Temporal and regional analysis of private pension rate

Estimating participant of numbers in the individual pension system is essential for countries which have relatively quite short sectoral history like just 12 years as Turkey. In this study, it was determined which one is the best forecasting model by using participant data from Pension Monitoring Center. Number of participants is estimated for period from January 2016 to January 2020 by using the best model. Selection of time series forecasting models, 118 months participant data for training, 25 months forecasts are generated by utilizing Minitab R17 and Forecast Pro Trac V4 programs. 25 months estimated and actual participant data are tested to compare. The best one is determined among 9 models by comparing MAPE (Mean Absolute Percentage Error), MAD (Mean Absolute Deviation) and TS (Monitoring Signal) values between two datasets with table value. Validity of models is statistically controlled by residual analysis. Box-Jenkins Model is applied on data. Also the relationship between Turkish Lira and American Dollar exchange rate is examined by Wavelet transformation of individual pension system using MATLAB.