Vous êtes sur la page 1sur 34
Behavioral Research in Accounting Volume 1, 1989 Printed in USA The Evolution of Behavioral Accounting Research in the United States, 1968-1987* David Burgstabler and Gary L. Sundem+ Untwersity of Washington ABSTRACT This paper reviews behavioral accounting research (BAR) from the perspective of nonbehavioral researchers (information economists). The role of BAR in identifying the demand function for accounting information is discussed. A brief review of trends in the types of BAR published in three major journals provides a background for general observations about the current state of BAR. Seven inherent difficulties of BAR are presented, followed by five areas where significant progress has been made in the past 20 years and seven areas of potential improvements. The paper concludes that BAR will have an important influence on the accounting products of the future. However, despite significant progress in the 1970s and 1980s, BAR is stil a broad, unfocused discipline, with increasing quality of research and researchers, but not yet to the point of having a unified set of research results that are generally accepted. In the 1960s and early 1970s behavioral accounting research (BAR) was isolated from other accounting research by its research methods and the topics addressed. But, in the last 10 or 15 years, the line between BAR and other areas of accounting research has become harder to draw. For example, researchers in experimental economics apply behavioral research methods to examine questions traditionally posed by economists, and analytic modelers are increasingly con- cerned with the consistency of their models with the results of behavioral studies. Because of the increased interaction of BAR and “Invited Paper. +We appreciate helpful comments by Jim Jiambalvo, participants in the European ‘Accounting Association's 1987 annual meeting, and two anonymous reviewers. +For examples of experimental economics research see Smith, Schatzberg, and Waller 11987) and the references ctted therein. An early paper combining analytic and behavioral approaches was Hilton {1980|. The concern of analytic researchers with behavioral research fs summarized by Baiman (1982, p. 157): “(Tihe degree to which we can have confidence that our normative suggestions will lead to the desired ‘outcomes depends upon how well we understand (can predict) how people behave within an organizational setting.” 15 Copyright © 2001. All Rights Reseved. 76 David Burgstahler and Gary L. Sundem other types of accounting research, it is an appropriate time to present an assessment of BAR from the perspective of nonbehavioral zncoumt. ing researchers, The purpose of this paper is to review the development of BAR over the past two decades. We were invited to write this paper because we have not been behavioral researchers, We profess interest in BAR, but neither expertise nor objectivity. Our perspectives and biases are probably different from those of behavioral accounting researchers, and therefore our assessment of the research is probably different. We will ty to point out our biases to help readers interpret our comments, which represent our subjective view of BAR. BAR has been reviewed by behavioral accounting researchers (for rrample Hofstedt [1976], Dyckman, et al. {1978}, Libby and Lewis 11978, 1982], and Snowball 11986), non-accounting behavioral Tgarchers (such as Einhorn 1976] and Weick {Swieringa and Weick, not usually forced to look as closely at the product as are the producers, Anis Paper chronicles our attempt to critically examine the product from a consumer's perspective. We will define BAR, identify character- isties of BAR and their changes over the last 20 years, point out inherent difficulties in BAR, assess the progress made during the last two decades, and point out continuing weaknesses that limit the contributions of BAR, all from our viewpoint rather than that of a behavioral researcher. Behavioral Accounting Research (BAR) Defined Our definition of BAR, which certainly influences our evaluation, may differ from that of many behavioral accounting researchers. We Percelve BAR from the perspective of how it fits into an information economics view of the world. Scientific research, such as that in physics or chemistry (or even in most social sciences such as psychology), seeks to determine underlying truths. But accounting research does not seek “truth.” Instead, it attempts to identify an economic commodity (accounting information) for which there is demand, either free-market demand or collective (soctal) demand, at a price that allows an economic return to the factors of production It is akin to the research undertaken by A common analogy compares accounting information toa map. Like a/mapvaocouvan Snformation ia judged on how wed it htipes unet soceesee 8a ioals, not of how “rue” a picture t provides. For an automobile driver a wap shone oe tight be more valuable than an arial photograph, cven though it oho co ones Picture. Accounting research determines the demand function for varioue ype cr