Forecasting the Twelve Month Treasury Bill Rates in Sri Lanka: Box Jenkins Approach

Abstract

In this study, univariate time series Autoregressive Integrated Moving Average (ARIMA) model is used to forecast government twelve month Treasury bill rates in Sri Lanka over the period June, 2008 to June, 2013. Box Jenkins methodology is mainly used to build four models and different diagnostic tests and criteria were applied to select the appropriate model. The accuracy of the forecasted values is compared with Mean Squared Error (MSE) and Mean Absolute Error (MAE). The empirical results reveal that the best ARIMA model for the twelve month treasury bill rates is ARIMA (1,1,2). The obtained model was used to forecast next five weeks period and the results showed that the slow decay of the T-bill rates. The decreasing of the interest rates implies that the increasing the considerable demand for the government T-bills. Therefore the findings of this study have been given some impression to the investors for planning their future investments.