article · Finance Research Open
An evaluation of exchange rate volatility for the Ghana Cedi against the US Dollar between January 2008 and December 2025 compares ten GARCH-family models with Facebook Prophet. Analysis of 216 monthly observations reveals that log returns are stationary with strong volatility clustering effects. A Threshold GARCH model with an autoregressive mean equation and Student-t errors emerges as the best-performing specification, passing all post-estimation diagnostic checks. This model captures notable leverage effects and indicates strong volatility persistence, with a calculated shock half-life of 25.74 months. Statistical testing confirms that the Threshold GARCH framework significantly outperforms Facebook Prophet, which displays high forecast error rates. Forward projections over twelve months anticipate conditional volatility declining from 5.31 in January 2026 to 3.09 in December 2026, offering relevant empirical guidance for currency stability measures.
Exchange rate fluctuations create severe balance sheet and planning risks for businesses, investors, and policymakers in import-dependent economies. Understanding the persistence of currency shocks allows institutions to design appropriate interventions. Proving that econometric methods outperform certain automated forecasting algorithms ensures market participants choose technically sound tools when protecting against foreign exchange turbulence.
The tested forecasting models could be directly applied by corporate treasuries, commercial banks, and central bank analysts to strengthen foreign exchange hedging and portfolio risk management. Although currently presented as applied empirical research rather than a standalone software product, the modelling framework is immediately usable within existing financial analytics platforms.
AI-generated from the published abstract. Always read the original work before citing.
This paper examines the volatility dynamics of the Ghana Cedi/US Dollar (GHS/USD) exchange rate in the period between January 2008 and December 2025 (n = 216 monthly average observations) using ten GARCH-family models and Facebook Prophet. The diagnosis of pre-estimation shows that the level series is non-stationary (I(1)) and the log returns are stationary (I(0)) with extremely significant ARCH effects (ARCH-LM Chi 2 = 45.83, p = 0.001). The best specification is determined to be the Threshold GARCH model with an AR(1) mean equation and Student-t errors (TGARCH(1,1)-AR(1)). All post-estimation diagnostics are acceptable at the 5 per cent level. Significant leverage effects are validated by the estimated asymmetry coefficient (η 1 1 = -0.490, p <0. 001). The volatility persistence of 0.9734 implies that the shock half-life is 25.74 months. The 12-month forecast indicates that conditional volatility will decrease from 5.31 (January 2026) to 3.09 (December 2026). Facebook Prophet has significantly greater out-of-sample forecast errors (MAPE = 41.69%) and the Diebold-Mariano test (DM = -4.278, p = 0.001) confirms that the TGARCH model makes statistically superior out-of-sample predictions, which provides significant implications for monetary policy, risk management and foreign exchange hedging in Ghana.
This page summarises published work. The authoritative version sits with the publisher.
DOI: 10.1016/j.finr.2026.100166
Is something wrong with this record? Report it or request removal.
Discussion
Have you built on this work, tried to replicate it, or seen it applied in practice? Share what you know. Verified researchers and MARATTO™ domain experts can open a discussion, and any member can reply. Contributions are reviewed before they appear.
No discussion yet. Open the first thread.
New to MARATTO™? Create a free account.