maps (information syatema, develope new maps, and discovers more cine eo producing any type of map. Copyright © 2001. All Rights Reseved. ‘The Evolution of Research 7 business firms (for example, automobile companies or electronics companies) to aid them in producing a competitive product. We will not debate in this paper whether accounting information is an ‘economic commodity. We accept the fact that it is, and our view of BAR reflects that perspective. ‘Accounting research, including BAR, has value if it enhances the production of accounting information. Three types of accounting research can do so: 1) market research, which identifies the current or potential demand for information, 2) new product development research, which explores new ways of measuring, transforming, or reporting data, or 3) production technology research, which develops efficient production methods for information. MARKET RESEARCH Market research is generally behavioral in nature. The demands for information, whether by individuals or organizations, depend on the characteristics of the people involved, their goals, and their opportu- nity sets. At least four types of research studies address information demand: 1, Some empirical researchers observe the information actually used in various situations, from which they induce a model of demand. For example, interviews or questionnaires could be used to ask profit center managers what information is used to motivate subordinates. Alternatively, field research could be used to directly observe the actual data used in such evalua- tions. 2. Laboratory experimenters control most variables, and they record the use of information as specific experimental vartables are manipulated. For example, loan officers might be given limited information and asked to approve or deny a loan application. Information use could be inferred from the deci- sions they make or a method such as protocol analysis could be used to more directly measure the use of information. 3. Researchers often observe the aggregate consequences of decisions, from which they infer information use. Studies of capital market reactions to disclosure of information are an example. 4. Analytic researchers assume behavioral characteristics, and then they derive the demand for information. Examples include statistical models for cost variance investigations and models of audit risk. The underlying objective of all four types of market research is to determine how people, individually or collectively, use information in making decisions. This is inherently a behavioral issue. However, Copyright © 2001. All Rights Reseved. 78 David Burgstahler and Gary L. Sundem most accountants would classify only the first two types as BAR. We will do the same to limit the scope of this paper. These two types of research account for the majority of BAR, and they are important ingredients in determining the demand function for accounting information. PRODUCT DEVELOPMENT RESEARCH Some BAR is new product development research. Most new accounting information products involve primarily a priori research (or, more accurately, creative thought), not BAR. Jjiri's (1982] triple-entry bookkeeping is a recent example. However, new products are often sought because an unfilled demand is observed or because a new production opportunity is created. For example, a few years ago the demand for inflation adjusted accounting information was increasing. BAR was undertaken to test the ability of alternative sets of information (i.e., alternative new products) to meet this demand. Laboratories served as test markets to assess general-price-level- adjusted statements vis-a-vis current-cost statements. and the use of cash flow data instead of (or in addition to) accrual data was explored. Today, the growth of multinational firms, dramatic changes in manufacturing environments, and use of just-in-time inventory methods are creating new information demands. BAR will play a large role in identifying new demands and testing products that might fill these emerging demands. Other new products arise because of advances in production technology. For example, the ability of managers to easily, quickly, and economically access data bases has created vast opportunities. Researchers in management information systems are exploring the benefits of such direct access systems.3 At the same time. the Importance of security to information demand is being examined. Without BAR, the introduction of new products would be more risky and hence more costly. PRODUCTION TECHNOLOGY RESEARCH Production technology research, defined as improving production methods for currently available accounting information, is primarily nonbehavioral. A large segment of such research focuses on technical aspects of computerized information systems.4 However, when people are part of the production process, BAR may be useful in refining the interface between people and the new technology. For example, the See, for example, Malone, et al. [1987] and references cited therein. Results of such research can be found in many