Comparative Analysis of Profit Forecasting Models for Companies Listed on the Palestine Exchange in the Insurance Sector رسالة ماجستير

Abstract

In the present thesis, the author has defined an objective, namely, to test the profit forecasting models of eight insurance companies listed on the Palestine Stock Exchange in the period between 2013 and 2023.It is with a view to four statistical models: the random walk model, the arithmetic mean model, the weighted average model, and the modified trend model. These models were scaled according to the lowest selection rules, i.e., the mean absolute error of the model, the root mean square, mean absolute error ratio, and coefficient of determination to obtain the most appropriate model that can be applied in servicing insurance industry. As the results of the experimental analysis show, the modified trend model was the most efficient of all the companies considered since it produced the lowest percent of error in predictive forecasting and the most effective determination factor value. The weighted average model came second and was characterized by its rapid response to movements in the values of earnings. The mean arithmetic model demonstrated average accuracy rate in the absolute error rate and the coefficient of determination. The random walk model produces the lowest estimate of the absolute error rate and coefficient of determination, since, like no other of these, it cannot determine any trend or structure in the data. The results affirm the existence of a deterministic structure and cyclicality of earnings behavior by Palestinian insurance companies. They emphasize the importance of selecting the right models to predict this behavior, owing to the real peculiarities of the financial statements. The scholar implies that the modified trend model may serve as a proactive tool of financial planning and strategy in that industry.