Abstract:

The consumer price index (TÜFE) is a real effective exchange rate (REK) that measures the competitiveness of the currency. Because the currency rate changes short intervals and is often downward output, investors need effective methods to reduce risks. This study aimed at comparing the predictive achievements of the artificial nerve network models, Box-Jenkins and the methods of top fixation with different architectures, determining the method that shows the highest predictive performance and producing monthly REK forecasts for the year 2019 with the help of the specified model. In the study, the real effective exchange rate data based on the 195 month period of January 2003 - March 2019 covered from the newsletter "Statistics of Currency Rates" published by the Republic of Turkey Central Bank. The estimated achievements of the models are assessed by the average absolute percentage error (MAPE) statistics.  As a result of the analyses carried out, the most successful resulting model among the various models applied was the Box-Jenkins Interseason-SARIMA(0,1,1)(1,0,0)12 model, with the help of the model obtained, the monthly forecast values for 2019 were produced.

Keywords: