Printer Friendly

The value of activity-based costing in competitive pricing decisions.

Abstract: This paper reports experimental evidence on the merits of activity-based costing (ABC) for price-setting in competitive markets that differ in their ability to provide informative feedback. Earlier research has shown that informative market feedback dominates the effects of cost-system design. In a multimarket context involving cost allocations, the present results suggest that cost-system refinement can play a significant role in price-setting, even in the presence of informative market feedback. Specifically, ABC provides benefits over volume-based costing in market segments in which biased cost allocations produce accounting losses that hinder learning from superior competitors. Compared to these informative settings, additional evidence also shows that performance is negatively affected by less informative market feedback. Yet in less informative settings, ABC still outperforms traditional costing, presumably because it helps to filter irrelevant competitor feedback from the decision process.

Keywords: pricing; cost allocation; competition; activity-based costing.

Data Availability: All data are available upon request from the authors.

INTRODUCTION

This study provides experimental evidence on the potential benefits of alternative cost systems for pricing decisions in a competitive market context. More refined costing systems, such as activity-based costing (ABC), are often claimed to have a value-enhancing effect on pricing decisions and profit performance (Cooper 1988; Kaplan and Atkinson 1998; Goebel et al. 1998) because they allow better price differentiation among products, customers, and markets. Gupta and King (1997) show that the benefits of cost system refinement for decision making apply to individual market settings in which firms act as monopolists.

However, other evidence suggests that the benefits of ABC for price-setting do not necessarily extend to more competitive markets. The competitive setting itself offers valuable information (Bruns and McKinnon 1993; Malmi 1997; Waller et al. 1999). Waller et al. (1999) show that markets provide opportunities to learn from the price offers of superior market players, regardless of cost-report type. Briers et al. (1999) demonstrate that participants with biased cost data perform better when they receive highly informative market feedback. Thus, opportunities to learn from market competition can sharply reduce any effect of cost-system choice for price-setting.

Given that neither Waller et al. (1999) nor Briers et al. (1999) directly investigate ABC, our primary objective is to study the extent to which the availability of better-informed competitors as a tool for learning overshadows the value of ABC for price-setting. Second, earlier studies have generally assumed that market feedback is highly informative (an exception is Callahan and Gabriel 1998). However, this assumption is not always warranted. (1) Therefore, a second objective of this study is to test whether ABC leads to improved pricing decisions when the market has relatively less informational value. We expect that when market feedback provides less opportunity for learning, the value of ABC will increase.

In this experiment, participants set prices in multiple market segments in a competitive duopoly. Accounting report type and market feedback are manipulated as between-subjects factors. Contrary to Callahan and Gabriel's (1998) findings, the present results show that ABC has a strong benefit when market feedback is not very informative. Second, contrary to our expectation that highly informative market feedback would dominate cost-system choice, the results suggest that ABC still outperforms biased cost data when biased cost allocations generate accounting losses for a particular market segment. In this market segment, participants hesitate to follow the price choices of a superior competitor, presumably because they are averse to economically prudent decisions that produce accounting losses (Kachelmeier 1996; Tversky and Kahneman 1991). In this regard, this study qualifies the conclusions of Waller et al. (1999), by showing that informative markets do not always make cost-system choice redundant for price-setting.

THEORETICAL BACKGROUND AND HYPOTHESES

Price-Setting, Accounting Data, and the Role of Market Feedback

Most studies on the role of accounting data for price-setting study decision makers in individual settings. Early studies concluded that cost system choice mattered, since decision makers relied heavily on a given cost system's output (Ashton 1976; Barnes and Webb 1986; Dyckman et al. 1982). Wilner and Birnberg (1986) and Moon (1990) proposed that certain types of feedback might mitigate such reliance on the cost system. Latter studies documented only small mitigating effects of outcome feedback (Hilton et al. 1988), financial rewards (Briers et al. 1997), or process feedback (Gupta and King 1997); most participants still relied on their cost system for subsequent adjustments of the decision process. Hence, accurate cost systems such as ABC continued to confer a significant benefit over distorted cost data in individual settings without other market agents (Gupta and King 1997; Briers et al. 1997). (2)

Waller et al. (1999) argued that these results would not necessarily extend to a competitive market context, due to the pressure from market competition to utilize superior market feedback. They showed that the effects of absorption versus variable costing on price-setting quickly disappeared in such an environment because decision makers learned from the price offers of successful sellers in the market. Price offers rapidly converged toward the competitive equilibrium. They concluded that the contribution of the cost system is limited because market feedback dominates accounting information. Waller et al. (1999) were only concerned with the effects of variable costing versus absorption costing, as opposed to the broader issue of whether alternative full-costing systems such as ABC affect pricing decisions in a competitive context (Foster and Gupta 1994, 58). Second, Waller et al. (1999) considered only one product in their design, such that cost allocation was not an issue. However, Waller et al. (1999, 735) conjectured that learning from informative market feedback would be more difficult in a multimarket context that involved cost allocations, presumably because (mis)allocations may have an effect on an agent's behavior in a competitive setting.

In a multimarket setting, traditional costing can produce biased cost estimates in heterogeneous markets (Cooper and Kaplan 1998). When such cost allocations overestimate (or underestimate) costs for a particular market segment, firms' cost systems may report accounting losses (profits) for markets that actually are profitable (unprofitable). Briers et al. (1999) investigated the effects of biased cost data in a multiproduct context. Although they found that participants with biased cost data made improved pricing decisions when they received market feedback in the form of a benchmark report about the best practice of other firms, their experimental task did not involve any direct competition (Briers et al. 1999, 80). Second, they studied volume-based cost reports, but they did not directly test the effects of an ABC system. (3)

Therefore, whether feedback of superior market rivals reduces the incremental value of ABC relative to biased cost data in a competitive multimarket context involving cost allocations, remains an open question. This important issue is the central focus of our investigation. If cost allocations produce accounting losses under volume-based costing but not under a more accurate ABC report, then ABC could facilitate competitive pricing even if market feedback is informative, to the extent that market agents underutilize market feedback under traditional costing in order to avoid accounting losses. The "Results" section reports evidence consistent with this interpretation, which we present as an exploratory finding that we did not anticipate in designing the study.

As a second contribution, we consider market feedback that is not always diagnostic. Some market rivals may act on limited information such that their pricing decisions fail to reflect actual demand and cost parameters (Coughlan and Mantrala 1992, 91). Iselin (1996, 150) argues that such a situation would reduce performance to the extent that decision makers act on such irrelevant market cues. In turn, performance would be enhanced if irrelevant market cues were filtered from the decision process (Iselin 1996, 150). In this respect, ABC offers an important advantage over biased cost data. The more accurate cost data means that participants with ABC would be more likely to detect and filter competitors' prices when these prices are a poor reflection of actual costs than would participants with biased cost data. Consequently, prices under ABC are likely to be based on more accurate cost data (Ashton 1976; Briers et al. 1997) rather than on less relevant market feedback. Therefore, we expect better performance under ABC.

