Assessing the managerial objectives of CPA firm partners.
The partner/manager responds to this complex environment facing the firm by balancing the firm's various objectives. These objectives include: (1) short-term profitability; (2) client service, which translates into longer-term profitability and increased competitiveness in attracting and maintaining clients; (3) declining work that cannot be handled, which affects the firm's risk of litigation; (4) managing the firm's labor force, which impacts the firm's cost efficiency; and (5) maintaining the skills and capacity of the firm through the training of personnel, which affects competitiveness and litigation risk.
Few studies have addressed the role of the partner as the manager of the firm. One study (Baker, 1977) investigated the overall management strategy of a single large CPA firm ming a participant observation methodology. Baker found that the management strategy of a CPA firm includes three components: "doing" -- performing client services for a fee; "representing" -- developing relationships with outside parties other than clients; and "being" -- developing the firm's image. Baker's research argued for a need to balance the strategic elements of client services, outside relationships, and firm image, but did not provide a model on which to base further research. Maister (1982) discussed the economics of a generic professional service firm, characterized by a multi-level staff structure, work performed by project teams, and fee revenue based on billable time (all of which describe a CPA firm). Such a firm interacts with the market for professional labor as well as the market for the firm's services. In dealing with the market for professional labor, the managers of the firm are concerned with hiring, promotion policy, growth planning, and turnover. In dealing with the market for the firm's services, concerns include the type of services to be offered and the team structure necessary to deliver the services. Management's role is one of balancing the firm's economic/organizational structure with the two markets in which it operates.
The professional literature on practice management emphasizes the notion of balancing multiple objectives. The AICPA's Management of an Accounting Practice Handbook (1985) discusses client service, financial administration, and personnel management as key functions in the management of the CPA firm. Earlier editions (MacNeill, 1962) began explicitly with chapters on fees, staff personnel (including hiring, training, and utilization), and client relations. Similarly, Miller (1987) cites the importance of the interaction of fees, a quality stafF, and timely client service in the management of a practice.
The academic literature has indirectly examined the behavior of CPA partner/managers via studies such as those on price premiums charged by national CPA firms (Palmrose, 1986; Rubin, 1988; Francis and Simon, 1987), initial price discounting ("low-bailing") to achieve a client (DeAngelo, 1981; simon and Francis, Schatzberg 1990, 1994), and the role of the firm's production cost function in setting audit fees (Stein et al., 1994: O'Keefe et al., 1994; Copley et al., 1995; Gigler and Penno, 1995). These studies involve consideration of short-term versus longer-term profit tradeoffs, the value of client retention, and the role of production costs in firm competitiveness. Most of these studies have approached the economic behavior of CPA firms from an external perspective, examining the interaction between auditors and auditees using publicly available data and questionnaire results. The Schatzberg (1990, 1994) studies took an internal perspective, using students as surrogates for firm decision makers in several laboratory market experiments. The study in this paper also takes an internal perspective, but uses 40 partners from national and local CPA firms in a multiple-criteria decision exercise.
By studying external indicators, such as audit prices, prior studies have indirectly addressed the managerial behavior of CPA partners. Our study directly examines how partners balance various objectives in a complex audit environment. Although auditing is only one of three subareas in a typical public accounting firm (tax and consulting being the others), the environment facing each subarea is more similar than different. We examine the specific solutions of 40 audit partners to a seven objective model involving considerations of profitability, client service, and personnel management. We find the behavior of partners consistent with the position that profitability, client service, and personnel management objectives are all important in managing the firm, and that the choices made in our experiment reflect a balancing of these factors. These results provide descriptive evidence on the behavior of CPA firm partners in a managerial Context.
Our experiment also offers insight on another dimension of accounting research. Various studies have explored differences in behavior between national/international CPA firms and local/regional firms (hereafter simply called "national" and "local" firms). Research on the quality of audit services (Francis and Wilson, 1988), the price premiums for Big Eight firms (Francis and Simon, 1987), and direct solicitation of clients by CPA firms (Chaney et al., 1995) lead one to expect differences in objective choices between partners from national firms and partners from local firms. We find significant differences between the objectives emphasized by national firm partners and those emphasized by local firm partners.(1) While we have not tested the relation of differences in managerial objectives to different audit decisions, our findings suggest that the different managerial objectives could be a factor in explaining different behavior reported in previous studies that have investigated national versus local CPA firms.
