Optimal cost drivers in activity based costing based on an artificial neural network

Abstract

This study focuses on the development of Activity Based Costing (ABC) system by using optimal cost drivers (OCD) for the Thai automotive parts industry. Recently, traditional cost accounting (TCA) has been used to calculate production costs. However, the difficulty of TCA appears in the indirect or overhead costs which can be considered as a distortion production cost. Although the factory used the ABC system, inappropriate methods were utilized in order to solve this problem. The selected cost driver may not be the only factor affecting production costs. However, it was found that using OCD in ABC calculation resulted in more accurate production costs. The estimated production cost using artificial neural networks (ANNs) as a tool for identifying optimal production costs, because this method is effective for resolving both linear and non-linear problems. ANNs are designed and tested to estimate production costs by using the input and output data in the activities and production costs, and utilize a multi-layered feed forward and a back-propagation. The testing results of the production cost and the estimated cost for product A were applied to ABC by OCD in December, 2013. The production cost, estimated cost and mean square error (MSE) are equal to 47.337, 47.282 Thai baht, and 0.000036017, respectively.