Abstract
This research examines the dynamics and forecasting performance of currency in circulation (CIC) in Iraq. It uses a theoretical framework that recognizes long-run equilibrium relationships, short-run adjustment, volatility clustering, and nonlinear patterns. These complexities limit the adequacy of traditional linear models and motivate hybrid approaches. Accordingly, a triple hybrid model is employed. This combines autoregressive distributed lag (ARDL), Generalized Autoregressive Conditional Heteroskedasticity (GARCH) with Generalized Error Distribution (GED) to capture heavy-tailed behavior, and Bidirectional Gated Recurrent Unit (BIGRU). Stationarity is assessed using ADF tests. Residual diagnostic tests confirm heteroskedasticity, thereby justifying the inclusion of a GARCH component. The proposed model is compared with ARDL–GARCH and ARDL–BIGRU benchmarks. Empirical results based on RMSE, MAE, and MAPE demonstrate the statistically superior out-of-sample forecasting performance of the triple hybrid model. Bank deposits negatively affect CIC by absorbing liquidity, while withdrawals positively impact CIC, reflecting the persistence of cash-based transactions. These findings highlight the importance of advanced hybrid models for monetary policy and liquidity management in rent-based economies. Future research should include macroeconomic variables and examine the possible impact of Central Bank Digital Currencies (CBDC).
DOI
10.33095/2227-703X.4343
Subject Area
Statistical
Creative Commons License
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License
First Page
38
Last Page
52
Recommended Citation
Akram, A., & Ali, O. (2026). Combined Hybrid ARDL-GARCH-BIGRU Model in Analyzing and Forecasting Currency in Circulation Issued by the Central Bank of Iraq. Journal of Economics and Administrative Sciences, 32(1), 38-52. https://doi.org/10.33095/2227-703X.4343
