Research Article 9
Journal of Finance & Economics Research - Volume 11, Issue 1 2026
By Samina Parveen, Dr. Muhammad AsadUllah, Nayab Jumani
https://doi.org/10.20547/jfer2611102
Keywords: ARIMA, Exponential Smoothing Naïve, Ukrainian Economy, NARDL

This study investigates the dynamics and forecasting performance of the Ukrainian exchange rate against the US Dollar using a range of time-series and econometric models. Monthly data covering the period from 2010 to 2024 are employed for model estimation, while 2025 observations are forecasted to evaluate out-of-sample predictive accuracy. Prior to empirical analysis, unit root tests based on bothAugmented Dickey–Fuller and Phillips–Perron approaches are applied to ensure stationarity of the variables. Linear time-series dynamics are examined using ARIMA modeling, whereas nonlinear and asymmetric long-run and short-run relationships with key macroeconomic fundamentals are explored through the Nonlinear Autoregressive Distributed Lag (NARDL) framework. In addition, exponential smoothing techniques are used to capture trend behavior under uncertainty. Forecast accuracy is assessed using the Mean Absolute Percentage Error (MAPE), and further improvements are examined through forecast combination methods based on equal weighting and variance–covariance schemes. The results confirm the presence of a long-run equilibrium relationship between the exchange rate and selected macroeconomic variables, with asymmetric effects playing a significant role. Among individual models, NARDL demonstrates the highest forecasting accuracy, followed by ARIMA. Combined forecasts, particularly those integrating ARIMA and NARDL under variance–covariance weighting, yield the lowest forecasting errors. Overall, the findings highlight the importance of combining structural economic relationships with time-series dynamics to improve exchange rat

Submission Date: 12 Mar, 2026 Reviews Completed: 8 May, 2026
Acceptance Date: 29 May, 2026 Publication Date: 30 Jun, 2026