Dale Buckmaster UNIVERSITY OF DELAWARE
and
Kok-Foo Theang ERNST & YOUNG

AN EXPLORATORY STUDY OF EARLY EMPIRICISM IN U.S. ACCOUNTING LITERATURE

Abstract: Little or nothing is said of empiricism in U.S. accounting literature during the first half of the twentieth century in accounting history literature. The objectives of this study are threefold: (1) to determine if an empirical accounting literature existed prior to 1950; (2) to determine if pre-1950 empiricism was extensive enough and substantive enough to have influenced the development of accounting thought; and (3) to compare pre-1950 empirical work with contemporary academic research. It is concluded that empirics were common prior to 1950 from examining a sample (approximately forty percent) of volumes (clusters) of The Accounting Review, The Journal of Accountancy, Michigan Business Review, The American Accountant and the N.A.C.A. Bulletin. One hundred eighteen articles and eleven books and monographs are classified as “empirical” in this study.

A sample was drawn from the books and monographs and classified using several recently developed taxonomies of accounting literature. This sample included works in several accounting specializations and also included works by both academic and non-academic authors from all of the journals. The empirics found in most of the studies were essential to the studies and not peripheral. However, inferential statistics were rarely used and the designs of the studies were very primitive. The sample yielded no evidence of a transition to a contemporary hypothetico-deductive paradigm. While not common, there were attempts at “positivism.” However, the authors of most financial accounting studies were concerned with normative theory. Empiricism was extensive enough and substantive enough to have had considerable influence on normative theorists and the development of the accounting literature of the period.

This research was made possible, in part, by a grant from the KPMG Peat Marwick Faculty Development Fund at the University of Delaware. The authors wish to thank Barbara Merino, Richard Vangermeersch, Araya Debessay, Scott Jones, Fred Stiner, and three anonymous referees for their valuable comments.

Most students of accounting research have some familiarity with the normative, non-empirical work of American accounting authors of the first half of the twentieth century. Paton, Littleton, Sweeney, and others are well known. However, little is known about the empirical literature of the period. Consider, for example, when Ball [1971] reported the bibliography of empirical work in accounting prepared at the University of Chicago, the bibliography contained only twelve entries dated prior to 1950. Even though most accounting historians recognize that Ball’s list is far from being complete, early empiricism has been ignored in accounting history literature. There seems to be no substantive discussion or investigation of the contribution of empiricism to the development of accounting thought in the United States. Thus, it is risky to make general statements as to the extent of early empiricism, the types of literature that utilized real data, or the relationship of the early literature to the contemporary empiricism that dominates modern academic accounting literature. This lack of knowledge about early empiricism provides unnecessary bounds on theories of the development of accounting thought.

This paper partially fills this void in accounting history literature by providing an exploratory examination of the nature of pre-1950 empiricism. This is accomplished by identifying and reviewing a sample of the empirical accounting literature published in the United States prior to 1950. Also, some characteristics of pre -1950 empiricism are compared with contemporary academic research. The objectives of this exercise are to provide evidence of the role of empiricism in the early literature and of the relationship of the early work to contemporary academic accounting literature. The relationship is examined by classifying the sample according to several taxonomies applied to accounting literature. The taxonomies are Brown and Vasarhelyi’s [1985] Mode of Reasoning, Research Method, and School of Thought taxonomies; Vasarhelyi, et al. [1988] Foundation Discipline, Accounting Area, and Information Sources taxonomies; Watts and Zimmerman’s [1986] four-class taxonomy of the transition of accounting re-search; and Antil’s [1989] two-class transition taxonomy. Brown, Gardner, and Vasarhelyi [1987] also use the Research Method and Foundation Discipline taxonomies.

Several other taxonomies of the development of accounting thought are found in the literature. In this research project, the authors would have liked to compare the sample with classifications suggested in these taxonomies. Unfortunately, the classifications are not compatible with the data obtained from the study. Furthermore the sample size of observations related to the taxons within a taxonomy is too small to be useful. For example, the sample of theory-based financial accounting studies used here is too small to support an evaluation of Flamholtz’s [1977] pre-1950 classifications and the data are not related to Previts and Merino’s [1979] discussions of pre-1950 accounting literature.

The sections describing the classifications of the sample are followed by an overview of the role of empiricism in pre-1950 accounting literature. The paper then speculates on some of the determinants of the development of accounting literature in the section preceding the “Summary.”

