Functional and Data Fixation
DATA-FIXATION RESEARCH IN ACCOUNTING
Data-Fixation Research Based on the Ijiri-Jaedicke-Knight Paradigm
Functional-fixation research in accounting generally has followed Ijiri, Jae- dicke, and Knight’s prescriptions, focusing on data rather than function, and has led to a series of data-fixation experiments. Ashton used M.B.A. students to assess the extent to which individual decision makers alter their decision proc- esses after the occurrence of an accounting change, from full-cost to variable- cost data, as evidenced by the effect of this cognitive change on subsequent decisions.19Ashton not only discussed the accounting change with the subjects
but also mentioned whether it reflected more or less important informational content, and consequently may have dictated a change in the decision behavior of the subjects. This result suggests that a large proportion of subjects in the experimental groups failed to adjust significantly their decision process in re- sponse to the accounting change, thereby providing evidence of the existence of functional fixation in accounting. The study was not met with complete ap- proval. First, Libby criticized it for an experimental design that might have become confounded with the effects of the accounting change.20He concluded
that
serious questions concerning the way in which the conceptual network was operation- alized, coupled with methodology deficiencies, question whether any conclusions can
change the subjects, the manipulation of the moderating variables information and im- portance, and the method of measuring the change in the subject’s decision process.21
Second, Pearson, a practitioner, simply rejected the study’s objectives and results as irrelevant to accounting.22 These criticisms, as might be expected,
motivated further empirical research.
Swieringa, Dyckman, and Hoskin looked into Libby’s criticisms and found that subjects tended to adjust their information processing as a result of the accounting change even though the significance of these adjustments differed depending on how they were measured.23The amount of information provided
was found to influence the subjects’ adjustments of their information processing. Swieringa, Dyckman, and Hoskin had made two modifications in Ashton’s ex- perimental design. One modification was to isolate the effects of the amount and form of the information about the accounting change. The second modifi- cation was to have the data received by the control groups be equivalent to the data received by the experimental groups.
A second study by Dyckman, Hoskin, and Swieringa merely replicated the earlier study by Swieringa, Dyckman, and Hoskin with subjects who, on aver- age, were older and had more exposure to accounting and business matters.24
The students used in the first study were enrolled in an introductory accounting course in a college of agriculture and life sciences and did not know what direct costing meant. In addition, the second study relied on a cross-sectional approach instead of a time-series approach to analyze the effects of the experimental conditions and demographic variables on the prices set by the subjects for each product. The results of the second study were found to be similar to those of the first one.
In their experiment, Chang and Birnberg provided M.B.A. students with a cost variance report and a cost standard.25The subjects were required to indicate
(1) whether they would investigate the production process, and (2) how large a variance would be necessary to justify an investigation. Their results pointed to the existence of a “weak form” of data fixity when a change in the variance amount was introduced. The “weak form” label was used to characterize a slight change in behavior; no change in behavior was evidence of the “strong form” of fixity. Two significant findings were noted by the authors:
First, fixity is not a phenomenon that is unavoidable. Research indicates that once we are aware of its presence, we can take steps to cope with it. The real question becomes one of finding the manner in which it can be reduced and efficient ways of doing so. Second, unfortunately, once alerted to the problem, there is reason to believe that the subject’s behavior will continue to reflect elements of past behavior—behavior which should have been forgotten along with the superseded data set. This then suggests two topics for future research. One is how past experience affects the subject’s behavior. The other is how to extinguish the older, now unnecessary patterns of behavior.26
Abdel-Khalik and Keller used bank investment offices and security analysts in their investigation of functional fixation.27 They articulated their research
problems as follows:
If investors are functionally fixated on the use of reported accounting earnings, then they will tend to ignore other accounting information which is not consistent with accounting numbers. The accounting signal which we chose to be inconsistent with reported earnings is the decision of management to switch the method of inventory valuation from First- in, First-out (FIFO) or from average cost to Last-in, First-out (LIFO) for both accounting and tax purposes.28
Because of the higher cash flows that result from change to LIFO in a period of rising prices, the investor using a cash-flow discounted model would value the firm higher, while another relying and fixated on earnings would value it lower. The results of the experiment showed evidence of functional fixation, as the subjects relied on the adjusted net income rather than cash flows in evalu- ating the securities. One problem with Abdel-Khalik and Keller’s study is the fact that the firms that switched to LIFO received qualified audit opinions, while those on FIFO obtained unqualified opinions. This could explain why the LIFO firms generally were viewed as having lower expected returns.
