Developing a Dynamic Regression Model for Predicting Future Operating Cash Flow

The purpose of this research is to develop a dynamic regression model for prediction of future operating cash flows of firms accepted in Tehran Stock Exchange. So, the information of 250 companies were considered during 2004 to 2017. In this study, operational and economic variables were added to the fundamental model of Bart, Cram and Nelson (BCN). Due to the simultaneous effect of sales growth rate on working capital accruals (independent variables) and future operating cash flows (dependent variable) to fit the model, the gray box method was used with the help of the Pade approximant. To estimate the model were used three meta heuristics algorithm, grey wolf optimization, particle swarm optimization and inspired optic optimization. The results showed that the model which estimated by gray wolff algorithm has the least cash flows prediction error among all algorithms. In order to investigate the superiority of the gray wolf algorithm, the Friedman test was used. The results of this test also confirmed the superiority of the gray wolf algorithm in predicting future cash flows.