Modeling and Forecasting the Volatility of the USD/IDR Exchange Rate Using an ARIMA-EGARCH Approach
DOI:
https://doi.org/10.33021/0pb6ga86Abstract
This study models USD/IDR exchange rate volatility using an ARIMA and EGARCH framework to capture stylized financial facts like volatility clustering, leptokurtosis, and leverage effects. Using 1,318 daily observations from April 2021 to April 2026 partitioned into training and testing sets, the log-return series was confirmed stationary. Model selection identified an ARIMA of order 4, 0, 1 with zero mean as the optimal mean equation. Due to extreme conditional heteroscedasticity, an EGARCH of order 1, 1 with Student-t innovations was selected as the superior variance model with an Akaike Information Criterion of -8.0370. A significant leverage parameter of 0.3894 at a p-value below 0.001 confirmed asymmetric volatility responses to market shocks. Post-estimation diagnostics verified the absence of serial correlation and ARCH effects. The 277-step-ahead out-of-sample forecast demonstrated strong accuracy with a MAPE of 3.98 percent and RMSE of 713.11 IDR. However, the resulting MASE of 7.1398, being greater than one, supports the semi-strong Efficient Market Hypothesis, illustrating the inherent limitations of long-horizon predictability. These findings offer a robust volatility modeling framework suitable for foreign exchange risk reserving and designing hedging strategies in actuarial applications.
Downloads
Published
Issue
Section
License
Copyright (c) 2026 NOVIAN RIANDANA

This work is licensed under a Creative Commons Attribution 4.0 International License.