Summary by Antoinette L. Lynch
Ph.D. Program in Accounting
University of South Florida, Spring 2002
A growing body of research is analyzing the success of activity-based cost management (ABCM). This paper examines the effect of using alternative approaches to measuring success. The two approaches examined are (1) an a priori classification approach (use in decision making, actions taken with ABCM information, dollar improvements and management evaluation as to overall success); and (2) a factor analysis approach.
ABCM success is a multi-dimensional construct. Broad-based success measures (especially those incorporating details of how ABCM data is used in decisions) yield the highest explanatory power. A common success measure in published research is a management estimate of dollar improvements (SM-$I) or of the overall net benefits (SM-ME) from adopting ABCM. The authors show that a broader-based success measure can improve the explanatory power of the models examined using SM-$I or SM-ME by over 50 percent. One implication is that using better specified success measures will yield more reliable inferences about ABCM success determinants. Consistent with the prior literature, the authors find support for success determinants such as top management support, and the use of ABCM data in performance evaluation/incentive systems. Two variables not included in prior studies (number of applications and time-in-use of application-individually) are also significant in explaining success differences across ABCM sites.
ALTERNATIVE ABCM SUCCESS MEASURES
Research to date has used a diverse set of ABCM success measures. At least four a priori types of measures can be distinguished:
1. SM-U. Measure based on the use of ABCM information in decision making. This measure assumes that the more extensive the use of ABCM information, the more successful the implementation. Examples include Cotton (1993), Lukka and Granlund (1994), Innes and Mitchell (1995) and Krumwiede (1997).
2. SM-DA. Measure based on decision actions taken with ABCM information. Using this measure, when an ABCM implementation causes a change in decisions, it is viewed as successful; when it causes no change in decisions, it is viewed as not being successful. Innes and Mitchell (1995) use this measure.
3. SM-$1. Measure based on the dollar improvements resulting from ABCM. This measure reflects either a summary management estimate, or an explicit dollar comparison of revenues and costs with and without ABCM. There may be a time dimension to this success measure if there is a delay between when ABCM is implemented and when dollar improvements become apparent. Examples include Shields (1995) and Krumwiede (1997).
4. SM-ME. Measure based on management evaluation as to the overall success of ABCM. This measure is typically based on an unspecified definition as to how success is to be interpreted. Examples include Shields (1995), Swenson (1995) and Mc Gowan and Klammer (1997).
The database used in this research consisted of potential ABCM sites. A survey was sent to each potential ABCM site identified.
The authors received 166 useable responses from 132 separate companies. All respondents were currently using ABCM when they completed the surveys. The authors tested for non-response bias by contacting 200 non-respondents.
Seventy-eight percent (130 out of 166) of the site surveys were filled out by finance/accounting personnel. The authors assume finance/accounting personnel are able to observe decision makers and thus make informed responses to questions. The cover letter to the survey asked respondents to "please select and use inputs from other people in the organization".
Field visits to 15 ABCM sites were also made as part of the research to gain additional insight into how to interpret the company responses and how they measured "ABCM success," and to develop individual site perspectives on why certain variables might be hypothesized to be determinants of ABCM success. Companies visited reported little systematic effort to either (1) quantify the expected benefits from using ABCM prior to implementation, or (2) conduct ex post audits to examine what benefits were actually achieved. All acknowledged that ABCM had to pass a cost-benefit test, but none of the 15 invested sizable resources into obtaining reliable measures of the ABCM benefits either ex ante or ex post.
1.0 An a priori classification (first approach)
1.1 Shields (1995) Model
The Shields (1995) study used the measurement of SM-$1 and SM-ME as a proxy for ABCM success. Each of the five significant independent variables reported in that study as an ABCM success determinant. These five variables were: (1.) Top management support; (2.) Implementation training; 3. Link to performance evaluation/compensation; (4.) Link to quality initiatives; and (5.) Adequacy of resources
First, using the responses from the survey, the authors identified individual questions that best aligned to those used by Shields (1995).
Second, the authors extended Shields (1995) by considering alternative measures of dependent and independent variables used by Shields (1995). The dependent variables are:
SM-ME: based on the six individual questions.
SM-U: based on the 31 individual questions.