Our prediction contrasts with the findings of Callahan and Gabriel (1998) that accurate cost data do not improve pricing decisions in a Bertrand duopoly in which competitors issue uninformative market cues. Their counterintuitive finding may result from all participants in their study receiving demand and expected cost parameters in addition to competitor feedback and cost reports. Using these parameters, even subjects with biased cost data in their study could calculate a best-price response, making accurate cost data redundant. Because precise information on demand and expected cost is typically not available (Day and Groves 1975, 137-138), we did not provide these parameters to the participants in our experiment.

Hypotheses

We administer a repeated price-setting experiment with participants playing against a computer-modeled competitor (Bertrand duopoly) in two market segments with heterogeneous costs. Aside from outcome feedback, participants receive imperfect cost reports (either ABC or traditional volume-based costing) and market feedback (either informative or uninformative) in the form of a report containing the price choices and the profit of their competitor. Market feedback is labeled "informative" when the rival is set to play optimally with full knowledge of market parameters. We label the feedback as "uninformative" when the rival acts randomly, given a participant's prices, because then a competitor report does not convey relevant cues of "optimal" behavior.

We focus on the participant's decision performance measured as the extent to which prices and profits converge to optimal market performance. Our first hypothesis predicts a main effect of market feedback on decision performance. Frederickson (1992, 653) argues that in settings where several players perform a similar task, participants tend to compare their performance against their rivals. In our duopoly, both the participant and their competitor face a similar pricing task. This environment reinforces the notion that price choices of the fully informed competitor provide a useful basis for comparison. We expect that these informative market cues strongly facilitate convergence to optimal performance, as is evidenced in prior work (Briers et al. 1999; Waller et al. 1999). Conversely, if a competitor's price choices are not informative, we assume that performance will decrease to the extent that decision makers are influenced by these irrelevant market cues (Iselin 1996, 150):

H1: Prices and profits are closer to optimal when participants receive informative market feedback than when they receive uninformative market feedback.

The second hypothesis addresses our main research question regarding the value of ABC in the presence of market feedback. Based on the previous discussion, cost-system accuracy matters when market feedback is not informative. ABC is then better able to filter out irrelevant competitor feedback (Iselin 1996, 150). Consequently, participants should base pricing decisions on more relevant cost data (Ashton 1976), which should positively influence performance. When market feedback is informative, earlier work (Waller et al. 1999) suggests that such feedback overwhelms the effects of alternative cost systems for price-setting. We test whether the benefits of ABC over biased traditional cost data would also disappear in a multimarket context in which learning from market feedback is more difficult. If this is the case, then the results should reveal an interactive effect for cost system and competitor feedback, whereby ABC should outperform biased cost data when market feedback has low informational value, but not when market feedback is highly informative:

H2: The benefits of ABC over biased cost data for price-setting fall as market feedback becomes highly informative.

EXPERIMENT

Experimental Market Environment

Due to the focus on a multimarket context, our experimental setting defines two market segments (denoted A and B) in which participants compete on price against a computer-modeled competitor. A typical Bertrand demand function for differentiated products (that is, products with different brand names) defines each market segment in a manner similar to that of Callahan and Gabriel (1998):

(1) [Q.sub.is] = [u.sub.s] - [P.sub.s][Q.sub.is] + [w.sub.s][P.sub.js] for s = A, B

In Equation (1), [Q.sub.is] is the quantity demanded for firm i, [P.sub.is] is the price of firm i (participant), and [P.sub.js] is the price of firm j (competitor) in market segment s. Parameter [u.sub.s] ([u.sub.s] > 0) represents the demand at a zero price in segment s, and [v.sub.s], [w.sub.s] ([v.sub.s], [w.sub.s] > 0 and [v.sub.s] > [w.sub.s]) are fixed parameters for each market segment as summarized in Table 1. (4)

Cost functions in each market segment are the same for the participant and the computerized competitor. Given identical cost functions, the task faced by participants is similar to that of their rivals, and in this way competitors may serve as a useful source for comparison (Frederickson 1992, 653). Participants were informed of this equivalence in cost functions. The common cost function is:

(2) C([Q.sub.is]) = [f.sub.s] + [y.sub.s] [Q.sub.is] + [Z.sub.s][Q.sub.is.sup.2] for s = A,B

where [f.sub.s] is the fixed cost parameter, and [y.sub.s] and [z.sub.s] are fixed linear and quadratic slope parameters, respectively. These parameters result in highly heterogeneous market segments. Specifically, Table 1 shows that market A is a higher cost-to-serve market because it has a much higher fixed cost (parameter f) and also exhibits greater cost increases as output increases (parameters y and z). As a result, market A incurs strictly higher costs per unit of production than market B and optimal pricing requires a much higher price for market A to recover these costs. However, at the start of the experiment we deliberately introduced a price distortion by setting a lower price for market A than for market B, although it was more costly to produce in market A. We selected starting values (see Appendix A) to ensure ample and equal room for profit improvement under both market feedback conditions.

Participants attempted to maximize profits by differentiating prices across market segments based on imperfect cost reports and market feedback. The firm's objective function for each market segment has the following form:

(3) [[pi].sub.is] = [P.sub.is] ([u.sub.s] - [v.sub.s][P.sub.is] + [w.sub.s] [P.sub.js]) - [y.sub.s]([u.sub.s] - [v.sub.s][P.sub.is] + [w.sub.s] [P.sub.js]) - [z.sub.s][([u.sub.s] - [v.sub.s][P.sub.is] + [w.sub.s][P.sub.js]).sup.2] - [f.sub.s] s = A,B

Manipulated Factors

We manipulated two factors between subjects. The first factor is market feedback in the form of a report containing the competitor's previous period price choices and corresponding total profit. Market feedback is either highly informative (modeled as a rival making optimal price choices) or uninformative (modeled as a competitor setting a random price close to the participant's price). The competitor always moves second, after participants make their price choices.

In the informative market feedback condition, the competitor charged an optimal price, given the subject's price choice. The competitor's optimal price response is calculated by solving the first-order condition (maximizing firm j's profit given [P.sub.i]), as stated in Equation (4). Decision makers have a strong tendency to compare their own performance against their rival (Frederickson 1992). If participants incorporate these informative market signals into their pricing decisions, then the value of the cost system could be sharply reduced (Briers et al. 1999; Waller et al. 1999).

