Developing a Grey Wolf Optimization-Based Gray Box Model for Cash Flow Forecasting: A Study on Tehran Stock Exchange Companies

Cash flow forecasting is a critical aspect of financial planning and has long been of interest to investors and creditors. The Gray Box (GB) method is a widely used tool for forecasting, but a significant challenge in developing a grey box model is parameter estimation. In this study, we introduce a novel approach to cash flow forecasting called the Grey Wolf Optimization-based Grey Box model (GWOGB). The GWOGB employs a GWO algorithm as a global search method to determine the parameters of the GB model. To enhance the accuracy of future cash flow forecasting, we incorporate firm-level control variables as well as market and financial control variables into the classical model. To evaluate the performance of the proposed model, we use a sample of 250 firms listed on the Tehran Stock Exchange. Our empirical findings indicate that the GWOGB outperforms the generalized method of moments (GMM). Additionally, we employ sensitivity analysis to discern the source of forecast error and find that the inclusion of the nonlinear effect of sales growth rate on working capital accruals and future cash flow significantly reduces forecast error. The results show that using a nonlinear form of the GWOGB model is a promising approach for modeling the complex relationships between cash flow and working capital accruals.