information systems textbooks, such as OBrien (1988), Copyright © 2001. All Rights Reseved. ‘The Evolution of Research 79 creation of expert systems to aid in the audit judgment process has a large BAR component. SCOPE OF THIS PAPER Although the preceding discussion suggests that most of the accounting literature could be covered under a broad definition of BAR, we limited the scope of this review to papers dealing directly with behavior. Papers assuming behavior (e.g., analytic modeling) and those looking only at consequences of decisions (e.g. capital market studies) are not generally regarded as behavioral, and are thus omittedS Examinations of the social impact of accounting, though regarded by some as behavioral, are also omitted because they are concerned with aggregate consequences, not behavior itself. We are mainly concerned with papers that include efther the methods or theorles of the basic behavioral sciences, primarily psychology. soctology, and organizational behavior. We focus more on empirical BAR than on theoretical analyses, because our background better equips us to assess developments in empirical research methods than in theory developments Behavioral accounting researchers have published their findings in a large number of journals, both in accounting and in the behavioral sciences. Our review will be restricted to publications in three accounting journals. Articles in The Accounting Review (AR) and the Journal of Accounting Research (JAR) will represent BAR presented to. the broad set of accounting researchers, and those in Accounting, Organizations, and Society (AOS) will represent communication primarily among behavioral researchers. Often the most significant advances in a field appear in a more specialized journal, such as AOS, before they are given a broader hearing in more general research journals. FOUNDATION FOR EVALUATION Most of our observations in this paper are subjective. They are based on our general reading of the literature: they are not derived from a systematic review of the literature. This is intentional. The paper's purpose is to assess our view of BAR, as possibly representative of the view of other nonbehavioral accounting researchers. However, to organize our thoughts we classified the BAR of the past two decades on several dimensions. It seems useful to share our classifications, since Such research is discussed in this introduction because our interpretation of BAR is, necessarily influenced by its relationship with analytic modeling and capital market research. Our lack of knowledge of competing theories from psychology. sociology. and other basic behavioral disciplines limits our ability to assess purely theoretical contributions. Copyright © 2001. All Rights Reseved. 80 David Burgstahler and Gary L. Sundem this may also reveal some of our biases. The summarized results are in Tables 1 and 2, and the details are provided in an appendix. These data are not intended to directly support our observations, but are presented primarily as potentially interesting descriptive data. They helped us to organize our thoughts, but are not “evidence” to support our conclusions. Panel A of Table 1 shows a growth in BAR over the past 20 years, both in absolute numbers of articles and in percentage of all papers, in AR and JAR. There was a year-by-year growth in BAR until the mid-1970s, followed by a slight drop-off, then an increase in the early 1980s, followed by another decline in the last two years. The first decline coincided with the introduction of AOS in 1976 and did not signal a drop in total BAR. The recent decline to 13 percent of the published articles in AR and JAR in 1986 and eight percent in 1987 is. due to the sharp drop of BAR papers in JAR. It is also interesting to note from Table 2 that the number of BAR papers in AOS has been declining. The mix of topics in BAR has changed considerably in the last 20 years. AR and JAR have experienced an extraordinary increase in audit articles since 1978, from under ten percent of the BAR papers to over 40 percent. The biggest decrease came in financial accounting, from 39 percent to 13 percent. AOS has a much smaller proportion of audit articles and a higher proportion in management accounting. The trend toward empirical BAR in AR and JAR is obvious, with 97 percent of the BAR being empirical in the last five years, up from 74 percent in 1968-72. Since 1978 very few nonempirical papers have been published. The same trend is evident for AOS, but even in the last five-year period over one-fifth of the AOS BAR papers were nonem- pirical. Panel B of Tables 1 and 2 contains data on the empirical articles only. The first two columns show a slight trend toward multiperson research in all three journals, with AOS having three times as much multiperson research as AR and JAR. Clearly, the organizational focus of AOS influences the types of BAR it publishes compared to AR and JAR. We expected a trend toward using fewer students and more real decision makers as subjects in BAR. This was only marginally confirmed. The use of student subjects peaked in the mid-1970s and has always been greater in AR and JAR than in AOS. While overall statistics do not show this, we observed a better match of subjects to research topics today than 10 or 20 years ago. Students continue to be used, but only in studies of basic human information processing, where expertise or organizational context are not major influences. Significant changes in the research setting have also occurred in the past 20 years. In AR and JAR, experimental studies have increased Copyright © 2001. All Rights Reseved. 