A much more extensive and varied literature than anticipated was discovered in the process of selecting the sample upon which this paper is based. Work falling within this study’s definition of empiricism was common and related to almost all areas of accounting. Early empiricism performed an information function and was extensive and substantive enough to have provided a basis for induction for much of the normative theory of the time. However, no beginning of the contemporary hypothetico-deduc-tive academic accounting literature was observed.

Comparison of the sample used in this research with contemporary literature suggests several variables that might affect the development of accounting literature. Data processing technology, the academic environment, the state of the art of other academic disciplines, the background and training of accounting authors, and the institutional environment of accounting are thought to be determinants of the form and content of accounting literature that might explain, at least partially, the differences between the sampled research and contemporary academic accounting literature.

OPERATIONAL DEFINITION OF “EMPIRICAL”

In this paper, “empiricism” is examined in the context that the typical contemporary academic accountant might use the word. There is no intent here to consider or confront any formal philosophical concepts of empiricism. Most of the pre-1950 works included in Ball’s index included some sort of tabulated real data or reported summary statistics extracted from studies; however, the Merriam-Webster Dictionary defines “empirical” more generally as “depending or based on experience or observation.” Ball’s definition, suggested by the absence of case studies and similar descriptive work, is unacceptably narrow. Yet the substantive definition provided by the dictionary is too broad to provide the bounds of what is normally considered empiricism by contemporary researchers. Therefore, for purposes of this paper, empirical research is operationally interpreted to include any work that uses, identifies, and reports observations of real events. Both descriptive studies and inferential studies properly fall within this definition.

The primary variation of the work identified by this study as empirical from those works listed by Ball is the admission of case studies. The absence of case studies from Ball’s list may be intentional in that he does not consider case studies as empiricism or, given the incompleteness of his list, the absence of case studies may be coincidence. Both explanations are plausible. In this paper, the authors take the position that case studies are properly classified as empirical. The authors of this paper refer to Brown, et al. [1987] and Vasarhelyi, et al. [1988] for support for this position. Both sets of these researchers include the taxon, “Empirical-Case,” in their “Research Method.

Judgment was necessary in classifying many articles as “empirical” or “non-empirical” even with our operational definition. An example of this type of article that was excluded, but might be classified as “empirical” is Bowles’ [1933] discussion of trends in statement presentation of “non-current gains and losses” and the proper treatment of these items. Another example is Danford’s [1931] description of variance analysis at ARMCO. Bias in the identification of sampling units is towards omitting appropriate observations rather than the converse.

THE SAMPLE

The publications examined are the twelve listed by Ball plus twenty-seven randomly selected publications. The theoretical population of interest is work coming within the operational definition of empiricism that was published in the United States prior to 1950. The only available list of such published empirical work is Ball’s “Index of Empirical Work in Accounting.” Recognition that this list is incomplete provided the original stimulus for this study. The original intention of the authors was to complete the list and review the identified work. It quickly became obvious that the quantity of early empiricism is much more extensive than had been anticipated. Completing the list and reviewing all entries would be much too costly since the text of each candidate for the empirical classification had to be examined and, if the taxonomies were to be utilized, perform a content analysis of each of the works. Thus, sampling was used in order to maximize the descriptive power of this study.

The frame (operationally defined population) from which the sample is drawn is pre-1950 issues of The Accounting Review, Journal of Accountancy, The American Accountant, N.A.C.A. Bulletin, and Michigan Business Studies. A two-stage sampling plan with cluster sampling comprising the first stage is the standard approach when there is no existing list of elements within the frame.1 Bound volumes of the journals making up the frame provide natural clusters, thus the articles and literature reviews were examined in randomly selected pre-1950 volumes of the journals making up the frame for accounting publications coming within the definition of empiricism. When a published work not published in one of the five journals, but reviewed in one of the volumes selected, appeared to be a likely empirical work the authors of this study obtained a copy of the work and examined it for possible classification as “empirical.

The clusters (volumes) amounted to about forty percent of the total pre-1950 volumes of the series examined. One hundred eighteen articles and eleven books or monographs have been identified from the clusters as being eligible for inclusion in the sample. The primary purpose in selecting clusters was to identify possible sampling units at reasonable cost without disturbing probability relationships. Once the partial list was complete, twenty-seven works were randomly selected from the list of articles and books. Appendix I is a chronological, annotated bibliography of the thirty-nine works (the twelve listed by Ball plus the twenty-seven randomly selected works) included in the study. A potential bias is intro-duced into the sample with the inclusion of the twelve entries from Ball’s bibliography. These twelve entries were included on the contingency that they might, somehow, be significantly different from the randomly selected entries. Inspection of the two classes of works reveals no discernible differences except for the absence of case studies in Ball’s bibliography; therefore, no formal analysis is performed or reported. Other than the potential sampling bias just described, the authors are not aware of any risk or bias other than those normally associated with sampling.

