نوع مقاله : مقاله پژوهشی
عنوان مقاله English
نویسندگان English
Beniesh model (1999) is one of the basic models for detecting earning manipulation, which follows a linear process and cannot predict random behavior of EM. All complex dynamic behavior is produced by at least a balancing and reinforcing loops, but because Beneish model consists of only one reinforcing loop, it can not stop the exponential growth of the stock variable. This research has analyzed the data of 81 companies over 2016-2022 by Vensim, in order to developing the boundaries of the Beneish model and adding the balancing loop of corporate governance to it. After designing the eight-loops stock and flow diagrams of the Beneish model, it was changed to a single-loop model of reinforcing for ease of use, and then the proposed two-loop model was drawn including balancing and reinforcing loops. A total of 1296 simulations were performed. Due to the higher simulation error than the acceptable limit of 15%, exogenous parameters were calibrated of 1296 times by free random changes of the upper and lower limits. The simulation accuracy of the Beneish and proposal model was 84.72% and 98.07% after calibration, respectively, and the the validation error roots reached to the ideal value, which confirms the developed model correctness.