Multiple-Objective linear Programming as a Tool
Multiple-criteria decision analysis provides a set of techniques to study decisions that are jointly influenced by several objectives. Numerous multiple-criteria decision techniques exist (Lin, 1987). In developing our model, We chose a multiple-objective linear programming technique (hereafter MOLP), where the participants do not need to specify their objective Weights in advance. Rather, from observing their choices in a practice management exercise, their implicit weighting of each objective can be determined. The participants are able to focus on the task at hand, the audit workload scheduling problem, without being distracted by or even aware of our interest in observing how much weight is placed on each objective.
In a MOLP model, several objective functions are established and are appropriately linked to linear constraints. The MOLP model does not have an "optimal" solution in the same sense as does a unidimensional linear programming (LP) model. However, a specific solution to the MOLP model reveals explicitly how all objectives are affected. This provides the decision maker with a useful device to observe the tradeoffs among the objectives as alternative solutions are proposed. This is not possible in the typical unidimensional LP model. The use of the MOLP model described in this article allows the decision maker to concentrate on the set of outcomes and not be burdened by the details underlying each solution outcome. That is, underlying any specific set of outcomes are the operating details necessary to implement the given solution. Thus the decision maker can concentrate on the potential payoffs (multiple objective function values) and make tradeoffs.
Balachandran and Steuer (1982) developed a single-period MOLP model of the audit staff planning process. In a prior paper, the authors (Gardner et al., 1990) extended the Balachandran-Steuer model by developing and testing a multi-period version, which serves as the basis for the experiment reported here.
Our MOLP model(2): consisted of seven objectives relating to profit, client service, and personnel management, specifically:
-- maximize current period profit
-- minimize late completion of work
-- minimize amount of work declined
-- minimize hiring of additional staff
-- minimize dismissals of staff
-- minimize underutilization (people doing lower level work) and nonutilization (idle time) of staff
-- minimize shortfall in meeting professional development targets -- that is, the number of staff hours scheduled to be devoted to training.
These objectives are consistent with the elements of economic behavior and with the practice management literature cited earlier. While it is possible that a Critical objective is omitted, nothing in the prior literature or in the reactions of the participants in our experiment suggest an untested objective.
The model was multi-period, coveting a four-quarter planning horizon which included a "busy season." The finn is assumed to operate under a variety of constraints that limit managerial flexibility along various dimensions. These constraints, along with particular limitations used in our application of the model, are summarized below.
Workload was measured in hours and characterized along two dimensions: the staff level scheduled to perform the work, and the time period in which the work is scheduled to be performed. In our application, six staff levels existed -- work could only be done by staff of the scheduled level or one level above; other substitutions were not allowed. Some workload rearrangement is permitted, enabling work to be done earlier or later than the scheduled period. In our application, the interim work limitation allowed up to 20 percent of a given quarter's workload to be performed in the preceding quarter; alternatively, work could be performed up to two quarters late. There was no limit on the quantity of late work, but minimization of lateness was established as one of the objectives.
Staff numbers at each level are established as a starting point. Additional staff may be hired or staff may be dismissed each period. Various limitations on staff changes may be imposed. In our application, staff changes were permitted only at the manager level (level 3) and below; the number of partners was considered fixed over the planning horizon. Limitations on numbers hired/fired during a time period could be imposed, and/or a hiring/firing cost could be reflected in the profit objective. We used the latter, and also established staff augmentation and staff reduction as objectives to be minimized. While it is tempting to constrain staff changes to integer values, this complicates the solution process. Non-integer outcomes may be interpreted in equivalent-unit terms, that is, as occurring during the period.
Staff Utilization is based on the amount of regular worktime available each period. Regular worktime reflects allowances for personal time (e.g., illness) and for nonbillable duties (e.g., administration, recruiting, practice development). Regular worktime may vary by staff level, reflecting the fact that partners have considerably more nonbillable time than juniors. Overtime work is allowed, and is subject to an upper limit, which may be different for different staff levels. Staff is underutilized if personnel are assigned duties below their normal work level, and staff ia unutilized if Idle time exists; we included minimization of these as an objective.
Supervisory Requirements incorporate the requirements of a CPA firm for certain relationships among the number of personnel at each level, reflecting the usual structure of audit work and the need for review and supervision. Minimum ratios of personnel at a given level to the total number of personnel at all lower levels were established.
Professional Development reflects the need for staff training, as required by professional rules and firm needs; these requirements may vary by staff level.
In total, 154 constraints are reflected in the model. In the development stage, the model was run by practitioners to ensure that it was easy to use and that reasonable outcomes resulted.
