Costing systems: a new study has found that the accuracy of the time estimates provided by employees is far from perfect, which may affect the use of time-driven ABC.
The research considered how the measurement error of time estimates is affected by the following variables:
* The level of aggregation in the definition of costing system activities--ie, how many activities are included in one cost pool.
* The timing of the notification--ie, whether the respondents are aware before they start working on an activity that a time estimate will be required.
* Task coherence--ie, whether the time estimate error is smaller when employees' activities present themselves in a structured and systematic sequence.
* Whether the time estimate is provided in percentages or in absolute time units.
In terms of aggregation, our findings suggest that there is a trade-off between the level of aggregation and measurement error. New costing systems are designed to use less aggregation--ie, more cost pools--in order to increase the accuracy of the reported product costs. This works fine in cases where hard data from software systems such as activity time logs are available to management accountants. But, where such information is not available, staff are asked to fill out time surveys. In these cases, the very act of disaggregating a costing system by defining more activities may reduce accuracy. This can be attributed to the cognitive limitations of the process--ie, the level of estimation error gets higher when the number of tasks to be estimated increases. We suggest that caution should be exercised when costing systems are being disaggregated.
Management accountants are well aware that disaggregating a costing system into finer cost pools requires a big investment in system development and consultancy fees. Our research suggests that they should also be aware that the resulting new system may not be as accurate as they might hope.
Our results show that prior notification ie, telling people in advance that a time estimation exercise is coming up--improves the accuracy of their estimates. It has been argued that, when people allocate some of their mental resources to the conscious timing of each activity, their performance of those tasks is adversely affected. We found no evidence of such an intrusive cognitive effect among our survey sample (clerical staff). The performance of those who were forewarned was not significantly different from that of staff who were retrospectively notified. We recommend, therefore, that management accountants should try to warn employees beforehand that time estimates will be required from them.
Our findings indicate that the level of task coherence alone has no effect on measurement error. Although a costing system designer has no influence over the coherence of the tasks performed by employees, the factor does have an important effect when taken in conjunction with costing system parameters such as the level of aggregation and the timing of notification--both of which are under management accounting control. First, we have found that the combination of high disaggregation and low task coherence ie, many activities presenting themselves in a random sequence--results in the highest level of estimation error. We would advise management accountants to focus investments on improving the measurement systems in these circumstances, because this should lead to significantly lower estimation errors. Our findings also show that in situations where tasks present themselves in an incoherent order, prior notification that time estimates will be required is especially important in reducing error.
We have found that employees tend to dramatically overestimate the time they work when they're asked to provide their estimates in hours and minutes. This presents a huge problem when using the time-driven activity-based costing (TDABC) method (see panel, right). Proponents of this approach claim that, by using estimates in absolute time units rather than percentages, TDABC overcomes the problem of allocating unused capacity to products by using only practical capacity costs rather than full capacity costs. But our results call this assertion into question: 77 per cent of our respondents consistently overestimated by an average of 37 per cent. They indicated that they felt much more confident giving estimates in percentages rather than absolute time units. This lack of confidence may lead people to question the costing data that emerges and ultimately, to refuse to use such data in the decision-making process.
The findings of our research have direct implications for costing practice. Given the widespread use of time estimates in costing products and services, decision-makers who use the figures should be aware of the extent of measurement error that affects the accuracy of their costing systems. There is a need for costing system designers to find ways to reduce the size of such errors. This can be achieved by:
* Finding the right balance between the levels of aggregation and the levels of measurement error.
* Improving the design of the costing systems so that task coherence is balanced with other costing system parameters such as the level of aggregation and prior notification.
* Notifying employees beforehand that their time estimates will be required.
* Exerting caution when applying TDABC, which requires people to give estimates in absolute units rather than percentages.
* Investing in automated (online) time measurement systems. This should pay off most in terms of increased accuracy when tasks are presented in an incoherent sequence and activities are disaggregated.
What is time-driven activity-based costing?
CIMA's official terminology defines TDABC as "an approach to ABC based on time required for each unit of activity.
The method avoids the use of interviews with operating managers in order to estimate the percentage of time spent on different areas of work. It is claimed that TDABC based on 'time per transactional activity' is simpler to install and update and can highlight unused capacity."
The approach was developed by Robert Kaplan and Steven Anderson in 2004. They claim that this is a simpler version of ABC, where time is the sole cost driver. The method uses duration drivers instead of transactional drivers.
The time spent inactive is not taken into account, which, according to Kaplan and Anderson, overcomes a particular problem with traditional ABC, where the costing system often allocates full capacity to cost objects. The problem is that traditional ABC uses a percentage response mode--ie, when employees are surveyed, they tend to allocate 100 per cent of their time over a set of activities, with only few admitting to spending some of their time at work idle. By using absolute time units such as minutes, the TDABC approach aims to overcome this problem.
A CIMA Mastercourse entitled "Time-driven activity-based costing" will be held in London on June 5. For details, visit www.cimamastercourses.com
Eddy Cardinaels is associate professor of accounting at Tilburg University. Eva Labro is a lecturer in accounting at the London School of Economics. This article is based on their paper "On the determinants of measurement error in time-driven costing", The Accounting Review, Vol 83, No 3, 2008 (www.snipurl.com/5emu5). For an executive summary of their research report, visit CIMA's web site at www.cimagtobal.com/researchexecsummaries.