Övgücan Karadağ Erdemir (ovgucan@hacettepe.edu.tr)
Hacettepe University, Faculty of Science, Department of Actuarial Sciences, Ankara, Türkiye
Murat Kırkağaç (murat.kirkagac@dpu.edu.tr)
Kütahya Dumlupınar University, Faculty of Applied Sciences, Department of Insurance and Risk, Kütahya, Türkiye
Abstract
In this study, the behavior of private pension investment funds in Turkey, one of the most important investment instruments, was examined using time series analysis methods over a six-year period. Daily price, daily number of shares in circulation, daily number of people, daily total fund value and daily logarithmic return data of selected low, medium and high risk pension investment funds were converted into weekly average data. The movements of the weekly average values of the funds over time were examined graphically using time series analysis methods. The stationarity of the weekly average logarithmic return values of ALZ, AZS and AMZ funds was examined with unit root tests, and the stationarity process was applied to non-stationary returns. Steady weekly average logarithmic return values were modeled with appropriate Autoregressive Integrated Moving Average (ARIMA) models, a one-year forecast was made and compared with the real values. It has been observed that in low risk ALZ funds, forecast values that are closer to reality and have lower errors are obtained with ARIMA methods.
Keywords: Pension Investment Fund, Risk, Time Series Analysis, ACF, PACF, ADF Test, ARIMA