A case problem was developed to provide a specific context in which the participants would make decisions. A fictional CPA firm was described, with initial numbers of personnel at each of six levels ranging from junior staff through partners. To establish realistic estimates of the various model parameters, we interviewed two former managers of CPA firms who had substantial familiarity with firm operations, including staff characteristics, cost factors, and billing rates. The projected audit workload was described in terms of quarterly hours required for each of the six staff levels. Also specified were available hours at each staff level, pay rates, billing rates, professional development targets, and hiring and firing costs. Limitations were provided regarding overtime, interim work, and supervisory requirements. Specific case data are described in the Appendix.
Partners from national and local CPA firms served as the participants in the experiment. Forty partners participated, 20 from national firms and 20 from local firms. All were either audit partners or managing partners. We excluded partners whose primary responsibility, was tax or consulting services. Experience varied from a newly appointed partner to one with 23 years of partner service. Since individuals typically have ten to fifteen years experience before achieving partnership, the total CPA-firm experience of our participants was in the range of ten to 35 years. The use of highly experienced participants lends particular credibility to the outcome of the exercise.
The Experimental Process
The partners were given, in advance, the case problem and descriptions of the objectives and the interactive process to aid them in understanding the experimental environment. They were advised that there was no "right" solution, as each person would assign different degrees of importance to the various objectives.
Each partner spent about 45 minutes to an hour on campus performing the experiment. A review of the process and a trial run for familiarization purposes preceded the "real" run of the experiment. The authors handled the technical requirements of loading and running the model.
At each iteration, the participant saw several alternative solutions on the screen. Each solution consisted of a particular outcome for each of the seven objective. For example, at the first iteration of the first run, each partner saw the seven alternative solutions presented in Table 1. The participant viewed the seven sets of choices on the screen and, after deliberation, chose the most preferred solution set. The computer used this selection to calculate the next set of alternatives to appear on the screen (this calculation process took a minute or less). In the subsequent iterations, the alternatives differed among individuals, based on their choices at earlier stages. Further, at each iteration beyond the first, an additional "alternative zero" existed, which was the solution selected at the preceding stage. Thus, the partner could either retain the previously chosen solution or move to a new solution. Seven iterations were allowed for the participant to reach the final solution. The partners deliberated on their choices, and appeared to understand the information set and the interrelationships.
TABLE 1 Initial Solution Presented to Partners Objectives Number Hours of % Work of Staff Profit($) Late Work Declined Added(*) 1. 3,152,696 6,262 0.0 0.0 2. 3,003,468 0 3.6 2.1 3. 3,232,527 6,457 0.0 0.0 4. 3,047,780 655 3.9 0.0 5. 2,966,412 0 0.0 11.6 6. 3,281,070 13,878 0.0 0.0 7. 3,220,764 0 0.0 3.0 Objectives Number Under Shortfall in of Staff Utilized Prof. Devel. Profit($) Reduced(*) (Hours) Pr(Hours) 1. 3,152,696 0.0 0 0 2. 3,003,468 0.0 1,198 0 3. 3,232,527 0.0 0 4,390 4. 3,047,780 0.5 0 0 5. 2,966,412 0.0 13,373 0 6. 3,281,070 0.0 4,846 4,304 7. 3,220,764 0.0 4,378 0 (*) Full-time equivalents; for example, hiring 0.5 individuals is interpreted as hiring one individual midway is the time period.
The process produced alternatives by calculating the implicit weights assigned by each participant to each objective. The Weight assigned to each objective measured the relative importance placed on that objective by the partners. The model began with weight ranges of width 1.0 for each objective, and successively narrowed the range as decisions were made at each iteration. After seven iterations, the weight range on each objective was narrowed to width .1429 (the entire range must fall within zero to one). An objective to which respondents reacted strongly results in a high range (e,g., 5153 to .6582) while an objective that respondents largely ignored results In a low range (e.g., 0 to .1429). Strong reaction means that the decision maker consistently picks a solution that has a high (or low) value for a given variable, and thus does not seem to be indifferent as to the value of that variable. The weight range for such an objective would be increased. If, for another variable, solutions are chosen with no consistent pattern, the decision maker appears to not view that as very important, or to not care what its outcome is, and the weight range would be decreased.
The final weight placed On each objective by a participant was a value, calculated by the model, within the .1429-width range (the total weights assigned to the seven objectives must sum to one). While the participants were able to observe the objective values at each iteration, they were never able to observe the weights or the weight ranges. The average weights ranged from a high of 20.29 percent for minimizing work declined to a low of 6.85 percent for minimizing staff underutilization. Each average weight was found to be significantly (at the .05 level) different from zero, suggesting that each objective was considered important by the participants. While examination of the pattern of weights suggests that balancing of objectives occurs, further analysis of the weights has limitations. The model reports a weight range for each objective for each participant. The specific weights reported for each participant represent a sample (chosen by the model) from the final range of weights. Taking a different sample from the same set of weight ranges could produce slightly different specific weights. For this reason, we do not use the calculated weights, but rather the standardized objective values for our tests.
As our goal h to test for the presence of multiple objectives in the partners' decisions, we defined three unidimensional decision strategies, described below, to serve as benchmarks. The authors ran the model (using the same starting point as used by the participants) following each of the unidimensional strategies. These provided the outcomes that a decision maker would achieve under this particular decision emphasis. The following decision rules were applied in implementing each unidimensional strategy.
The profit-maximization strategy was defined as always choosing the alternative with the highest current period profit. Since all behavior may ultimately have profit implications, the first strategy is maximizing profit in the short term, but for convenience we label it simply as "profit maximization." Certainly, the client-service strategy may be viewed as seeking long-term profits through customer satisfaction, and the personnel-management strategy may be viewed as focusing on intermediate-term profits through control of production costs. In case of ties (i.e., equal profit levels two or more alternatives), the alternative with the lower amount of late work was chosen.
The client-service strategy was defined as having a first priority of timely completion of all work. In the case of ties (typically at zero lateness), the next goal was to minimize work declined. If a further tiebreaker was needed, the highest profit was chosen.
The personnel-management strategy was defined as having a first objective of meeting professional development targets, followed in turn by minimization of layoffs, minimization of underutilization, and minimization of hiring during the period. If a further tiebreaker was needed, the highest profit alternative was chosen.
We then compared the results generated by these strategies to the results generated by the 40 partners. We examined whether the partners' results coincided with any of the output patterns produced by our three benchmark strategies. We further examined the behavior of the national firm partners versus the local firm partners, relative to the benchmark strategies and to each other.
Partner Outcomes and Standardization
Table 2 displays the mean, standard deviation, and range of the outcomes for the model's seven objectives (profit, work completed late, etc.) as chosen in the final iteration by the partners. To use these values in subsequent analyses, we standardize them by expressing each as a location in the range of objective values encountered by the 40 partners. That is, we set the lowest encountered value equal to zero and the highest encountered value equal to one, and express a given objective value as a decimal relative to the standardized range of one. For each of the seven objectives, the range is defined by the highest and lowest values ever encountered by the participants throughout the experiment, at any iteration. Keeney and Raiffa (1976) argue that using a measure relative to the range actually encountered by the subjects is more efficient than using a measure relative to the range of values theoretically achievable, Further, some i theoretical outcomes, such as declining all work, are essentially unachievable by running the model, as they run counter to the model's logic (of maximizing profit while minimizing all other constraints). To approximate the range of achievable values, we ran 120 simulations of the process (i.e., 120 simulated participants times seven iterations per run, or a total of 840 iterations), making random selections at each iteration. For five of the seven variables, the simulated range differed from the participants' encountered range by less than five percent; the largest difference was 11.5 Percent. Thus, the range encountered by the participants appears to be a good approximation of the achievable range.
TABLE 2 Final Objective Values Chosen Standar- Standard dized Mean Deviation Range Value Profit ($) $3,115,992 74,230.00 $2,965,540 .7087 to $3,247,761 Work completed late (hours) 2,676.46 2,269.46 0 to 7,166.75 .1594 Work declined (percent) 1.35 1.75 0 to 4.38 .2454 Staff added (people) 3.04 3.46 0 to 11.67 .2055 Staff reduced (people) 0.88 2.72 0 to 10.92 .0412 Staff underutilized (hours) 2,294 2,230 0 to 8,930 .0964 Professional development shortage (hours) 83.55 388.46 0 to 4,195.79 .0154
The final column of Table 2 displays the average standardized objective values chosen by the 40 partners. The standardized values show that the Farmers chose a relatively high level of profit (on average, an amount that represented over 70 percent of the standardized maximum value) and relatively loW" levels of the other six objectives (ranging from about two percent to about 25 percent of the standardized maximum). These results are consistent with the exPected maximization of profit and the expected minimization of the other six objectives.
The standardized objective values produced under the three benchmark strategies, along with the standardized mean results achieved by our 40 Farmers, are shown in Table 3. The profit-maximization strategy (shown in the first line of Table 3) resulted in a profit figure (1.102 of the encountered range) that exceeds by ten percent the highest profit figure ever encountered by any of the participants (recall that the highest encountered value was standardized to 1.0). This implies that not even one of the 40 partners pursued a strategy to drive the profit outcome as high as possible in the current period, and hence the theoretical upper bound of the profit range was never encountered at any stage of their selections. The profit-maximization strategy also resulted in a high degree of work completed late (.6 of the encountered range) and a significant shortfall in achieving professional development targets (.8); no current financial penalties attached to either in the model. No work was declined, and no staff were laid off (the costs of dismissal are fairly high).
TABLE 3 Relative Objective Values: Benchmark Strategies vs. Participants Objective Work Add Profit Lateness Declined Staff (Max.) (Min.) (Min.) (Min.) Profit max. 1.102 0.600 0.00 0.154 Client service 0.560 0.000 0.00 0.785 Personnel mgt. 0.876 0.375 0.00 0.00 Participants 0.709 0.159 0.245 0.206 Objective Reduce Under- P.D. Staff Utilize Shortage (Min.) (Min.) (Min.) Profit max. 0.00 0.184 0.804 Client service 0.00 0.375 0.000 Personnel mgt. 0.00 0.000 0.000 Participants 0.041 0.096 0.015 Note: table values are expressed as a percentage of the maximum value encountered by the participants. For each objective, the range of values encountered is standardized so that the maximum value encountered (see Table 1) equals 1.0 and the minimum value equals zero. A figure greater than 1.0 (i.e., the profit outcome in the profit maximization strategy) means that the value chosen exceeded the highest value ever encountered by the subjects. None apparently pursued a strategy to drive outcome as high as possible, and hence encountered the theoretical upper bound of the profit range.
The client-service strategy (second line of Table 3) resulted in all work being accepted and all work being completed on time, though at some cost in profit (only .56 of the encountered range is achieved). High levels of hiring (.785) and corresponding Underutilization of staff (.375) were necessary in order to provide the desired client service.
The personnel-management strategy (third line of Table 3) resulted in all personnel objectives being fully met (minimized) -- no professional development deficiencies, no staff changes, and no underutilization. Profit was fairly high (.876 of the range) as a result of minimizing production costs. Client service suffered, as late work was more common (.375 of the range).
The relative objective values generated by the 40 partners are shown on the fourth line of Table 3. Some degree of lateness was tolerated (.159) and some work was declined (.245). Some "inefficiencies" in production cost were accepted, as hirings (.206) and layoffs (.041) occurred along with some underutilization or idleness of staff (.096). Profit was moderate (.709 of the encountered range), falling below that of the current profit maximization or personnel cost minimization strategies, but above the level of the client-service-at-any-cost approach.
Comparison of Partners to Benchmark Outcomes
To evaluate whether the participant group seemed to follow a profit-maximization, client-service, or personnel-management strategy, the objective values for each benchmark strategy were compared to the objective values generated by the partners. As shown in Table 4, for each strategy, at least six of the seven objective values are significantly different (at least at the .05 level) from the partners values.
TABLE 4 Summary Statistics of Partner Objective Results Compared to the Objective Results of Three Benchmark Strategies A. Partner Objective Results Compared to Profit-maximization Strategy Objective Work Add Profit Lateness Declined Staff ($000) (hours) (%) (people) Difference -169.81 -7313.54 1.34 0.77 t-statistic -14.71(*) -21.25(*) 5.09(*) 1.66 Objective Reduce Under- P.D. Staff Utilize Shortage (people) (hours) (hours) Difference 0.88 -20.70 -4280.54 t-statistic 3.47(*) -65.60(*) -51.89(*) B. Partner Objective Results Compared to Client-service Strategy Objective Work Add Profit Lateness Declined Staff ($000) (hours) (%) (people) Difference 149.98 2691.54 1.34 -8.56 t-statistic 12.99(*) 7.82(*) 5.09(*) -18.52(*) Objective Reduce Under- P.D. Staff Utilize Shortage (people) (hours) (hours) Difference 0.88 -66.40 83.55 t-statistic 3.47(*) -210.00(*) 1.01 C. Partner Objective Results Compared to Personnel-management Strategy Objective Work Add Profit Lateness Declined Staff ($000) (hours) (%) (people) Difference -36.31 -3570.71 1.34 3.04 t-statistic -3.14(*) -10.37(*) 5.09(*) 6.57(*) Objective Reduce Under- P.D. Staff Utilize Shortage (people) (hours) (hours) Difference 0.88 2.90 83.55 t-statistic 3.47(*) 72.60(*) 1.01 (*) Significant at the 5% level for a two-tailed test.
These results indicate that the partner group did not follow any of our three benchmark strategies: short-term profit maximization, client service or personnel management. As each of these emphasized a single dimension of firm performance, the results imply that partners balanced considerations of current profit, client service; and personnel management in managing the firm. Thus, the behavior of the partners is consistent with the position that multiple objectives impact their managerial decisions.
Comparison of National to Local Partners
As discussed earlier, some research identifies differences in the behavior of national firms versus local firms. Half of our 40 participants were national firm Partners, and half were local firm partners,(3) In comparing each of these two subgroups to the results for the benchmark strategies, we achieved the same results as for the entire group. In each comparison, at least six of the seven objectives show significant (at least .05) differences between the values chosen by the subgroup and the values produced by each of the three benchmark strategies. Thus, neither the national firm subgroup nor the local firm subgroup appears to follow a pure current profit-maximization, client-service, or personnel-management strategy.
Not finding differences in comparing homogeneous subgroups to certain benchmarks does not imply a lack of significant differences between the subgroups themselves. A direct comparison of the objective values chosen by the national firm partners to those chosen by the local firm partners was made. As shown in Table 5, there were significant differences on the profit and lateness dimensions.(4) Local firm partners chose higher current profit levels than did the national partners, while allowing more work to be completed late. The results suggest that local partners may be more motivated by short-term profitability, or at least are less concerned with timeliness of completion of audit work relative to national partners. Completing work on time, especially during the busy season, is costly, thus yielding lower current profits for the national firm participants. National firm partners may be more sensitive to performing client services on time, as many of their clients may be SEC registrants facing a ninety-day filing deadline. Local firms, with few if any SEC clients, do not face this element of time pressure. In addition, national firms may trade off profitability in the audit function for profits in other services (particularly consulting) offered to clients. Local firms, often not having significant consulting capabilities, may not be in a position to allow audit services to be a "loss leader," thereby resulting in our outcome of higher audit services profit maximization. The differences in behavior observed between national and local firm partners appears consistent with differences in the nature of their practices.
TABLE 5 Summary Statistics of Differences Between Objective Results Local Partners Compared to National Partners Objective Work Add Profit Lateness Declined Staff ($000) (hours) (%) (people) Difference 18.09 621.61 -0.18 -0.48 t-statistic 3.44(*) 3.98(*) -1.52 -2.28 Objective Reduce Under- P.D. Staff Utilize Shortage (people) (hours) (hours) Difference 0.23 -0.158 -167.10 t-statistic 1.96 -1.09 -4.47(*) (*) Significant at the 5% level for a two-tailed test.
This paper reported on an experiment designed to measure the managerial objectives of CPA firm partners in a realistic situation where current profit maximization could be balanced against client service and personnel management considerations. Current client service is not only an objective in itself, but also has implications for the firm's future profitability. Personnel management can also impact future profitability through the maintenance and training of staff. The use of 40 partners from national and local CPA firms as the participant lends credence to the conclusions.
A previously developed interactive multiple objective linear programming model involving a workload scheduling problem served as the basis for this experiment. Seven objectives included maximizing profit, minimizing lateness of work, work declined, staff additions, staff reductions, underutilization of staff, and professional development deficiencies. Each partner independently selected from alternative combinations of objective values for a case problem in an interactive process which refined the available alternatives based on the individual's prior decisions. Results were then calculated for certain predefined strategies -- a current profit-maximization strategy, a client-service strategy, and a personnel-management strategy -- to serve as a basis of comparison.
The results of the experiment produced two major findings. First, partner behavior appears to be consistent with the position that multiple objectives impact their managerial decisions. The partner group did not follow one of the benchmark strategies; the objective values produced under each of the three predefined strategies differed significantly from the objective values produced by the group of 40 partners. Second, there appear to be differences between national and local firm partners as to managerial objectives. A direct comparison of the results for the national versus local partners suggested that the latter were somewhat more current-profit motivated and/or less motivated by on-time client service.
There are, of court, limitations to the study. The somewhat limited sample size -- 20 national firm partners and 20 local firm partners -- and the limited geographic makeup of the sample may limit the external validity of the conclusions. The decision exercise, while as complex and realistic as we could make it, was nonetheless artificial. Participants appeared to take the exercise seriously, but we cannot be assured that they would behave the same in an actual decision context.
Our findings have implications for future research. In modeling the behavior of CPA firms or CPA firm partners, a profit objective alone is not sufficient. Client service and personnel management objectives should be considered as well. From a methodology viewpoint, the use of multiple objective linear programming coupled with a decision case may be a useful approach to other behavioral studies. To our knowledge, this approach has not been widely used as a behavioral research tool.
Our findings have managerial implications as well. We provide some insight into the managerial behavior of professional service firms. Understanding the managerial objectives of CPA firms may be useful to various parties who interact with such firms. By understanding what motivates the decisions of the owner-manager of a CPA firm, clients, regulators, and other parties who deal with such firms may be better able to manage their interactions with CPA firms. On another dimension, participants suggested that an exercise of the type used here might be a useful training and evaluation tool within CPA firms. As individuals progress toward the partner stage, participating in such an exercise would enable an assessment of their likely behavior in a managerial role.
As the environment of the CPA firm grows more complex, issues concerning the management of the firm will attract increased attention. This study adds some insight into the managerial behavior of CPA firm partners. The growing importance of service firms makes this a fruitful area for further research.
Personnel may be expanded beyond the initial number, at a hiring cost of $20,000 for managers, $10,000 for seniors, $5,000 for semiseniors, and $2,000 for juniors. Only levels 3-6 can be hired; new partners cannot be added. Supervisory requirements specify that there be at least one engagement partner for every four managers, at least one manager for every three seniors, at least one senior for every semisenior, and at least one semisenior for every two juniors. Oversupervision at one level may be offset at higher levels, provided the cumulative requirements are met. At some point, to hire at one level it may also be necessary, to hire at upper levels. Personnel may also be dismissed during the year, at a cost of $60,000 to dismiss a manager, $30,000 to dismiss a senior, $15,000 to dismiss a semisenior, and $6,000 to dismiss a junior.
Appendix: Case Data The audit firm has six levels of personnel, as follows: Initial Available PD Level Description Number Hours Required 1 Supervising Partner 1 800 40 2 Engagement Partner 3 900 40 3 Manager 9 1400 75 4 Senior 18 1800 100 5 Semisenior 18 1800 80 6 Junior 27 1800 80 Billing Annual Level Description Rate Pay 1 Supervising Partner $225 $125,000 2 Engagement Partner 150 85,000 3 Manager 110 40,000 4 Senior 80 27,000 5 Semisenior 60 22,000 6 Junior 50 18,000
The available hours shown above reflect the chargeable time likely to be available after allowing for vacations, sick time, other idle time, and other nonbillable time. Overtime may be worked, up to a maximum of 50 percent of available hours for levels 1 and 2; 20 percent for levels 3, 4, and 5; and 10 percent for level 6. Levels 4 through 6 are paid for overtime at 150% of their regular rate; levels 1 through 3 are not paid for overtime.
The required professional development time each year represents the number of hours to be spent either in taking courses or in teaching them. It is assumed that seniors and managers often conduct in-house training for staff.
The billing rates shown are the nominal rates. Frequently, the effective rate is less, because extra hours are incurred that cannot be billed, or because the job is essentially a fixed-price job. In addition, there are some variable overhead costs (fringe benefits, suppliers, travel, etc.). To allow for both price concessions and variable costs, we assume that the effective net billing rates are 60% of the nominal rates.
The projected workload, in hours, for the four quarters of the coming year is as follows:
Level 1 Level 2 Level 3 Level 4 Quarter 1 132 446 2079 5346 Quarter 2 352 1188 5544 14256 Quarter 3 220 743 3465 8910 Quarter 4 176 594 2772 7128 Total 880 2971 13860 35640 Level 5 Level 6 Total Quarter 1 5346 8,019 21368 Quarter 2 14256 21,384 56980 Quarter 3 8910 13,365 35613 Quarter 4 7128 10,692 28490 Total 35640 53,460 142451 Normal work hours are assumed to be divided equally over the year, making the available hours at each level as follows: Level 1 200 hours/qtr 800 hours/year Level 2 675 2700 Level 3 3150 12600 Level 4 8100 32400 Level 5 8100 32400 Level 6 12150 48600 Total 32375 129500 Overall workload is 142,4517 hours and normal work hours available from present staff is 129,500.
(*) We appreciate the helpful comments of Robert L. Hagerman and workshop participants at the University at Buffalo and the University of Rhode bland.
(1) There is a question of whether these subgroups are sufficiently homogeneous; we address this issue later.
(2) The mathematical formulation of the model may be obtained from the authors or by consulting Gardner et al. (1990).
(3) The 20 rational firm participants came from nine different national firms and from offices in three cities, so there should be little predominant firm bias or office bias in this subgroup. The 20 local partners came from 13 different firms, all in the same city. The diversity or firms from which our
(4) There was also a significant difference for the professional development objective. No conclusions are drawn for this variable as only one of the forty partners chose a non-zero level for the professional development deficiency. The level chosen, however, was quite large.
American Institute of Certified Public Accountants. 1985. Management of an Accounting Practice Handbook New York: American Institute of Certified Public Accountants.
Baker, C.R. 1977. "Management Strategy in a Large CPA Firm." The Accounting Review (July): 576-586.
Balachandran, K.R., and R.E. Steuer. 1982. "An Interactive Model for the CPA Audit Staff Planning Problem with Multiple Objectives." The Accounting Review (January): 125-140.
Chaney, P.K., D.C. Jeter, and P.E. Shaw. 1995. "Direct Solicitation and Large Audit Firm Dominance in the Audit Market." Auditing: A Journal of Practice & (Spring): 19-33.
Copley, P.A., J.J. Gaver, and K.M. Gaver. 1995. "Simultaneous Estimation of the Supply and Demand of Differentiated Audits: Evidence from the Municipal Market." Journal of Accounting Research (Spring): 137-155.
DeAngelo, L.E. 1981. "Auditor Independence, Low Balling and Disclosure Regulation." Journal of Accounting and Economics (August): 113-127.
Francis, J., and D. Simon. 1987. "A Test of Audit Pricing in the Small-Client Segment of the U.S. Audit Market." The Accounting Review (January): 145-157.
--, and E. Wilson. 1988. "Auditor Changes: A Joint Test of Theories Relating to Agency Costs and Auditor Differentiation." The Accounting Review (October): 663-682.
Gardner, J.C., R.J. Huefner, and V, Lotfi. 1990. "A Multiperiod Audit Staff Planning Model Using Multiple Objectives: Development and Evaluation." Decision Sciences (Winter): 154-170.
Gigler, F., and M. Penno. 1995. "Imperfect Competition in Audit Markets and its Effect on the Demand for Audit-Related Services." The Accounting Review (April): 317-336.
Keeney, R.L., and H. Raiffa. 1976. Decisions with Multiple Objection: Preferences and Value Tradeoffs. John Wiley and Sons.
Lin, T.W. 1987. "Multiple Criteria Decision Making in Auditing." The Auditors Report (Winter): 1, 4-6.
MacNeill, J.H. (ed.) 1962. Accounting Practice Management Handbook. New York: American Institute of Certified Public Accountants.
Maister, D.H. 1982. "Balancing the Professional Firm," Sloan Management Review (Fall): 15-29.
Miller, R.E. Jr. 1987. How to Compete Successfully in a Public Accounting Practice. Englewood Cliffs, NJ: Prentice-Hall, Inc.
O'Keefe, T.B., D.A. Simunic, and M.T. Stein. 1994. "The Production of Audit Services: Evidence from a Major Public Accounting Firm." Journal of Accounting Reseaarch (Autumn): 241-261.
Palmrose, Z. 1986. "Audit Fees and Auditor Size: Further Evidence." Journal of Accounting Research (Spring): 97-110.
Rubin, M.A. 1988. "Municipal Audit Fee Determinants." The Accounting Review (April): 219-236.
Schatzberg, J.W. 1990. "A Laboratory Market Investigation of Low Balling in Audit Pricing." The Accounting Review (April): 337-362.
--. 1994. "A New Examination of Auditor `Low Ball' Pricing: Theoretical Model and Experimental Evidence." Auditing: A Journal of Practice & Theory (Supplement): 33-55.
Simon, D., and J. Francis. 1988. "The Effects of Auditor Change on Audit Fees: Tests of Price Cutting and Price Recovery." The Accounting Review (April): 255-269.
Stein, M.T., D.A. Simunic, and T.B. O'Keefe. 1994. "Industry Differences in the Production of Audit Services." Auditing: A Journal of Practice & Theory (Supplement): 128-142.
John C. Gardner Dean and Professor of Accounting Oakland University
Ronald J. Huefner Distinguished Teaching Professor State University of New York Buffalo
Vahid Lotfi Professor of Management Science University of Michigan--Flint