(4) [P.sub.js] = (2 [u.sub.s][v.sub.s][z.sub.s] + [u.sub.s] + [v.sub.s][y.sub.s]) + (2 [v.sub.s][[w.sub.s][z.sub.s] + [w.sub.s]) [P.sub.is]/2([v.sub.s] + [v.sub.s.sup.2][z.sub.s])

Under uninformative market feedback, the competitor randomly sets a price close to the subject's price. The competitor's price is based on a random draw from a uniform distribution over an interval around the participant's price, as specified by Equation (5). (5) We set the parameter a at 2 percent to make it difficult for subjects to infer that the rival's price strategy was randomly based on their own price choices. (6) Such a strategy is uninformative because it is not based on cues of optimal market performance. In fact, starting prices of this random competitor were (as for the participant's firm) irrelevant because they were far out of line with the actual costs (see Appendix A). More accurate ABC information is valuable because it reveals that the competitor's prices are a poor reflection of actual costs, thereby helping to filter irrelevant competitor cues from the decision process.

(5) [P.sub.js] = [(1 [+ or -] a)[P.sub.is]]

The second factor is the accounting report type. Instead of actual cost data, participants received imperfect cost reports as shown in Appendix A. Half of the participants received an ABC report. The system used a two-stage procedure (Kaplan and Atkinson 1998), which first allocated customer costs to three marketing activities (ordering, software handling, and delivery). Next, the ABC procedure allocated the cost of each activity across the two market segments via their respective activity drivers (number of orders, software licenses, and deliveries). Market A consumed more of these activities than market B, rendering it more costly to serve. Appendix A shows that ABC figures approximate the actual cost of serving the two market segments more closely than does the traditional cost system. The remaining error in the ABC cost data reflects the use of linear drivers to approximate the nonlinear cost function of Equation (2) (Christensen and Demski 1995).

The other half of the participants received a traditional accounting report, which allocates customer costs to market segments using sales volume in units as a cost driver. This driver is less effective in differentiating the cost of servicing both market segments. In general, traditional costing overestimates actual costs for market segment B, and underestimates actual costs for market segment A (see Appendix A).

Experimental Procedures

The participants consisted of 131 students from a master's cost-accounting course at the University of Leuven, Belgium. The course had covered the differences between ABC and traditional cost reports, and had discussed the use of ABC for segment profitability analysis and price-setting. We randomly assigned participants to the experimental cells upon entering the computer lab. Each session lasted one hour.

At the start of the experiment, subjects reviewed computer screens describing the case company and the pricing task. The case company imports a particular brand of portable PCs, and operates in two heterogeneous market segments. We informed participants that customers in market segment A were more demanding in terms of ordering, delivery, and software requirements. Subjects did not receive the actual parameters of the market segments, but they did receive an initial cost report and a report on the performance of a competitor at the beginning of the session. We informed participants that they would compete on prices against this competitor, who operated under a differentiated brand name in the same two market segments, and faced identical cost structures. In the uninformative market feedback condition, the case material described the competitor as a new market player, whereas in the informative condition, the competitor was an established market player.

Participants attempted to maximize profits by setting selling prices for PCs in each market segment. As motivation, subjects were notified that the four best players, based on the highest average profit over all experimental trials, would receive a 20 [euro] coupon exchangeable for books or CDs. (7) To provide room for improvement, we set the company's initial price strategy of 1650 [euro] for market A and 1710 [euro] for market B to deviate sharply from the profit-maximizing prices (see Table B1 of Appendix B). We required that prices fall in the range of 1100 [euro] to 2200 [euro] to ensure that demanded quantities remained positive at all feasible prices. After each of 10 trials, participants received an updated cost report for their own firm (traditional or ABC) and an updated report of the competitor's price choices and his profit performance (informative or uninformative). In addition, both the participant's and the competitor's price choices and profits for the preceding five trials remained on the screen. A multi-item exit questionnaire determined that motivation among participants was high (average: 4.22 on a five-point scale). No significant effects on self-perceived motivation were detected for accounting report ([F.sub.(1,127)]: 0.01; p > .50) or market feedback ([F.sub.(1,127)]: 0.72; p = .40).

RESULTS

Effects on Profit Performance

Our primary dependent variable is the participant's best profit score, i.e., his or her highest profit over the ten periods. To check the robustness of this approach, we also report results based on the average of the subject's best five scores. We focus on best or better scores in our setting because they were less sensitive to the few outlying observations resulting from trial and error behavior in the initial phases of our experiment. Table 2 reports an ANOVA model using the percentage difference between actual profit and the optimal profit (Table B1 in Appendix B) as the dependent variable (%dev.profit), (8) and accounting system (A) and market feedback (M) together with their interaction (AM) as the explanatory variables.

Panel B of Table 2 shows that the main effect of market feedback is significant at the p [less than or equal to] .01 level. Panel A of Table 2 confirms that the best (five best) score(s) of participants who receive highly informative market feedback are much closer to optimal profit than the scores of participants who receive uninformative feedback. We thus find support for H1's prediction that participants will perform better when they receive informative market feedback.

The interaction term between accounting report and market feedback (H2) is not significant (Panel B of Table 2). The second hypothesis, which predicts that ABC is beneficial to profit performance when market feedback is uninformative (competitor sets price randomly), but not when market feedback is informative (competitor sets price optimally) is therefore not supported. Only the main effect of accounting report type is significant, consistent with ABC having value in both the uninformative market and in the highly informative market. Similar inferences can be made from the means (Panel A). Apparently, the opportunity to learn from a well-informed competitor did not make accurate ABC cost data redundant.

To obtain a better understanding of this unexpected result, we conduct separate analyses for the two market segments in which subjects set prices. In addition to the relative distance from optimal profit in each market segment ([%dev.profit.sub.A] and [%dev.profit.sub.B]), we also test how close participants were to optimal prices ([%dev.price.sub.A] and [%dev.price.sub.B]) in each market. (9) The ANOVA models reported in Table 3 analyze the subjects' best (five best) profit score(s) in each market and the respective price(s) associated with these scores. The accounting system (A), market feedback (M) and their interaction (AM) are the explanatory variables.

Panel B of Table 3 indicates that the main effect of market feedback is significant in each market segment, thereby reinforcing H1. However, differences between the two market segments are evident with respect to H2. In market segment A, the interaction term is at least marginally statistically significant for all profit and price models (Panel B). (10) For market segment A, the means of Panel A of Table 3 suggest that consistent with H2, the interaction reflects that ABC has incremental value when a competitor provides little opportunity for learning, but not when market feedback becomes highly informative. In contrast, there is no support for H2 in market B. The interaction term of the ANOVA analyses for market B (Panel B) is not significant in any of the models. The means in Panel A of Table 3 indicate that ABC is apparently closer to the optimal profit and prices in both market feedback conditions.

In summary, the results for H2 appear to differ sharply across the two market segments. Contrary to the conclusions of Waller et al. (1999), we conclude that a well-informed competitor does not necessarily make cost-system choice completely redundant, because the effect of cost system still exists in one particular market segment. We next explore a potential explanation for this unexpected result.

Supplementary Analysis

We speculate that the unexpected result in market segment B may reflect participants with biased cost data underutilizing informative market feedback and setting prices too high due to loss aversion (Tversky and Kahneman 1991). By overstating the cost in market segment B, traditional cost allocations report a unit cost of around 1400 [euro]. Therefore, the optimal competitive price of 1362 [euro] generated accounting losses in market segment B. Tversky and Kahneman (1991) argue that participants are averse to loss-generating scenarios and try to adjust their decision process to avoid negative outcomes. According to this scenario, to avoid accounting losses in market segment B, participants with biased cost data would charge prices significantly above this cost even in the presence of informative market feedback. They would not consider that such losses resulted from biased cost allocations. The resulting performance of participants with traditional reports in segment B was inferior to that for participants with ABC. This result is consistent with the ABC system not generating accounting losses under an optimal competitive pricing strategy in market segment B. (11)

These findings bear some similarity to those in Kachelmeier (1996) where accounting methods that emphasized sunk (irrelevant) costs led sellers to seek significantly higher prices for their assets in an attempt to cover costs and avoid accounting losses. While these different market negotiation dynamics did not materially affect actual trading prices in Kachelmeier (1996), accounting losses resulting from biased cost allocations did influence actual prices in the current study.

We note that these effects were not observed in market segment A, apparently because traditional cost allocations produced accounting profits at optimal price levels in that segment. We added market segment as a repeated measure to the ANOVA models on profits to test whether the three--way interaction among accounting system, market feedback, and market segment is significant. Such a three-way interaction would corroborate our conjecture that informative market feedback removes the benefits of ABC over biased cost data when optimal pricing decisions produce a profit (segment A), but not when optimal competitive prices produce an accounting loss under traditional costing (segment B). The three-way interaction is indeed significant for both the best ([F.sub.(1,126)] = 4.49; p = .04) and the five best profit scores ([F.sub.(1.126)] = 3.8; p = .06). It appears that in a multimarket context, learning from competitor feedback is more difficult when traditional cost allocations produce accounting losses. This finding extends the single-market results of Waller et al. (1999) by suggesting a role for cost-system refinement in multimarket settings even in the presence of informative market feedback.

Finally, we analyzed the benefits of ABC when market feedback is not informative. In this case, we expect ABC to assist in filtering out irrelevant competitor behavior. As explained earlier, we initially set the prices of a competitor playing at random such that [P.sub.A] < [P.sub.B] while the optimal state requires [P.sub.A] > [P.sub.B]. We then checked whether participants would continue to mirror this price distortion ([P.sub.A] < [P.sub.B]) in further experimental trials. Participants followed the price distortion of the competitor on average in 7.9 out often trials for the cell with traditional cost data. This frequency was significantly higher than the corresponding frequency of 4.8 out of ten trials in the ABC cell ([F.sub.(1,64)]: 12.88; p < .01).

These results support the conclusion that irrelevant market cues are effectively filtered with more accurate cost data because participants with ABC reverse the price distortion sooner than participants with traditional costing.

Manipulation Check

As a manipulation check, the exit questionnaire probed whether participants considered the price choices of their competitor important for price-setting. We report an ANOVA with the participants' responses to this question as the dependent variable. Although there were no significant effects of accounting system ([F.sub.(1.127)] < 1, ns) and market feedback ([F.sub.(1,127)] < 1, ns), the interaction of these factors was significant ([F.sub.(1,127)]: 4.90; p = .03). The absence of a main effect from market feedback and the question's high overall rating (average of 4.16 on a five-point scale) are consistent with subjects using even irrelevant competitor data (Iselin 1996). More importantly, the significant interaction and the means per cell show that, although participants with traditional cost data found competitor data to be less important than did their ABC counterparts when facing an informed competitor, this relation was reversed for uninformative feedback. (12) This is in line with the results reported earlier. Indeed, compared to traditional costing, market feedback under ABC is used to a greater extent when it is informative (and hence considered more important) while it is less used when it is not informative (and hence considered less important).

DISCUSSION

This study investigates the potential benefits of ABC for price-setting in competitive markets that differ in the degree to which a competitor's results provide informative feedback. Prior research (Waller et al. 1999) concluded that the effects of alternative cost systems do not persist when there is opportunity to learn from the market. An important contribution of our study is demonstrating that such conclusions are not always valid in a multimarket context involving cost allocations across markets. We show that in a market segment in which optimal pricing decisions produce accounting losses under volume-based costing but not under ABC, market agents with biased cost data underutilize informative market feedback. Due to the fact that subjects are averse to such losses (Tversky and Kahneman 1991), prices are based on biased cost data, rather than on competitive market feedback and performance is worse than under more accurate ABC costing. Conversely, in a market setting in which both volume-based costing and ABC produce accounting profits under optimal pricing, our results corroborate those of Waller et al. (1999) that informative competitor feedback can dominate more accurate cost data.

We also studied the issue of whether ABC can improve prices when competitor feedback is less informative. Although performance decreased because participants relied heavily on uninformative market feedback, ABC reports had a strong benefit over biased traditional cost data, presumably because ABC enabled decision-makers to filter out less relevant competitor prices from the decision process. Our findings are contrary to those of Callahan and Gabriel (1998), who did not find any value of ABC in a similar uninformative market setting. A likely explanation is that unlike their study, we did not provide participants with ex ante information on actual cost and demand parameters.

Future work may examine whether users of cost reports are generally oversensitive to accounting losses. Although cost system bias likely occurs under traditional costing (Cooper 1988), there might exist scenarios in which ABC produces accounting losses (for example, the optimal state requires an acceptance of a loss because a market segment is very costly), whereas traditional costing might show a biased but positive profit figure. In such settings, aversion to losses by individuals using ABC could plausibly lead to poorer performance than under traditional costing. However, because ABC systems also typically provide diagnostic feedback on specific activities, they are likely to help managers understand the reasons for the losses (Bonner 1994). This improved understanding may reduce the effect of loss aversion in case of ABC.

Further research may consider varying the intensity of competition. In the current study, because traditional costing produced accounting losses in one market segment, informative market feedback did not completely dominate cost systems. However, in our study participants do not lose all sales when they fail to follow market feedback, in contrast to Kachelmeier's (1996) more competitive market setting in which failure to follow market discipline can lead to exclusion from the market. In Kachelmeier's (1996) setting, cost-report variances influenced only the bid-ask dynamics of market negotiation. Actual trading prices were driven by competitive equilibrium predictions.

We conclude by noting certain limitations that also deserve further investigation. First, our experimental design does not incorporate capacity costs of serving markets (e.g., the common cost of a distribution center from which different markets are served). The problem of whether cost reports should report (un)used capacity in a competitive context is, however, a fascinating issue. Just as with accounting losses, participants may focus on eliminating unused capacity, which may not be optimal because competitive strength is lost when the firm would attract additional demand from other rivals. Second, for reasons of experimental control, we employed a computerized rival who responded either optimally or randomly to the prices of a participant. To validate our results, future research could evaluate how ABC costing interacts with other sources of feedback in a more natural competitive setting with participants interacting with each other.

APPENDIX A

Actual Cost Structure and the Types of Cost Reports

This appendix shows how we determine the costs for both market segments under ABC or traditional costing. Only the uninformative market feedback setting is illustrated. Analysis is similar under informative market feedback. At the start of the experiment, we set prices of the participant's firm at 1650 [euro] for market A and 1710 [euro] for market B (Panel A of Table A1). Panel A of Table A1 shows the actual results for a participant's firm when his rival randomly sets [P.sub.jA] = 1635 [euro] and [P.sub.jB] = 1740 [euro] as shown in Panel B of Table A1. In fact, these initial starting prices are not consistent with actual costs because market segment A with the highest cost receives the lowest price. The results for the participant's firm are based on the functions and the parameters in Table 1. The profit figure in the competitor report (Panel B) is based on the competitor's profit function. Only the competitor feedback is issued to participants. The results for the participant's firm are replaced by either a traditional cost report or an ABC report, as displayed in Panels A and B of Table A2.

Both cost reports assume that the participant firm's actual cost (4,393,672; Panel A of Table A1) includes the cost of goods sold. Products are imported at a fixed price that is slightly higher for market B (694.8) than for market A (659.35):

Cost of goods sold = 659.35 x [Q.sub.iA] + 694.8 x [Q.sub.iB] = 659.35 x 2,349 + 694.8 x 720 = 2,049,069

However, the cost reports differ markedly with respect to the allocation across market segments of the remainder of actual costs (4,393,672 - 2,049,069 = 2,344,603), defined here as "customer costs." Traditional volume-based costing allocates customer costs using sales volume as the driver (Panel A of Table A2). Compared to actual cost data, a traditional volume-based cost report produces highly biased unit costs. Specifically, the volume-based allocation assigns too much cost to market B and too little to market A, as detailed below.

ABC uses a two-stage procedure to allocate the customer costs (see assumptions in Panel C of Table A2). The first stage allocates customer costs to three cost-to-serve activities--ordering, delivery, and software handling--on the basis of the time that the activity consumes. The second stage allocates the cost of each activity to the two market segments on the basis of activity drivers, assuming that market A requires more activities (i.e., orders, deliveries, and custom design) than market B. As a result, the ABC report in Panel B of Table A2 shows that market A incurs more costs per unit than market B. This pattern more closely fits the actual cost data of Table A1.

APPENDIX B

Optimal Prices

This appendix derives price levels that maximize the decision-maker's profits in each type of market feedback condition. Because the competitor moves second, maximum profit levels are achieved when the decision maker anticipates the competitor's pricing strategy and employs the game theoretical equilibrium, given that strategy. The "Results" section uses these optimal profit and price levels as benchmarks against which to compare the actual performance of a participant. To derive optimal responses, we begin with the participant's profit function in Equation (B1):

(B1) [[pi].sub.is] = [P.sub.is] ([u.sub.s] - [v.sub.s][P.sub.is] + [w.sub.s] [P.sub.js]) - [y.sub.s]([u.sub.s] - [v.sub.s][P.sub.is] + [w.sub.s][P.sub.js] - [z.sub.s][([u.sub.s] - [v.sub.s][P.sub.is] + [w.sub.s][P.sub.js]).sup.2] - [f.sub.s]

When market feedback is informative, the competitor's fixed pricing strategy is given by Equation (B2). Participants can anticipate this pricing strategy by substituting [P.sub.js] from formula Equation (B2) in Equation (B1). The profit objective function is then determined completely in terms of [P.sub.is]. The first-order condition for [P.sub.is] yields the optimal price response for a decision maker in the informative-market feedback condition, as shown in Equation (B3).

Competitor Response in the Informative Market

(B2) [P.sub.js] = (2 [u.sub.s][v.sub.s][z.sub.s] + [u.sub.s] + [v.sub.s][y.sub.s]) + (2 [v.sub.s][w.sub.s][z.sub.s] + [w.sub.s]) [P.sub.is]/2([v.sub.s] + [v.sub.s.sup.2][z.sub.s]

Participant's Optimal Pricing Strategy

(B3) [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.]

If market feedback is uninformative, then the competitor sets a price within a random range of the participant's price as given by Equation (B4). In the long run, such a rival charges, on average, the same price as a participant; see expected value in Equation (B5). Substituting [P.sub.js] for [P.sub.is] from Equation (B5) in Equation (B1) and solving the first-order condition yields the participant's best response Equation (B6) that maximizes profits when market feedback is uninformative.

Competitor Response in the Uninformative Market

(B4) [P.sub.js] = [(1 [+ or -] a)% [P.sub.is]]

(B5) Expected value: E([P.sub.js]) = [P.sub.is]; since interval [1 [+ or -] a] is uniformly distributed.

Participant's Optimal Pricing Strategy

(B6) [P.sub.is] = [u.sub.s] + [v.sub.s][y.sub.s] - [w.sub.s][y.sub.s] + 2[u.sub.s][v.sub.s][z.sub.s] - 2[u.sub.s][w.sub.s][z.sub.s]/ 2([v.sub.s] - [w.sub.s] + [v.sub.s.sup.2][z.sub.s] - 2[v.sub.s][w.sub.s][z.sub.s] + [z.sub.s][w.sub.s.sup.2])

Table B1 displays optimal performance levels in each market feedback condition. The maximum profit level associated with the subject's best reply in an uninformative market can vary depending on the competitor's random price response.
TABLE A1
Actual Results at the Start of the Experiment

Panel A: Actual Results for the Participant's
Firm (not issued to participants)

                             Segment A    Margin   Segment B

Price ([P.sub.is])               1,650                 1,710
Sales Volume ([Q.sub.is])        2,349                   720
Revenue                      3,875,850             1,231,200
Actual Cost (a)              3,480,696     89.8%   9,129,976
Profit                         395,154     10.2%     318,224
Cost/Unit                     1,481.78              1,268.02

                                Margin      Total     Margin

Price ([P.sub.is])
Sales Volume ([Q.sub.is])                   3,069
Revenue                                 5,107,050
Actual Cost (a)                  74.2%  4,393,672      86.0%
Profit                           25.8%    713,378      14.0%
Cost/Unit

Panel B: Competitor Feedback Issued to Participants (b)

Competitor Report

Price Market A ([P.sub.jA])      1,635
Price Market B ([P.sub.jB])      1,740
Total Profit                   661,445

(a) For example, we calculate the cost for market A using
the functions and the parameters of Table 1; C([Q.sub.iA])
= 1,750,000 + 220 (2,349) + 0.22 (2,349)2 = 3,480,696.

(b) This figure is based on the competitor's total profit
function in combination with the parameters of Table 1
([MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.]
for s = segment A, B).

TABLE A2
Types of Cost Reports Issued to the
Participants at the Start of the Experiment

Panel A: Traditional Costing Report

                            Segment A      Margin   Segment B

Sales Volume ([Q.sub.is])       2,349                     720
Price ([P.sub.is])              1,650                   1,710
Revenues                    3,875,850               1,231,200
Cost of Goods Sold          1,548,813       40.0%     500,256
Customer Costs (a)          1,794,550       46.3%     550,053
Profit                        532,487       13.7%     180,891
Cost/Unit                    1,423.31                1,458.76

                               Margin       Total      Margin

Sales Volume ([Q.sub.is])                   3,069
Price ([P.sub.is])
Revenues                                5,107,050
Cost of Goods Sold              40.6%   2,049,069       40.1%
Customer Costs (a)              44.7%   2,344,603       45.9%
Profit                          14.7%     713,378       14.0%
Cost/Unit

Panel B: ABC Costing Report

                            Segment A      Margin   Segment B

Sales Volume ([Q.sub.is])       2,349                     720
Price ([P.sub.is])              1,650                   1,710
Revenues                    3,875,850               1,231,200
Cost of goods sold          1,548,813       40.0%     500,256
Customer Costs (a)          2,038,315       52.6%     306,288

Cost Details (b)           Drivervol.        Cost  Drivervol.

Order processing               352.35     730,988       43.20
Software handling            5,402.70     808,539      864.00
Delivery                       164.43     498,788       28.80
                                        2,038,315

Profit                        288,722        7.4%     424,656
Cost/Unit                    1,527.09                1,120.20

                               Margin       Total      Margin

Sales Volume ([Q.sub.is])                   3,069
Price ([P.sub.is])
Revenues                                5,107,050
Cost of goods sold             40.60%   2,049,069       40.1%
Customer Costs (a)             24.90%   2,344,603       45.9%

Cost Details (b)                 Cost  Drivervol.        Cost

Order processing               89,623      395.55     820,611
Software handling             129,302    6,266.70     937,841
Delivery                       87,363      193.23     586,151
                              306,288               2,344,603

Profit                          34.5%     713,378       14.0%
Cost/Unit

Panel C: Assumptions of ABC (not shown to participants)

Stage 1: Customer Costs to Activities

                            % of time

Order processing               35
Software handling              40
Delivery                       25

Stage 2: Number of Resources per 100 Units Sold

                            Segment A   Segment B

Number of orders                15           6
Number of licenses             230         120
Number of deliveries             7           4

(a) Allocated via sales volume; e.g., segment A is
assigned [2,349/(2,349 + 720)] * 2,344,603 = 1,794,550.

(b) Allocation via assumptions of Panel C. We show an example
for the order processing cost. In stage one, 35 percent of the
customer costs (0.35 * 2,344,603 = 820,611) are allocated to
order processing. In stage two, driver volumes for market A
[2,349 * 15/100 = 352.35] and B [720 * 6/100 = 43.20] are
calculated. Segment A is then assigned [352.35/(352.35 + 43.20)]
* 820,611 = 730,988 of order-processing costs.

TABLE B1
A Subject's Optimal Price and Profit
Levels in Each Market Feedback Condition

                                        Informative
Optima                                     Market

Price of segment A ([P.sub.A.sup.*])      1,833.8
Price of segment B ([P.sub.B.sup.*])      1,362.4
Profit of segment A ([p.sub.A.sup.*])     595,823
Profit of segment B ([p.sub.B.sup.*])     345,361
Total Profit ([[phi].sup.*])              941,184

                                           Uninformative
Optima                                         Market

Price of segment A ([P.sub.A.sup.*])          1,951.6
Price of segment B ([P.sub.B.sup.*])          1,476.5
Profit of segment A ([p.sub.A.sup.*])    [605,612; 686,600]
Profit of segment B ([p.sub.B.sup.*])    [352,582; 370,718]
Total Profit ([[phi].sup.*])            [958,194; 1,057,318]

The figures were verified by a spreadsheet program
that returned the same optima as displayed here.

TABLE 1
Demand and Cost Parameters in Each Market Segment

             Demand: [Q.sub.is] =     Cost: C([Q.sub.is] =
            [u.sub.s] - [v.sub.s]    [f.sub.s] + [y.sub.s]
            [P.sub.is] + [w.sub.s]   [Q.sub.is] + [z.sub.s]
                  [P.sub.js]            [Q.sub.is.sup.2]

             u       v       w         f        y       z

Segment A   5,500    3.0     1.1   1,750,000     220    0.22
Segment B   2,250    1.2     0.3     700,000     195    0.14

[P.sub.is] (P.sub.is]) is the price for subject i (competitor j);
[u.sub.s], [v.sub.s], and [w.sub.s] are demand parameters; [Q.sub.is]
is the quantity demanded for subject i, and [f.sub.s], [y.sub.s], and
[z.sub.s] are cost parameters for market segment s (s = segment A, B).

TABLE 2
Effects of Accounting System and Market Feedback on Profit
Performance Panel A: Mean Deviation from Optimal Profit

                         Market Uninformative     Market Informative

                       Traditional     ABC      Traditional    ABC
Variable (a)              n = 32       n = 33      n = 33     n = 33

% dev. Profit
  Best score             22.98%       17.83%       2.70%      1.10%
  Best five scores      [25.85%]     [20.75%]     [4.03%]    [1.74%]

Panel B: F-values and Significance Levels of the ANOVA-Analyses (b)

                     Y= % dev.profit based on

                                    Best Five
Source of Variation    Best Score     Scores

Accounting System (A)    7.58 ***     9.17 ***
Market Feedback (M)    228.21 ***   279.72 ***
Interaction (AM)         2.10         1.33
F-model                 78.96 ***    96.37***

*, **, ***, Significant at the p [less than or equal to] .10,
p [less than or equal to] .05, and p [less than or equal to]
.0l levels, respectively.

(a) % dev.profit is the percentage difference between a
participant's realized profit and the optimal profit. Each
cell contains the means of two metrics in which realized
profit is either based on the subject's best or an average
of his five best profit scores.

(b) ANOVA-models use the two profit metrics from Panel A as
dependent variables and the experimental factors and their
interaction as sources of variation.

TABLE 3
Effects of Accounting System and Market Feedback by Market Segment

Panel A: Mean Deviation from Optimal
Profits and Prices by Market Segment

                                Market Uninformative

                                Traditional      ABC
Variable (a)                      n = 32       n = 33

Market A: % [dev.profit.sub.A]
  Best score                      24.36%       17.45%
  Best five scores               [28.29%]     [22.10%]
Market A: % [dev.price.sub.A]
  Best score                      10.67%        7.95%
  Best five scores               [11.64%]      [9.26%]
Market B: % [dev.profit.sub.B]
  Best score                      11.55%        8.69%
  Best five scores               [15.97%]     [12.48%]
Market B: % [dev.price.sub.B]
  Best score                      12.76%        9.72%
  Best five scores               [14.69%]     [11.82%]

                                 Market Informative

                                Traditional      ABC
Variable (a)                      n = 33       n = 33

Market A: % [dev.profit.sub.A]
  Best score                       1.39%        1.43%
  Best five scores                [2.30%]      [2.12%]
Market A: % [dev.price.sub.A]
  Best score                       1.83%        1.78%
  Best five scores                [2.49%]      [2.35%]
Market B: % [dev.profit.sub.B]
  Best score                       3.50%        0.02%
  Best five scores                [5.61%]      [0.44%]
Market B: % [dev.price.sub.B]
  Best score                       4.05%        1.03%
  Best five scores                [5.67%]      [2.03%]

Panel B: F-values and Significance of
the ANOVA-models for Markets A and B (b)

                             Y = % [dev.profit         Y = % [dev.price
                             .sub.B] Based on          .sub.B] Based on

                                         Best                    Best
                            Best         Five        Best        Five
Market A                    Score       Scores       Score      Scores

Source of variation
  Accounting System (A)    3.23 ***    3.53 *      4.67 **     4.78 **
  Market Feedback (M)    168.17 ***  113.83 ***  189.15 ***  140.63 ***
  Interaction (AM)         2.88 *      3.62 *      3.67 *      4.42 **
F-model                   57.85 ***   40.11 ***   65.53 ***   49.67 ***

                             Y = % [dev.profit         Y = % [dev.price
                             .sub.B] Based on          .sub.B] Based on

                                         Best                    Best
                            Best         Five        Best        Five
Market B                    Score       Scores       Score      Scores

Source of variation
  Accounting System (A)    8.52 ***   6.60 **     12.72 ***  11.37 **
  Market Feedback (M)     56.17 ***  45.81 ***   105.90 ***  93.64 ***
  Interaction (AM)         0.32       0.07         0.18       0.01
F-model                   21.81 ***  17.42 ***    39.45 ***  34.84 ***

*, **, ***, Significant at the p [less than or equal to] .10,
p [less than or equal to] .05, and p [less than or equal to]
.01 levels, respectively.

(a) We report the means of two profit deviation metrics
([dev.profit.sub.A] and [dev.profit.sub.B]) for markets A and B
based on a subjects' best or an average of the five best profit
scores for that market segment. In addition, we report the means
of two price deviation metrics ([dev.price.sub.A] and [dev.price
.sub.B]) for markets A and B defined as an absolute percentage
difference of the price linked with the best (five best) profit
score(s) relative to the optimal price for that market segment.

(b) ANOVA-models use the four defined deviation metrics as
dependent measures and the experimental factors and their
interaction as sources of variation.


We are grateful to two anonymous reviewers for helpful comments and suggestions. We also thank Gustaaf Van Herck and participants of the doctoral seminar in accounting (K.U.Leuven, 2002), the seminar of behavioral research (Durbuy, 2002) and the 25th EAA congress (Copenhagen, 2002) for their advice on earlier drafts. Preliminary work related to this study appeared as a tribute in honor of our retiring colleague, Prof. Marcel Van Aeoleyen, in the Tijdschrift voor Economic en Management (vol. XLVI, November 2001).

(1) Evidence suggests that competitors often act with limited information and rationality (Coughlan and Mantrala 1992). Competitors also sometimes deliberately choose to issue uninformative data to the market (Gal-Or 1986).

(2) In a similar fashion to laboratory markets (e.g., Gupta and King 1997), survey research (e.g., Shim and Sudit 1995; Drury and Tayles 2000) has shown that firms rely heavily on cost data for price-setting (cost-plus pricing) when other information is limited (Noble and Gruca 1999).

(3) We note that Briers et al. (1999) did include a second group of participants with biased cost data who received process feedback on the resource consumption of products. Although this feedback contained ABC-like data, such as aggregated driver and activity cost information, their focus was on whether such feedback could signal potential cost system bias. However, this process feedback is still far from a full ABC-costing system since the ABC-like data was not formally used to perform any specific cost allocation to the individual product level.

(4) In the present study, setting [v.sub.s] > [w.sub.s] makes the firm's price-demand effect stronger than the competitor's cross-price demand effect, similar to the setting of Callahan and Gabriel (1998). In such a Bertrand duopoly, players are left with residual demand if the rival charges a lower price.

(5) New entrants with limited information are likely to follow the actions of other market players in the market (Coughlan and Mantrala 1992). Such behavior can also be observed in situations in which smaller firms follow an existing market leader (Cooper 1996).

(6) We assessed the 2 percent parameter value in a pre-test. None of the subjects involved in the pre-test determined that the strategy of his or her rival was based on the participant's own price choices.

(7) We actually granted one coupon in each of our four treatments. Although this incentive structure is not common in experimental economics, it has precedence in psychological studies of similar decision tasks (McIntyre and Ryans 1983). It is effectively a form of "tournament," and as such, may lead to dysfunctional effects among losing participants. For example, because only four out of 131 participants win a prize, participants have incentive to make risky decisions in order to win the prize (Bonner et al. 2000). To minimize such effects, we instructed participants to perform well on every trial because average profit was the basis of reward. In addition, it is less likely that dysfunctional effects interacted with the treatment factors in the experiment because overall motivation was high and the incentive structure led to similar scores on motivation across experimental cells.

(8) The distance from optimal profit is represented by %dev.profit = ([[pi].sup.*] - [[pi].sub.i])/[[pi].sup.*], where [[pi].sup.*] is the optimal profit and [[pi].sub.i] the subject's realized profit based on either the best profit score or an average of the five best profit scores. Because optimal profits fluctuate within an interval due to random responses in the case of an uninformed competitor (see Table B1 of Appendix B), the upper limit of this interval is used for [[pi].sup.*].

(9) The profit metrics, [%dev.profit.sub.A] and [%dev.profit.sub.B], are defined for each market segment in a similar manner to the total profit metric. The price metrics use following formulas: [%dev.price.sub.A] = abs([P.sub.A.sup.*] - [P.sub.iA])/[P.sub.A.sup.*], and [%dev.price.sub.B] = abs([P.sub.B.sup.*] - [P.sub.iB])/[P.sub.B.sup.*], where [P.sub.A.sup.*] and [P.sub.B.sup.*] are the optimal prices and [P.sub.iA] and [P.sub.iB] are the participant's prices asso with their best (or best five) profit score(s) in market A and market B. The absolute value is used because prices above and below the optimum are possible. Optimal values by market segment can be found in Table B1 of Appendix B.