The sample selection process reveals a much more extensive and varied literature than anticipated. There is a much larger number of sources that report actual observations than is implied by accounting history literature. This early empirical literature provided inductive possibilities for the normative theorists of the period. The research design used here does not, however, provide any evidence of specific use of reported empirics in inducing normative positions. To gain such insight, the authors would have to review the early literature on the particular problem and place the empirical work in the temporal context of that literature to provide evidence of the contribution of the empirical work to the normative theory. This is beyond the scope of this study. However, this study does indicate some possibilities for such future research.

Another interesting question arises concerning trends in methodology and other taxons. Again, the research design does not permit any useful conjecture about trends. To do so would require stratifying the sample by period and drawing a much larger sample. Enough is learned in this study, however, to permit an interested researcher to design a study within reasonable cost constraints to search for trends in methodology or other characteristics.
THE VASARHELYI, BAO AND BECK TAXONOMIES Accounting Area

Table 1 describes the sample by number of works in each accounting area in each journal. The classification by accounting area provides the first application of one of the literature taxonomies. Table 1 contains a modified version of Vasarhelyi, et al. Accounting Area taxonomy. The literature sample classified by Vasarhelyi, et al. was selected so that education and professional development articles were a null set, thus there was no need for that particular taxon. Yet the sample used here suggests that education and professional development articles represented a substantial portion (20% of the sample) of early empiricism, so an Education and Professional Development taxon was added.

Vasarhelyi, et al. do not provide any statistics describing their sample by accounting area so no comparisons are made with contemporary literature, but two interesting characteristics are suggested by the classification of the sample. First, real data were used by authors of articles in almost all accounting areas. Second, there was a large proportion of articles on non-accounting problems in the accounting literature. The subjects of four of the articles in the “other” classification are management topics, one is on proper organization of natural monopolies, one was a call for regulation of public holding companies, and the only accounting article in the “other” classification deals with governmental accounting. Thus six (15% of the total sample or 23% of the sample drawn from accounting journals) of the seven articles in the “other” classification have “non-accounting” subjects.

Research Method

Classification of the primary research method for the publications is reported in Table 2. Four of the eight “Archival-Primary” classifications used financial statements as the primary information source and the other article derived descriptive statistics from university catalogues. The works in the “Archival-Secondary” taxon were either normative arguments that abstracted data from other publications to support a position or reviews of studies that were based on descriptive studies.

Vasarhelyi, et al. [1988, p. 57] indicate that the three main Research Method taxons for contemporary academic research are Analytical-Internal Logic (51%), Archival-Primary (22%), and Empirical-Laboratory (10%). The sample definition used here restricts the sample to works that report real observations and thus limits comparability with Vasarhelyi, et al.; particularly since the definition eliminates most studies where internal logic was the primary research method. Yet, if the differences in the sample definition is recognized, comparing the two samples provides evidence of the differences between contemporary and early empiricism.

There are several striking differences in the two samples. “Analytical-Simulation,” a taxon that accounted for sixty-four studies in the Vasarhelyi, et al. study is, of course, a research approach only feasible after the advent of computers, thus we would not expect our sample to contain any works using this approach even if such studies came within our definition of “empirical.” There are no “Empirical-Laboratory” studies in this sample, yet this taxon contained ten percent of the Vasarhelyi, et al. sample. Other notable differences are the “Archival-Secondary,” “Empirical-Case,” and “Opinion-Survey” taxons. Several of the sampled works are classified as “Archival-Secondary” were review articles that abstracted data from other publications and three of

Buckmaster and Theang: Exploratory Study of Early Empiricism
Table 2 Classification by Research Method

Modern
Academic
Taxonomy Sample of Early Work
Empiricism Reported
by
Vasarhelyi,
et al.
Number %of
of Works Works % of Works
Analytical-internal logic -0- -0-% 51%
Archival-primary 8 21% 22%
Archival-secondary 8 21% not reported
Empirical-Case 6 15% not reported
Empirical-field 2 5% not reported
Empirical-laboratory -0- -0-% 10%
Opinion-survey 9 23% not reported
Mixed 6 15%

39

100%

Classification of Mixed Research Methods — Early Empiricism

Internal Logic and Archival-primary Internal Logic and Opinion-survey Archival-primary and Opinion-survey
Total

the articles were no more than abstracts of academic studies published in non-academic journals. Review articles are still common in academic literature, but the abstract-type articles are not. Elimination of the abstract-type articles makes the early and contemporary proportions of “Archival-Secondary” more comparable. The greater frequency of “Empirical-Case” and “Opinion-Survey” articles in this sample may be accounted for by the greater participation of “non-academic” authors in the sample of early empiricism and by the contempt of contemporary editors of academic journals and dissertation chairmen for these research methods.

