Süleyman Demirel Üniversitesi, Sosyal Bilimler Enstitüsü Dergisi, vol.34, pp.78-103, 2019 (Refereed Journals of Other Institutions)
a measure of the currency's competitive power. Because exchange rates
change over short periods and are often ups and downs, speculators need
effective methods to reduce risks. In this study, it was aimed to determine the
method with the highest estimation performance by comparing the estimation
successes of Artificial Neural Network models with different architectures, BoxJenkins and exponential smoothing methods and to produce monthly real
effective exchange rate based on CPI estimates for 2019 with the help of the
determined model. The study benefit 195 monthly data between January 2003
and March 2019 which was obtained from the "Foreign Exchange Rates
Statistics" bulletin published by the Central Bank of the Republic of Turkey.
Forecasting performances of the models were evaluated by the MAPE statistics.
As a result of the analyzes performed, it was found that Box-Jenkins
Multiplicaptive-seasonal ARIMA (0,1,1)(1,0,0)12 model was the most successful
one among the alternative models applied. With the help of the selected model,
monthly real effective exchange rate forecasts were made for the year 2019.