Management Accounting
Advanced Management Accounting
In principle it would be misleading to use any of the established cost functions to estimate total costs for ranges, which are different from the one you computed the cost function with. The cost functions are only valid within the considered range of data. It could even be that computed fixed costs become negative, which is impossible in practice.
In general, staff costs correlate, as expected. Growing sales lead to increasing staff costs. It is statistically demonstrated that there is a very close match between the calculated line and the actual data with a pretty high coefficient of determination R2 of 0,856. Considering the R2 we have explained 85,6% of the variability of staff costs. The remaining 14,4% can be explained by other random variables. The cost function y=621.191,81+0,0149*x and the chart do not show any strange abnormalities. As the cost function shows us, there are pretty high fix costs of £ 621.191,81 and variable costs, which increase by £ 0,0149 with every pound of invoiced sales.
Having a look on the calculated cost function of the fuel costs, which is given by y=-2.827,34+0,0026*x, you will notice that the fixed costs are negative. As already mentioned this is because the cost function is calculated only for a specific range of data. The given data can be described the best with a cost function, which includes negative fix costs. The good fit of the cost function with the given data is demonstrated with the pretty high coefficient of determination of 0,961. Therefore, invoiced sales are a pretty reliable cost driver for predicting fuel costs. Whereas for predicting the warranty/refund costs the invoiced sales of the same month are not the best cost driver.
The very small value of the coefficient of determination (R2 = 0,001) shows that there is a limited fit between the calculated line and the given data. With such a small R2