Table 3 describes the professional affiliation of authors in each of the journals. The journals in the sample cannot be classified as academic and non-academic if the criterion is the professional affiliation of the authors. The professional affiliation of authors in The Accounting Review, The American Accountant, and the Journal of Accountancy is similar. Although we provide no comparative statistics, few will dispute that the appearance of a non-academic author in a contemporary issue of The Accounting Review is indeed a rarity. Those non-academic authors publishing in The Accounting Review and other accounting journals of the pre-1950 period wrote about things they knew — the operations and methods of firms with which they were associated.

The presence of non-academic authors in the journal articles sampled in this study, partially accounts for the greater proportion of case studies in the sample, but the question of most significance in explaining the development of academic literature is why non-academic authors were present in academic journals then and not now. Of course, non-academics were as competent with the dominant methodology as academics in the early period and probably had more access to data. On the other hand, few non-academics have the research capabilities required to put together a paper for a contemporary academic journal.

However, the authors of this paper propose the hypothesis that increased competition for journal space by academics would create an effective barrier to would-be non-academic authors even if they had the requisite research skills. In general, the divergence of the academic and practice cultures over the years has resulted in the disappearance of common goals. Also, what is currently considered empirical accounting research lies outside the domain of accounting practice. Nissley’s [1929] article on the Bureau of Placements is interesting within the context of the recent Bricker and Previts [1990] article on the current practitioner/academic schism. Bricker and Previts identify an earlier schism partly precipitated by the Bureau of Placements’ concentrating primarily on liberal arts colleges that did not have accounting curricula. Nissley’s article is an obvious attempt to placate academics.
School of Thought

The classification of the sample using the “School of Thought” taxonomy produced twenty-three observations in the “Accounting Theory” taxon and one in the “Institutional.” The “other” taxon contains fifteen observations despite very liberal interpretation of the content of the “Accounting Theory” taxon. The authors of this paper interpreted the Accounting Theory taxon to include works that had some connection with financial accounting theory, managerial accounting theory, and auditing theory.

Table 4 School of Thought

Taxon
Behavioral (All Taxons) Statistical Modeling (All Taxons) Accounting Theory Accounting History Institutional Other
Total

Modern
Academic
Reported by
Sample of Vasarhelyi,
Early Empiricism et al.
Number
of Works Percentage Percentage
-0- -0-% 18%
-0- -0-% 37%
23 56% 27%
-0- -0-% 3%
1 3% 9%
15 41% 6%
100%
39
100%

Composition of the “Other” Taxon — Sample of Early Empiricism
Number of Works

Institutional (but not accounting or accounting
regulatory institutions) Management Professional Development and Education
Total

3 6 6
15

Explanations for the differences for some of the taxons ap-pear obvious. Most of the behavioral statistical modeling tech-niques were not developed or were developed late in the period from which the “Early Empiricism” sample was drawn. Vasarhelyi, et al. define the “Institutional” taxon narrowly in that it applies primarily to accounting regulatory agency literature and there were fewer accounting regulatory agencies and the primary accounting regulatory groups currently existing did not exist during the first twenty-seven years of the forty-three year period from which the sample was drawn. Education articles are outside the bounds of Vasarhelyi, et al. operational definition of their sample. The frame for their sample was articles in the main sections of The Accounting Review, The Journal of Accounting Research, Accounting, Organizations, and Society, Journal of Accounting and Economics, and Auditing: A Journal of Theory and Practice for the period 1963 through 1984. The Accounting Review had moved education articles to the “Education” section by 1963 and the other journals have specific missions that exclude education.

All of the early accounting history articles which the researchers reviewed used a narrative style without use of statistical data. Thus, the absence of “history” from our sample is expected. Also, there could be a body of non-empirical accounting literature published prior to 1950 that deals with the Committee on Accounting Procedure and the SEC, thus properly classifiable as “Institutional” in the “School of Thought” taxonomy.

There is no assurance that accounting researchers would have used the approaches indicated by the “Behavioral” and “Statistical Modeling” taxons even if the methods had existed prior to 1950. The basic methods of classical statistical inference had been or were developed in the early twentieth century. Yet none of the authors in the sample reported testing for statistical significance even though it would have been appropriate in several of the studies. Two of the more obvious explanations for the absence of the use of inferential statistics for theoretical predictions are that accountants (academic and practicing) were not trained in classical statistics and editors avoided reporting statistical tests of predictions. Both explanations probably frequently operated jointly.
Mode of Reasoning

Table 5 reports the application of the “Mode of Reasoning” taxonomy to our sample and the corresponding percentages for the Vasarhelyi, et al. sample where such percentages have been reported. The “Mixed” taxon was, of course, used when two “modes” were considered of approximately equal importance for the publication.

The presentation of data in accounting literature prior to 1950 was generally for one of two reasons. The most common reason was to describe the accounting treatment of some element of financial statements. The next most common reason was to provide descriptive statistics to support a theoretical position. Yet even those works that were primarily presentations of descriptive statis-

Table 5 Mode of Reasoning

Taxon
Descriptive Statistics
Correlation
Quantitative-Analytical
Qualitative
Mixed
Total

Modern
Academic
Reported by
Sample of Vasarhelyi,
Early Empiricism et al.
Number
of Works Percentage Percentage
15 38% 11%
1 3% Not reported
-0- -0-% 28%
13 33% 36%
10 26% Not reported
39
100%

Mixed Modes of Reasoning

Descriptive Statistics and Qualitative Descriptive Statistics and Correlation
Total

Number of Works
9 1
10

tics contained a discussion of the theoretical merit of the various treatments or opinions.

Table 6 describes the distribution of the data evaluation and presentation techniques. The New York State Society Committee on Inventory Verification [1930], Magor [1942], Williams [1947], and Child [1949] did not present any data or discuss any summary statistics. With the exception of Warshaw [1924], those that used graphs or basic statistical tools also provided tables and tabulations of raw data. Warshaw only used graphs of inventory and price movements to illustrate his income smoothing argument for base stock inventory methods. Those that used basic statistical indicators were: (1) Burrell [1929] reported the correlation coefficients of his accounting aptitude test and standardized general aptitude tests with the final grades in the first accounting course of his test groups of college freshmen; (2) the review of the Ohio State study of the operating ratios of transient hotels in The American Accountant [1931] reported and discussed the meaning of the standard error of the estimate as well as the means the ratios; (3) Wood, Traxler, and Nissley [1948] disclosed means and ranges as well as graphs; and (4) Jones [1949] used means, ranges, standard deviations, and graphs. None of the studies had a statistical test for the significance of variables.
Table 6

Data Evaluation and Presentation Techniques used in the Sample of Early Empiricism

Techniques Used Number of Studies
No data presented or discussed 4
Verbal description and indication of selected
summary indicators only 15
Tables and/or tabulations of Raw Data 19
Some basic statistical indicators 4
Graphs 2
Total 44*
*Some studies used more than one technique.

The use of data and its presentation (or lack thereof) was clearly adequate for the objectives of a number of the publications. For example, additional data would have contributed little to the case studies of company systems and procedures. Also, Mitchell’s case studies of financial statements [November, 1906; December, 1906; June, 1907] and Pierson’s [1907] case study of the Philadelphia Gas Works were complete and would not have benefitted from more elaborate design. Stockwell’s [1910] and Warshaw’s [1924] data supplement their articles adequately. However, when one moves from case studies, many of the publications would have benefited from even the most primitive design considerations and/or data disclosures. For example, Williams [1947] provides examples of unusual financial statement treatments, but he provides no indication of the number of statements he examined or of the frequencies with which he observed specific treatments. The deficiencies of Williams’ paper were typical of several of the publications that utilized more than a small number of observations.
Foundation Discipline

Table 7 contains the “Foundation Discipline” classification and the corresponding proportions reported by Vasarhelyi, et al.