(10) Some effects for market A in Panel B of Table 3 are marginally significant due to one outlier. Deleting this outlier would make all these effects significant at the 5 percent level.

(11) Given informative market prices and traditional volume-based costing, both the mean price over all trials (= 1472.6) and the participants' mean price in the most profitable period in market B (= 1411.3) exceeded their reported unit cost of about 1400 [euro]. This is consistent with participants being reluctant to accept accounting losses at optimal price levels. In contrast, participants with ABC apparently continued to use informative market feedback. This conclusion is supported by the fact that their mean price over all trials (= 1396.1) and the mean price related to their best profit score (= 1364.2) were closer to the optimal price of 1362.4 in market B.

(12) In the case of informative feedback, the average score under traditional costing was only 4.12, whereas under ABC it was 4.33. When market feedback was uninformative, participants with ABC found market feedback less important (3.91) than their volume-based costing counterparts (4.31)

REFERENCES

Ashton, R. H. 1976. Cognitive changes induced by accounting changes: Experimental evidence on the functional fixation hypothesis. Journal of Accounting Research 14 (Supplement): 1-17.

Barnes, P., and J. Webb. 1986. Management information changes and functional fixation: Some experimental research from the public sector. Accounting, Organizations and Society 11 (1): 1-18.

Bonnet, S. E. 1994. A model of the effects of audit task complexity. Accounting, Organizations and Society 19(3): 213-234.

--, R. Hastie, G. Sprinkle, and M. Young. 2000. A review of the effects of financial incentives on performance in laboratory tasks: Implications for management accounting. Journal of Management Accounting Research 12 (Fall): 19-64.

Briers, M., P. Luckett, and C. Chow. 1997. Data fixation and the use of traditional versus activity-based costing systems. ABACUS 33 (1): 49-68.

--, C. Chow, N. Hwang, and P. Luckett. 1999. The effects of alternative types of feedback on product-related decision performance: A research note. Journal of Management Accounting Research 11 (Fall): 75-92.

Bruns, W. J., and S. M. McKinnon. 1993. Information and managers: A field study. Journal of Management Accounting Research 5 (Fall): 84-108.

Callahan, C. M., and A. E. Gabriel. 1998. The differential impact of accurate product cost information in imperfectly competitive markets: A theoretical and empirical investigation. Contemporary Accounting Research 15 (Winter): 419-455.

Christensen, J., and S. J. Demski. 1995. The classical foundations of modern costing. Management Accounting Research 6 (March): 13-32.

Cooper, D. 1996. Barometric price leadership. International Journal of Industrial Organization 15:301-325.

Cooper, R. 1988. The rise of activity-based costing: When do I need an activity-based cost system? Journal of Cost Management 2 (Fall): 41-48.

--, and R. S. Kaplan. 1998. Cost and Effect. Boston, MA: Harvard Business School Press.

Coughlan, A. T., and M. K. Mantrala. 1992. Dynamic competitive pricing strategies. International Journal of Research in Marketing 9: 91-108.

Day, R. H., and T. Groves. 1975. Adaptive Economic Models. New York, NY: Academic Press.

Drury, C., and M. Tayles. 2000. Cost System Design and Profitability Analysis in UK Companies. London, U.K.: Chartered Institute of Management Accountants.

Dyckman, T., R. Hoskin, and R. J. Swieringa. 1982. An accounting change and information processing changes. Accounting, Organization and Society 7:1-12.

Foster, G, and M. Gupta. 1994. Marketing, cost management and management accounting. Journal of Management Accounting Research 6 (Fall): 43-77.

Frederickson, J. R. 1992. Relative performance information: The effects of common uncertainty and contract type on agent effort. The Accounting Review 67 (October): 647-669.

Gal-Or, E. 1986. Information transmission--Cournot and Bertrand equilibria. Review of Economic Studies 53: 85-92.

Goebel, D. J., G W. Marshall, and W. B. Locander. 1998. Activity-based costing: Accounting for a marketing orientation. Industrial Marketing Management 27:497-510.

Gupta, M., and R. R. King. 1997. An experimental investigation of the effect of cost information and feedback on product cost decisions. Contemporary Accounting Research 14 (Spring): 99-127.

Hilton, R. W., R. J. Swieringa, and M. J. Turner. 1988. Product pricing, accounting costs and the use of product-costing systems. The Accounting Review 63 (April): 195-218.

Iselin, E. R. 1996. Accounting information and the quality of financial managerial decisions. Journal of Information Science 22 (2): 147-153.

Kachelmeier, S. J. 1996. Do cosmetic reporting variations affect market behavior? A laboratory study of the accounting emphasis on unavoidable costs. Review of Accounting Studies 1: 115-140.

Kaplan, R. S., and A. A. Atkinson. 1998. Advanced Management Accounting. Third edition. Englewood Cliffs, NJ: Prentice Hall Inc.

Malmi, T. 1997. Toward explaining activity-based costing failure: Accounting and control in a decentralized organization. Management Accounting Research 7: 459-480.

McIntyre, S. H., and A. B. Ryans. 1983. Task effects on decision quality in travelling salesperson problems. Organizational Behavior and Human Performance 32: 344-369.

Moon, P. 1990. Some experimental evidence on functional fixation: A research note. Accounting, Organizations and Society 15 (3): 193-197.

Noble, P. M., and T. S. Gruca. 1999. Industrial pricing: Theory and managerial practice. Marketing Science 18 (3): 435-454.

Shim, E., and E. F. Sudit. 1995. How manufacturers price products. Management Accounting 76 (2): 37-39.

Tversky, A., and D. Kahneman. 1991. Loss aversion in riskless choice: A reference-dependent model. Quarterly Journal of Economics (November): 1039-1061.

Waller, W., B. Shapiro, and G. Sevcik. 1999. Do cost-based pricing biases persist in laboratory markets? Accounting, Organizations and Society 24 (8): 717-739.

Wilner, N., and J. Birnberg. 1986. Methodological problems in functional fixation research: Criticisms and suggestions. Accounting, Organizations and Society 11 (1): 71-80.

Eddy Cardinaels

University of Tilburg

Filip Roodhooft

Katholieke Universiteit Leuven and Vlerick Leuven-Gent Management School

Luk Warlop

Katholieke Universiteit Leuven
COPYRIGHT 2004 American Accounting Association
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2004 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Cardinaels, Eddy; Roodhooft, Filip; Warlop, Luk
Publication:Journal of Management Accounting Research
Geographic Code:1USA
Date:Jan 1, 2004
Words:9821
Previous Article:Managing value creation within the firm: an examination of multiple performance measures.
Next Article:Lessons lost in linearity: a critical assessment of the general usefulness of LEN models in compensation research.
Topics:

Terms of use | Privacy policy | Copyright © 2019 Farlex, Inc. | Feedback | For webmasters