Bloom, Elgers, and Murray extended the Ashton study by examining both individual and group decisions in response to a fully disclosed, cosmetic change in depreciation method.29 The results of the study showed a moderate shift in
the decision behavior of individuals, a phenomenon similar to what Chang and Birnberg called the weak form of fixation. In addition, they found that groups exhibited a higher degree of fixation than did individuals. Among the reasons given for this difference were the following: “One explanation is that the group process inhibited the collective or individual intellectual functioning of its mem- bers; yet another is that the groups incurred a higher cost in developing a new decision rule in response to the accounting change than did the individuals.”30
Another explanation was that the difference could be a reflection of the nature of the task, which consisted of the need both to reach a decision within the group on a decision rule and to make a decision on the task.31
Another accounting study provided evidence of functional fixation without being based on Ashton’s and Ijiri, Jaedicke, and Knight’s paradigms. A National Association of Accountants (NAA) research study on the effects of software accounting policies on bank lending decisions and stock prices showed clear evidence of fixation by loan officers making a decision on a loan to two fictional firms: the Campbell Corporation, which capitalized software expenditures; and the Edwards Corporation, which expensed all software costs.32 Without men-
tioning data fixation per se, the results were indicative of the presence of the phenomenon. Witness the following:
Campbell was favored over Edwards by 62.2% of the respondents; Edwards was favored by 11.1%; 13.3% would treat the companies equally, but did not give any reason for the
equal treatment; and 13.3% would treat the companies equally because a company’s software policy would not influence the lending decision. Only 27.3% of the bankers would grant a $3 million, five year unsecured loan to Edwards; compare with 61.4% for Campbell. Of those respondents that gave an interest rate for both companies, 55% would charge Campbell a lower rate, 5% would charge Edwards a lower rate, and 40% would charge the same rate to both companies.33
A similar finding was made in another study. Belkaoui conducted an experiment in which bank loan officers evaluated a loan application that was accompanied by financial statements based on either accrual or modified cash accounting.34
The loan officers in the experiment believed that the loan applicant presenting accrual accounting financial statements (1) was more likely to repay the loan, (2) was more likely to be granted the loan, (3) was given a different interest rate premium, and (4) had statements that were more reliable and freer of clerical errors.
Other Data-Fixation Research
Other accounting research studies have used the Ijiri-Jaedicke-Knight para- digm to explain their own results. This strategy has taken place both in the research of investor decisions and in capital market research.
In the research of investor decisions, a cross-sectional orientation was given to functional fixation as it was applied to alternative accounting methods rather than to changes in accounting methods over time. Jensen examined the impact of alternative depreciation and inventory costing methods on investor deci- sions.35 To explain his findings that alternative accounting techniques affected
decision making, he suggested that his subjects might be functionally fixated on net earnings. Livingstone examined the effects of alternative, interperiod tax- allocation methods on regulatory rate-of-return decisions affecting the electric utility industry.36In light of his findings that some rate-making books focus on
“raw” rates of return and ignore the effects of alternative tax-allocation methods, he offered the explanation that some predictions might be functionally fixated on net operating revenue. Livingstone stated the following:
It is therefore hypothesized that the reason that original-cost jurisdictions have been so much slower to adjust for alternative treatments of deferred taxes is that they are func- tionally fixated with respect to financial statement data. Since normalizing changes the amount but not the name of net operating revenue, it is intended that original-cost juris- dictions tend to view net operating revenue under normalizing as being the same as without it.37
Livingstone also suggested that users of accounting information could have formed a learning set after having experience with a significant number of dif- ferent problems, all of which can be solved in the same manner. One solution went as follows: “If the hypothesis of a learning set with respect to alternative
accounting methods is valid, multi-informational accounting statements would tend to stimulate learning and reduce functional fixation by providing users with information on accounting alternatives.”38 Mlynarczyk examined the effect of
alternative tax-accounting methods on common-stock prices of electric utility companies and related functional fixation to his work.39
In capital market research, the functional-fixation hypothesis has been used to explain the lack of efficiency in the capital market. Beaver argued, however, that the market is not functionally fixated.40He stated the following:
In essence, the implication of the functional fixation hypothesis is that two firms (secu- rities) could be alike in all “real” economic respects and yet sell for different prices, simply because of the way the accountant reported the results of operations. The impli- cation is that the market ignores the fact that observed signals are generated from different information systems. Hence, it does not distinguish between numbers generated by dif- ferent accounting methods either over time or across firms. Needless to say, this implies market inefficiency. . . . The functional fixation hypothesis as described above is a rather extreme form of the market inefficiency argument, in that it implies that disequilibrium could exist indefinitely and presumably permanently.41
DATA FIXATION AND FUNCTIONAL FIXATION IN