SM-DA: based on the 11 individual questions. These were the individual questions that were aligned with those used by Shields (1995): (1.) Top management support; (2.) Implementation training; (3.) Link to performance evaluation/compensation; (4.) Link to quality initiatives; and (5.) Adequacy of resources
1.1.1 Findings from extended Shields (1995):
The SM-U success measure-based regression has an adjusted R2 of 0.64 which is marginally higher than the adjusted R2 of 0.61 when all four success measures are combined into a composite.
When all four success measures are combined into a composite, each of the five independent variables is statistically significant (at the .05 level) in the same direction as reported by Shields (1995).
The ABCM sites examined in this study differ greatly in both their number of primary applications and their time in use. The results for adding both of these as independent variables to the current existing five independent variables increases the R2 from 0.61 to 0.71.
One explanation for the significance of the number of primary applications variable is that it is only successful implementations that increase the number of applications over time.
Alternative explanations also exist for the significance of average years of primary applications. One explanation is that benefits do not appear until after a sizable time-in-use has elapsed (either because of implementation problems or because the effects of changed decisions take time to appear). To probe into these alternative explanations, the authors add "increase in number of applications" and a dummy variable approach (0 if the site average time-in-use is 1 year or less and 1 otherwise) as independent variables to the current model of seven independent variables.
Results show that the addition of the increase in primary application variable and the dummy variable for applications with average use <= year does not add to the explanatory power of the model. These results are consistent with the hypotheses that (1) more ABCM applications lead to more success, and (2) the longer the time period ABCM is used the higher the benefits and that these benefits start appearing in a relatively short time period after use commences.
1.2 Shields and Young (1989) Model
Shields and Young (1989) present a "Seven Cs model" for implementing cost management systems. The seven factors proposed as success determinants in that model are-culture, controls, champion, change process, commitment, compensation and continuous education. Using the database from the survey, independent variables capturing each of the seven factors were developed.
The results were similar to those found in extending the Shields 1995 model. When all four success measures are combined into a composite as the dependent variable, four of the seven Cs as independent variables are statistically significant (at the .05 level). Expanding the success measures beyond SM-$I or SM-ME to include SM-U and SM-DA increases the explanatory power of a model of ABCM success determinants. Similar to prior findings discussed in this paper, the two additional variables-number of primary applications and average time in use-are both statistically significant and have positive coefficients when added to the overall model. The adjusted R2 is 71 for the broadest based success measure (based on 55 questions covering four categories).
2.0 Factor-Analysis Based ABCM Success Measures (second approach)
Four factors were identified:
Factor 1: Decision use
Factor 2: Product/customer applications
Factor 3: Function/manager applications
Factor 4: Manager group success perceptions
There is considerable overlap between several of the above factors and the a priori groupings examined previously in this paper. Pairwise correlations between each of the four success factors and each of the four a priori success measures are:
The highest correlations (both 0.89) are between the decision use factor (F1) and the decision use measure (SM-U) and between the function/manager applications factor (F3) and the decision use measure (SM-U).
Next, the authors examine the effect of using these factor-based success measures in the two previously examined models of ABCM success determinants-i.e., Shields (1995) and Shields and Young (1989).
2.1 Findings: The study reports the adjusted R2 when each of the four factor-based success measures is separately used as the dependent variable and when a composite of the four factors (equally weighted) is the dependent variable. The results reinforce those previously reported in this paper. The adjusted R2 when the broader-based composite success measure is used as the dependent variable is .70.
Cooper, R. and R. S. Kaplan. 1992. Activity-based systems: Measuring the costs of resource usage. Accounting Horizons (September): 1-13. (Summary).
Foster, G. and S. M. Young. 1997. Frontiers of management accounting research. Journal of Management Accounting Research (9): 63-77. (Summary).
Keys, D. E. 1994. Tracing costs in the three stages of activity-based management. Journal of Cost Management (Winter): 30-37. (Summary).
Keys, D. E. and R. J. Lefevre. 1995. Departmental activity-based management. Management Accounting (January): 27-30. (Summary).
Martin, J. R. Not dated. Activity based management models. Management And Accounting Web. https://maaw.info/ABMModels.htm
Martin, J. R. Not dated. The CAM-I conceptual design. Chapters 1-3. Management And Accounting Web. https://maaw.info/CAM-ISummary.htm
Martin, J. R. Not dated. The CAM-I conceptual design. Chapter 6. Management And Accounting Web. https://maaw.info/CAM-ISummaryChapter6.htm
Thorne, H. and G. Gurd. 1999. Activity-based costing: Improved product costing or activity management? Advances in Management Accounting (8): 173-194. (Summary).