2.3 Management Accounting and Technology
2.3.3 The Accounting View of Technology and of
Currently , accountants entertain a very simple view of technology (Bromwich and Hong, 1999; Noreen, 1991). Implicit in their view is a technology which is unlikely to reflect that employed by modern firms. This technology is portrayed in Figure 2.3 .
The available technology is shown by the area AOB. With the assumed technol- ogy, each unit of output requires so many units of material, say five, and so many units of labour, say three, in fixed proportions. This mixture of the two inputs is the only one that can be used. A lack of one input cannot be overcome by using more of the other input. No production is possible if either input is unavailable in the required amounts. Higher volumes of output simply require a larger bundle of the inputs to be used in the same fixed proportions. Thus both costs and technology are linear.
The above assumptions lead to there being only one efficient recipe in Figure 2.3 for the assumed volume. This recipe uses M 1 units of material and L 1 units of labour to produce our assumed 100 units of output, as shown by point O. With the assumptions, each unit of production requires 100/M 1 units of material and 100/
L 1 units of labour. Thus this technology allows no substitution between material and labour – adding more units of either input just wastes the inputs. The total cost
of the inputs is given by multiplying the quantity of each input by its price and adding these costs together. With a technology involving multi-inputs, it is implicit that there is no jointness between the inputs. This means that the technological rela- tionship between any two inputs is not affected by how much of a third input is used. The assumed technology therefore cannot capture the intricacies of modern cost functions. It rules out the possibilities of economies/diseconomies of scale and scope, and that resources may be used jointly by a number of cost objects.
2.3.3.1 ABC
It is not intended here to review the workings of ABC systems. These are well known and well handled in the text books. Nor is it intended to review the degree of usage in practice further than was done in Chapter 1 (a very good review of the studies of this and ABC generally is Gosselin, 2006). Here we look briefly at the assumed technology of ABC and consider whether this helps to explain its restric- tive use. ABC represents an attempt to understand the technology underlying some fixed overheads by relating that technology to activities rather than product vol- ume. Ideally with ABC the assumed relation between the inputs should be empiri- cally validated, but often the application of these assumptions is just a matter of belief. The accounting technology described above is also used by ABC. Thus, it is assumed that the relationship between inputs is fixed and is just scaled up for vol- ume of activity. Just as above we assumed a single product, a single and unique cost driver for cost pools is assumed in ABC. It is also assumed that activities in cost pools do not affect other cost pools. This assumption was implicit in the above anal- ysis, where it was assumed that our product was isolated from all other activities of the firm. In a multi-input situation, there is assumed to be no jointness between
A
B Material
Labour M1
L1 O
Figure 2.3 Technological Assumptions of Accounting.
the inputs used in a cost pool. These assumptions, together with fixed input prices, yield the familiar linear relationship between cost pool activity volumes and cost pool cost (Noreen, 1991). Cost pools cannot contain fixed costs as the assumptions of ABC assume that all inputs are fully variable.
These assumptions limit the possible application of ABC in practice. Many overheads do not behave in the assumed way. Thus while ABC and its variants are useful additions to the management accountant’s toolbox they are not a general panacea for all accounting problems. One possible explanation for the poor take- up perceived to be shown in empirical studies is that different industries may face different levels of overheads that conform with the assumptions of ABC and, of course, different intensities of overheads in costs. Al-Omiri and Drury (2007) sug- gest this may be the case when they found that 68% of the financial services firms in their sample used ABC whereas only 20% of manufacturing firms did so.
Recent research has suggested that practical ABC costing systems can be made more efficient by seeking to improve the firm’s cost architecture (see Datar and Gupta, 1994; Labro and Vanhoucke, 2007). This architecture comprises resource pools, cost pools and the number of pools entertained, the type of driver used for each pool and the precision of the measurement of each resource usage and cost driver (measurement error). Each of the above choices changes the structure of the costing system and alters the likelihood of errors and the types of error that can be expected.
Datar and Gupta (1994) for research purposes assume that the ideal cost structure is known, then derive variances from this ideal. Looking solely at cost pools and the products serviced therein, there are three possible types of product cost variance that cause the total costs of products to vary from their costs under the ideal system.
■ A specification variance is generated because the cost drivers used differ from those used in the ideal system, assuming that the ideal number of cost pools is used.
■ An aggregation variance arises where the number of cost pools employed is dif- ferent to that under the ideal system.
■ A measurement variance is generated where a less precise measurement of cost driver usage than is ideal is used in practice 3 .
Generally we would expect that more refined systems with more correct speci- fications and more cost pools will produce better results, even in the absence of knowledge of the ideal costing system. This is generally confirmed by Labro and Vanhoucke (2007), who simulate a large number of costing systems. Management, although without access to the ideal, will have ideas as to how the cost system may
3 More technically: specification variance ⫽ ideal product cost – product cost with ideal number of cost pools and known cost pool costs allocated using actual cost drivers for each cost pool. Aggregation Variance ⫽ Product cost with ideal number of cost pools – actual product cost.
be improved. Calculating variances using this system relative to using the bench- mark existing system may provide indications of whether the revised system is a better approximation to the ideal than the existing system. Where the input mix in a cost pool is fixed, it does not matter which of the inputs is used as the cost driver. Any input, or driver correlated with an input, may be used as the cost driver for the pool in this case.
Similarly, cost pools that have the same proportional input mix can be combined.
But proceeding by trial and error may make things worse. First, it is possible that difficulties in measuring correctly the cost driver usage by products will give rise to measurement errors. The likelihood of such errors increases with refinement of the accounting architecture. Thus with increased refinement, the measurement variance may offset improvements in the specification and aggregation variances (see Datar and Gupta, 1994). Secondly, Datar and Gupta (1994) show that seeking to improve some cost pools in a poor system may make things worse as previously variances may have been offsetting and refining part of the system, increases the total variance from the benchmark system. Thus, the general guidance is to con- tinue to refine systems unless the improved variances generated by refinements are negatively correlated with other known errors.