Forecasting stock prices by using alternative time series models: The case of Istanbul Securities Exchange

Abstract

This study compares the forecast performance of alternative time series models at the Istanbul Securities Exchange (ISE). Considering the emerging market characteristics of ISE, stock prices are estimated by using money supply, inflation rate, interest rate, exchange rate, and government deficits. First the time series properties of the data set are examined and cointegration is tested. Next, univariate ARIMA models, VAR’s in levels and differences, and error correction models are specified and estimated using monthly data from 1986(1) through 1995(12). According to out- of-sample forecasting exercise it is found that the models assuming the existance of seasonality performes poor, the more parsimonious univariate ARIMA model have better performance than multivariate models.