81 ‘The Evolution of Research (or) ca we vo z9-c96t ze-ez6t LL-e161 z2-8961 joao maps Womg «toed | pond ones Baes worse Aquo oreaeoy ondurg Jo sadtwary sg TouVE (so) te) so 8999 e's 28-6861 (so) te) #0 869s eet ze-eer (vo) (807 v0 | ott srt Luter (0) (60 yo OT wit z2-e961 BRO Ter wav wemeg | pond SdOTWad UVIA-FALA YO (NOLLUOdOUd) UVAK Wad SHIOLLUV JO UAGWON SOVAAAV—HOAVISTY ONILNAOIOV 40 TWNUNOL GNV AMGIATA ONILNNOIOV AHL NI ONDIVAddY HOYVASTY ONTLNNODOV TVAOIAVHGE dO NOLLVOLAISSV10 1 aTavL Copyright © 2001. All Rights Reseved. David Burgstabler and Gary L. Sundem ‘Q261 Uh paystignd ys14) sem SOV :poliad seaK-OML, (ey) ue) (ee) wo) ue) an @s) ter) + © ze ve ve 0 v2 ot zs ve Le-c96t tgs’) (or) (6z) (60) ue) au (we) (89°) 9 we zw zt 2% zo 9 #0 ve vs ze-ecst (es) (so) (tr) (so) (62) (or) (ar) (es) er % os so oF so sh st ov ss euL-ou6t Tey Ram amg png woupedsg “og Teun Auspmg Woeng omg —ponied «ty wep Pista “nowy BER omg TomaPH Fares wine aaoTaTS roo yoreseoy reopridaag Jo medpeuy 2a found Ze) (62) tor) (uy (6r) wr) ¥% 0 ot vr os ot vu 29-ce6t (Ze) (es) r) cn) (or) (er) a9 oe ee oT 99 9% ver zeae (er) as) (bo) a) (se") (er) 06 v6 sy oz v9 se ver 21-9161 TeopydarsuON roprdag PRO yok Tepaieren ToOUwTT ‘woder popes oO Toy srodeg feopagog MV :V [ued SdoNiad UVIA-GAM UO (NOLLUOdOUd) UVAA Ud SATOLLUV AO HAGWOAN AOVAAAV—ALAIOOS CNV ‘SNOILVZINVOUO ‘ONILNNODOV NI ONTAVAddV HOUVASTA ONILNNOOOV TVAOIAVHAE JO NOILVOIAISSV1O ta1ave Copyright © 2001. All Rights Reseved. ‘The Evolution of Research 83 from about 60 percent of empirical BAR in 1968-72 to over 80 percent in 1983-87. In 1968-72, most of the laboratory studies in AR and JAR were conducted in an experimenter created environment (e.g. a classroom or a specially designed behavioral lab), but by 1983-87 over nalf were either conducted in the subject's work environment or mailed to the subject” The increased use of on-site and mailed experiments probably reflects a recognition that people behave differently when taken out of their usual decision environment. About three-fourths of AOS experimental studies were conducted in an experimenter created environment. Field studies have been a decreasing percentage of articles in AR and JAR, but an increasing percentage in AOS. In 1983-87, the proportion of field studies was about nine times greater in AOS than in AR and JAR. Survey articles have decreased in all three journals, but they still constitute over one-third of the empirical BAR papers in AOS while comprising less than half that proportion in AR and JAR. In summary, the BAR in AR and JAR is primarily empirical and experimental, while AOS publishes more nonempirical and nonexperimental studies. The last two columns show the median number of references in the papers, and the percentage of the references that cited accounting literature. These statistics proved to be less revealing than anticipated. The median number of references increased during the 1970s, but has been relatively stable since. The percentage of accounting references has been relatively stable throughout the two decades. It seems that many early BAR papers cited accounting references because the research was not based on behavioral theories from other disciplines. Later BAR papers cited accounting papers that summarize behavioral literature in an accounting context. In between, there was a larger proportion of citations from non-accounting behavioral literature. Evolution of Behavioral Research The simple model in Figure 1 represents our overall assessment of how BAR has evolved during the last two decades. First, behavioral research methods were used for accounting issues without adoption of any underlying behavioral theories. An example is research on whether student subjects make different decisions with FIFO reports than with 7We operationally defined laboratory studies as those in which the researcher controls and manipulates experimental variables whether on premises controlled by the researcher (lab). the business premises of the subjects with the researchers in attendance (on-site lab), or through an instrument completed at the subjects’ convenience (mailed lab). Studies where subjects filled out questionnaires that simply revealed attitudes and bellefs (which then might be compared to characteristics of other subjects) were classified as surveys. To be a lab study. a questionnaire must include some manipulation, such as presentation of a scenario about which the subject makes a decision or judgment. Copyright © 2001. All Rights Reseved.

Vous aimerez peut-être aussi