FORECASTING STOCK MARKET PRICES OF BENUE AND ASHAKA CEMENT INDUSTRIES USING THE BOX-JENKINS APPROACH

This study used Box-Jenkins approach to forecast stock market prices of Benue cement and Ashaka cement.Weekly data were collected for five months periods. Time series plot of the data for each of the companies under investigation was created. The time series indicated that there are possibly some events that influenced the price per share for cement because of the inconsistency of the behavior of the stock which displayed lowest price in week 26. The autocorrelation function for Benue cement stock prices tails off while the partial autocorrelation function cuts off after lag one thereby suggesting an autoregressive process of order 1, AR(1) for Benue cement stock prices. The model for the Benue cement stock prices sampled was Zt = 1.7759 + 0.9632Z_t-1 + a_t.This shows that this week’s price is a function of last week with a coefficient of 0.9632 and a constant of 1.7759 and some random error. The output result showed that all the parameters including the constant are significantly different from zero, because they have p-values that are significantly smaller than 0.05. The probability plot of the residuals reveals that they are essentially normal and the time series plot of the residuals contain only noise. These diagnostics indicate that a reasonable model has been found.The resulting models for Ashaka Cement sampled was AR (1). The model for sampled Ashaka cement is Zt = 0.2723 + 0.9916z_t-1 + a_t. The output result showed that the p-values are significantly smaller than 0.05 which implied that the parameters are significantly different from zero. The probability plot of the residuals also reveals that they are essentially normal and the time series plot of the residuals contain only noise.When examining the forecasts for Benue cement, for the data sampled, we noticed that the fits were very close to the actual. For the five months periods, the fits had a range of 49.50- 48.87. However, the actual prices had a range of49.50- 43.23. There was not much change in the fits because the coefficient of the autoregressive terms is 0.963 and a constant of 1.776.The fits of the model show slight downward trend because only a portion (0.963) of the last week prices is used to predict the forecasts. This is the same in behavior with the actual values, which also decreased during this time frame. The model can therefore be concluded to be an appropriate fit statistically the results would be desirable for an investor. Examining the actual versus the fits for the sampled data, their behavior was the same. The fits increased, while the

actual values also increased. This model would also be desirable for an investor.