The comparative numbers are surprising, not for the differences, rather for the similar proportions of economics/finance-based articles. If we drop education articles which were excluded from the Vasarhelyi, et al. sample, then the “Economics/Finance” taxon contains 23% of the Early Empiricism sample compared to 18% for the contemporary academic sample.
Table 7 Foundation Discipline

Modern Academic
Sample of Early Empiricism Vasarhelyi, et al.
Number
Taxon of Works Percentage Percentage
Psychology -0- -0-% 12%
Economics, Finance 6 15% 18%
Engineering,
Communications 1 3% Not Reported
Mathematics/Decision/
Game Theory -0- -0-% 6%
Accounting 23 59% 44%
Management 6 15% Not Reported
Education 3 8% Not Investigated
Total 39 100%

The early articles classified as “Economics/Finance” were those that were obviously relying on those foundation disciplines. The 15% sampled in this research may be an understatement. This can be and probably should be tested by drawing a sample of economics and finance literature and making a comparison. However, such a test is beyond the scope of the objectives of this paper and, in the authors’ opinion, should be approached as a separate study with the accounting and economics sample being drawn from the body of literature existing in the first half of the twentieth century and not limited to empirical based studies.
The “Management” based articles dealt with management problems, not accounting problems. Thus accounting authors were not drawing on management methods to investigate accounting problems. When this is considered, “Accounting” based works are 71% of the portion of the sample compared to 44% in the Vasarhelyi, et al. sample. The accounting researchers of the early period were much less likely to draw upon the research methodologies of other disciplines to solve accounting research problems than contemporary researchers. The greater likelihood of contemporary researchers drawing on methods from other disciplines must be attributed, at least in part, to the relatively recent development of the borrowed methods.

WATTS AND ZIMMERMAN’S TAXONOMY OF EVOLUTION

Watts and Zimmerman [1986] provide the following periods of the evolution of financial accounting theory (pp. 4-7):
Late nineteenth and early Researchers observed practices
twentieth century and developed pedagogical clas-
sification rules.
After the Securities Acts Normative theory development,
until the later 1960s
Late 1960s until mid-1970s Tests of existing normative
theories and the relatinships of accunting numbers and economic variables.
Mid-1970s until now Positive theory development.

Watts and Zimmerman’s first two classifications do not hold for works sampled for this study. For example, the three analyses by Mitchell [November, 1906; December, 1906; June, 1907] all compared financial statements with theoretically correct standards. Five of the nine entries prior to the passage of the Securities Acts that dealt with financial accounting gave substantive consideration to theoretical aspects of the problem being considered. Conversely, six of the eleven papers published after the Securities Acts and concerned with financial accounting were descriptions of practices without substantive consideration of theory. Watts and Zimmerman’s classification scheme would have been more precise if their first two eras had been combined and they had observed that both normative theory and observation of existing practices were common in the literature during the first half of the twentieth century. Obviously their fourth “era” description is not valid either. Market studies continue to flourish and historical and non-positivist studies may be increasing in importance in the evolution of financial accounting theory. Therefore the authors suggest that the fourth classification reflects either extreme naivete or is a clear example of the “self-interest” assumption that is basic to Watts and Zimmerman’s work. The authors suggest further that Watts and Zimmerman are not naive.

Two authors (Hornberger [1929] and Clendenin [1941]) did investigate why managers made certain accounting choices. Previts and Merino [1978] also recognize early positivism (or explanatory theory) in the surveys conducted by the N.A.C.A. research staff in the late 1930s and early 1940s. They note a 1940 movement in which surveys of management accounting methods were “extended to find out not only what practice was being used but also why such practices were used” (p. 278). So there was some attempt to explain why accounting choices were made, but the sample used in this research suggests that such attempts at “positive” theory were infrequent and not a common attribute of early empiricism.

ANTIL’S TAXONOMY OF EVOLUTION

Antil [1989] provides a somewhat similar, but more comprehensive, classification with his assertion that accounting modeling has moved from implementation to explanation (p. 103). The “implementation” classification includes Watts and Zimmerman’s periods through the late 1960s and “explanation” would include contemporary positive theories. With the exception of two articles, all works on accounting in the sample are either directed toward developing normative accounting methods or describing accounting practices without attempting to explain (predict) why the object of study takes the particular form being observed. All of the studies that were directed towards identifying frequency of use of specific accounting methods or techniques provide the opportunity for authors to speculate on motivational determinants of such usage. Such induction would reflect the infancy of contemporary explanatory (positive) research. Yet only two of the authors bother to speculate on why particular methods were used. Thus, the data in the sampled research supports the Antil contention that early literature was primarily concerned with implementation.

THE ROLE OF EARLY EMPIRICISM

It is concluded that, based on research results, early empiri-cism contributed to accounting in two ways: (1) it fulfilled an information function that has been taken over by other sources now such as the AlCPA’s Current Trends and Techniques and (2) it provided a basis from which certain normative concepts were induced. The information function is obvious from the research sampled in this study. For example, Stockwell [1910], Hornberger [1929], Clendenin [1941], and the Research and Technical Service Department, N.A.C.A. [1941] provide surveys of accounting methods used to account for specific events.

