of instruments such as U.S. Treasury (UST) securities, which are recognized as safe-haven assets and global financial benchmarks. The UST yield curve serves as a key indicator in identifying market expectations and economic risks, including the likelihood of a recession. This study aims to develop a yield curve projection model using a hybrid approach that combines the Dynamic Nelson-Siegel (DNS) framework with Long Short-Term Memory (LSTM) networks. The decay parameter in the DNS model is estimated using the Newton-Raphson method (yielding a value of 0.59776), followed by the estimation of other parameters through Ordinary Least Squares (OLS). These DNS parameters are then used as input features in the LSTM model, with various hyperparameter configurations. Evaluation results based on the Root Mean Squared Error (RMSE) indicate that the DNS-LSTM model achieves higher accuracy compared to the DNS-ARIMA approach across all tenors. These findings suggest that the hybrid model is more adaptive in capturing the complex dynamics of the yield curve, especially in the post-pandemic period. This study is expected to offer a more adaptive, data-driven approach to support foreign reserve management, strengthen exchange rate stability, optimize investment strategies, and ensure Indonesia’s economic resilience in the face of increasingly complex global challenges.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Obligasi, Imbal Hasil, Dynamic Nelson-Siegel, Long Short-Term Memory,Bonds, Yield, Dynamic Nelson-Siegel, Long Short-Term Memory
Subjects:	H Social Sciences > HA Statistics H Social Sciences > HA Statistics > HA30.3 Time-series analysis H Social Sciences > HB Economic Theory H Social Sciences > HB Economic Theory > Economic forecasting--Mathematical models. H Social Sciences > HC Economic History and Conditions H Social Sciences > HC Economic History and Conditions > HC441 Macroeconomics.
Divisions:	Faculty of Science and Data Analytics (SCIENTICS) > Statistics > 49201-(S1) Undergraduate Thesis
Depositing User:	Alya Shafira
Date Deposited:	01 Aug 2025 07:41
Last Modified:	01 Aug 2025 07:41
URI:	http://repository.its.ac.id/id/eprint/126059

Actions (login required)

View Item