Several regression models were tested that could be suitable for cost behavior prediction and for decision making. We used multiple linear regression models with point and interval estimates of the model parameters. We carried out comparisons of regression models for the cost behavior and their reliability, using the quality of the data collected for the case of basic and adjusted data of a concrete manufacturing company.
Our main area of interest is focused mainly on overhead costs. The classical approach of direct (variable) cost management is essentially determined by only one factor: the volume of production. The area of the overhead costs’ behavior and management is much more interesting and still unexplored; uncontrolled overheads alone can represent a significant cost reserve in many enterprises.
The overheads were divided into several groups of relevant costs, and their dependences were examined on multiple factors (rather than solely examining production volume) using the correlation matrix. We believe that asymmetric cost behavior ensuing from the results of the transformed model is affected by asymmetric behavior of the chosen factors. Also, the model representing the cost changes in time over one month was used and tested. This model can be used for examining shifting costs over time in a short period (e.g. months) and thus it is possible to prove cost asymmetric behavior, called “sticky costs.” We used the model adapted according to Anderson, Banker and Janakiraman (ABJ model) and we kept the model clearly transformed and assembled so that there remained only variables that had a statistically significant effect on the dependent variable. Based on these results, it is possible to perform the graphic analysis of deviations and cluster analysis to find out the similarities in costs through cost centers and/or through the examined periods.