The sample of research studies also indicates that early em-pirical works provided the data from which some positive, as well as normative, concepts could have been induced. All of the surveys mentioned above plus case studies, such as those of Mitchell ([November, 1906], [December, 1906], [1907]), provide such data even though the authors do not utilize the data to make inferences. In addition, there are works in the sample in which the authors have made inferences that may have influenced accounting and management thought. For example, Pierson [1906] concluded that private operation of utilities with government regulation provides the best organizational arrangement for such enterprises. Fullington [1916] concluded that proper disclosure of government activities will result in public pressure that brings about efficient government operation. Van Vlissinger [1929] identified benefits from centralization and decentralization of accounting functions and Cannon [1948] recognized the need for capitalizing financial leases. While the authors of this study cannot conclude that studies such as these directly led to specific changes in accounting and business thought, the authors believe that the studies were of such substance that it is likely that they influenced the work of normative theorists.

SPECULATION ON SOME DETERMINANTS OF ACCOUNTING LITERATURE

Watts and Zimmerman [1979] attracted a great deal of attention with their “Market for Excuses.” Although the explanatory power of their theory has been effectively questioned in the literature, there has been no further attempt at explaining the development of accounting theories or accounting literature. Lowe, Puxty, and Laughlin [1983] provide a strong argument against oversimplifying the complex process of the development of accounting theory and literature. The accounting literature of any period is certainly the result of a complex process, yet the final objective of any study of that literature must be to increase the understanding of the factors that operated to create that literature. It seems appropriate to go beyond description and to suggest some determinants of the content of accounting literature that are consistent with the descriptions in this paper.

The most obvious determinant of changes in accounting lit-erature is the advent and availability of the computer. The Vasarhelyi, et al. taxon, “Analytical-Simulation,” which generally is composed of non-empirical studies according to this paper’s operational definition was a null set prior to development of computers. Simulation was just not feasible for accounting studies without the existence and availability of computers. Computer development and accessibility also contributed to the use of more sophisticated design and statistical analysis in accounting literature. Thus, these developments suggests that available technology is an important determinant of the form and content of accounting literature.

If accounting research borrows theories and research meth-ods from other academic disciplines, then accounting literature is surely influenced by the state of art in related disciplines. Some of the taxons identified by Brown, et al. and Vasarhelyi, et al. are dependent upon borrowed methods, particularly from agency theory and information economics, that did not exist during the period from which the research sampled in this study is drawn.

There were, however, techniques available prior to 1950 that were not used, but would have been helpful in several of the studies. Therefore another variable, training and background of academic accountants, in addition to those already identified helps to explain the degree to which accounting researchers exploit methods and theories from other disciplines. Only a very small portion of the faculty in Briggs’ [1930] sample were Ph.D.s. These early faculty surely had less training in (borrowed) research methodology than most contemporary accounting faculty. Also, their training and background most likely resulted in their interests being more closely aligned with those of practitioners and more distant from other academic disciplines and the university as a whole than contemporary accounting faculty.

One might reasonably hypothesize that senior accounting faculty that remain active in research are doing research primarily for psychic reward rather than direct financial reward. Because they are operating in an environment that values the methods of natural and social sciences and humanities more than methods (whatever they might be?) that would bring approval of the practice community, these senior faculty will emulate their more prestigious university colleagues in other disciplines. The academic journals and Ph.D. programs are in the hands of this particular set of faculty; therefore, new accounting faculty must emulate them if they are to survive as academics. Both the awareness and interest required to stimulate borrowing from other disciplines to the extent that it is done today appears to have been absent during the formative years of modern academic accounting literature.

Another comparison suggesting an explanatory variable is the change in the “Institutional” taxon within the “School of Thought” taxonomy. Only one of the works in the sample fits the rather restrictive definition of “institutional” used by Vasarhelyi, et al, yet that empirical work on regulation constitutes an extensive and important body of work in contemporary accounting literature. Much of the increase in interest in research on institutions can be attributed to change in the institutional environment within which accounting operates. While we do not subscribe to the Watts and Zimmerman [1979] thesis that justification for regulatory purposes accounts for the demand for accounting theory, the institutional environment of accounting and business certainly affects the content of accounting literature. Therefore the authors are convinced that the “Market for Excuses” variable is operating as one of many variables that determine the body of accounting literature.

