1
Department of Accounting, Ker.C., Islamic Azad University, Kermanshah, Iran.
2
Department of Economics, Ker. C., Islamic Azad University, Kermanshah, Iran.
10.22099/ijes.2026.55803.2111
Abstract
This study examines the asymmetric and nonlinear effects of exchange rate fluctuations on Iran’s stock market index and industrial production over the period 2009–2024, with the explicit aim of quantifying the differential sensitivity of financial and real sectors to currency depreciation and appreciation shocks under varying macroeconomic regimes. A hybrid analytical framework is employed, integrating the Nonlinear Autoregressive Distributed Lag (NARDL) model for asymmetric cointegration analysis, a Hidden Markov Model (HMM) for endogenous regime identification, Gaussian Process Regression (GPR) for Bayesian uncertainty quantification, and a Transformer architecture for temporal attention-based forecasting. Monthly data spanning 2009–2024 are utilized, with all variables confirmed as through ADF, PP, and KPSS unit root tests. Bounds testing confirms stable long-run cointegrating relationships in both models ( and , respectively). Long-run asymmetry is significant in both sectors; negative shocks exert effects approximately times larger than positive shocks in the stock market ( vs. ) and times larger in industrial production ( vs. ). The HMM identifies two structurally distinct regimes — stable and turbulent — wherein turbulent-regime coefficients exceed stable-regime estimates by a factor exceeding two. The hybrid framework achieves RMSE reductions of 38% and 24% over benchmark models for the stock market and industrial production, respectively. Exchange rate shocks in Iran operate through asymmetric, regime-dependent, and nonlinear channels, with financial markets exhibiting substantially greater sensitivity than the real sector. Symmetric currency management policies are structurally inadequate; preemptive, regime-sensitive intervention instruments are required.
Malihi, S. A. , Moradi, A. , BaghfalakI, A. and Mohammadi Yarijani, F. (2026). Deep Learning-Enhanced Cross-Sectoral Analysis of Exchange Rate Asymmetry: Evidence from Sectoral Stock Markets and Industrial Output. Iranian Journal of Economic Studies, (), -. doi: 10.22099/ijes.2026.55803.2111
MLA
Malihi, S. A. , , Moradi, A. , , BaghfalakI, A. , and Mohammadi Yarijani, F. . "Deep Learning-Enhanced Cross-Sectoral Analysis of Exchange Rate Asymmetry: Evidence from Sectoral Stock Markets and Industrial Output", Iranian Journal of Economic Studies, , , 2026, -. doi: 10.22099/ijes.2026.55803.2111
HARVARD
Malihi, S. A., Moradi, A., BaghfalakI, A., Mohammadi Yarijani, F. (2026). 'Deep Learning-Enhanced Cross-Sectoral Analysis of Exchange Rate Asymmetry: Evidence from Sectoral Stock Markets and Industrial Output', Iranian Journal of Economic Studies, (), pp. -. doi: 10.22099/ijes.2026.55803.2111
CHICAGO
S. A. Malihi , A. Moradi , A. BaghfalakI and F. Mohammadi Yarijani, "Deep Learning-Enhanced Cross-Sectoral Analysis of Exchange Rate Asymmetry: Evidence from Sectoral Stock Markets and Industrial Output," Iranian Journal of Economic Studies, (2026): -, doi: 10.22099/ijes.2026.55803.2111
VANCOUVER
Malihi, S. A., Moradi, A., BaghfalakI, A., Mohammadi Yarijani, F. Deep Learning-Enhanced Cross-Sectoral Analysis of Exchange Rate Asymmetry: Evidence from Sectoral Stock Markets and Industrial Output. Iranian Journal of Economic Studies, 2026; (): -. doi: 10.22099/ijes.2026.55803.2111
)