The Selection of the Best Estimation Model of the Composite Index on Stock Exchanges in Five ASEAN Countries Using Box Jenkins Method

This paper aims to obtain the Estimation Model of the Composite Index on Stock Exchanges in five countries of ASEAN. Those countries are Indonesia, Malaysia, Singapore, The Philippines, and Thailand. The estimation method used in this paper is Box Jenkins Method. The used data in building this estimation model are weekly data of composite index from 2000 to 2012 obtained from Blomberg. The best model is expected to be usable as one of references in estimating the composite index on those five stock exchanges. The best model selection is respectively begun by looking at the stationary data. Differencing will be conducted if the data are not stationary until they are. The next step is to do Autocorrelation Function (ACF) and Partial Autocorrelation Function (PAFC) analysis to gain ARIMA model. The best model selection is conducted by observing the coefficient significance on each model and select model with the smallest sum square error (SSE). Then, to ensure the selected model meets the assumption of white noise and normally distributed, the model diagnosis and normality test will be done using Kolmogorof Smirnof. Based on the analysis above, it can be decided that the estimation model of the best composite index from the five stock exchanges is ARIMA (2,1,1) for Indonesia, ARIMA (6,1,2) for Malaysia, ARIMA (5,1,5) for Singapore, ARIMA (2,1,0) for The Philippines, and ARIMA (4,1,0) for Thailand