The authors of this paper have proposed that some of the variables appropriate for an explanatory (predictive) model of accounting literature are the development and availability of data accumulation and analysis technology (hardware), the body of knowledge of related disciplines, the background and training of accounting researchers, and the institutional environment of accounting. It is easy to identify other candidates for inclusion in such a model, but the descriptions of early empiricism in this paper are consistent with these proposed variables.

SUMMARY

Shank’s [1981, p. 100] observation that “Empirical research in accounting is essentially a phenomenon of the past twenty years” (now thirty years) is accurate only if descriptive empiricism is ignored. If one is willing to accept descriptive statistics and case studies as “empiricism,” then empiricism was common in accounting literature prior to 1950. A more precise statement by Shank is that the use of inferential statistical tests of theoretical predictions is essentially a phenomenon of the past thirty years.

Early empiricism was the work of both academics and non-academics. Academics published in non-academic journals and non-academics published in academic journals. While the data reported in some publications added little to the work in which it was included, empiricism was central to most of the items in the sample. The studies were primitive by contemporary standards because little attention was paid to experimental design and statistical tools in these early studies.

Accounting specialties represented in the sample were diverse. Several of the studies were directed toward determining reporting techniques or methods. Studies related to management or government were frequently case studies. Articles on accounting education and professional development described curricula and professional opportunities. Three of the articles on professional development were “advocacy” type articles that were intended to promote and defend activities of the AIA or of an academic institution.

The objectives of using and reporting real world observations were primarily to report the use of accounting methods, to describe the application and functioning of information systems, to publicize the activity of institutions, or to support a normative position. In a few cases the observations were used to make normative inferences and two authors attempted to discover the rationale for the variations in treatment of an accounting event. However, interest in explanatory theory appears minimal and no formal tests of theory were found.

The design limitations of this paper prevent the identification of the specific contribution of empiricism to the induction of theoretical positions. However, the study does reveal that the work of early empiricists was of broad enough scope and detail to have permitted the induction of a wide range of theoretical concepts. Therefore, empiricism may have made a much greater contribution to accounting thought than is reflected in historical accounting literature.
Further investigation of the role of early empiricism in the development of accounting thought will be interesting. A poten-tially useful avenue for investigation is to evaluate the temporal relationship between the publication of certain empirical-based works and the appearance of normative positions. For example, did Cannon’s [1948] article on the materiality of financial leases precede the call for recognition of these lease obligations as a liability in the normative literature? Potentially even more rewarding is an investigation of empiricism from 1950 through 1965. It is with the literature of this period that one will be able to observe the transition from sole use of descriptive statistics to early inferential tests.

APPENDIX

Annotated Bibliography of Works Included in the Sample

1906

Mitchell, Thomas W., “Reviews of Corporation Reports,” Journal of Accountancy (November, 1906): 42-47.
This is the first of a series of eleven sets of case studies that appeared in each issue of the Journal of Accountancy through September, 1907. The statements of American Locomotive Company (fiscal year ended June 30, 1906) and Distillers Securities Corporation (1905 and 1906) were analyzed with the objective of determining if “the reports rendered to stockholders contain such information as will enable them to exercise intelligent judgment with respect to the fidelity, efficiency and economy of corporate trustees and agents.” (p. 42)

Mitchell, Thomas W., “Reviews of Corporate Reports,” The Journal of Accoun-tancy, III, 2 (December, 1906): 146-153.
The second of the series of case studies. This case study has the same objectives and procedures as the preceding analysis. The statements of the Chicago, Rock Island, and Pacific Railway Company from 1901 through 1906 were the subject of this analysis.

Pierson, Ward W., “Municipal Control of the Philadelphia Gas Works,” The Journal of Accountancy, 2, 2 (June, 1906): 118-131.
This is a case study of the private and public operation of Philadelphia Gas Works (1834-1906). He conludes that the best arrangement for operating a similar franchise is by private operation with government acting in a “watchdog” role.

1907

Mitchell, Thomas W., “Reviews of Corporate Reports,” The Journal of Accountancy (June, 1907): 144-150.
Mitchell discusses perceived material items specific to the particular statements being analyzed. This analysis was of the 1906 annual report of the Tennessee Coal, Iron and Railroad Company.

1909

Stockwell, Herbert G., “Depreciation, Renewal, and Replacement Accounts,” Journal of Accountancy (December, 1909): 89-103, (January, 1910): 189-210. Stockwell discusses the theory, practices, and bookkeeping for acquisition and maintenance of long-lived operating assets. The empirical portion of the paper, the description of practices, is a relatively minor part of the paper.

1916
Fullington, E. M., “The Budget System in Ohio,” The Journal of Accountancy, XII, 2 (February, 1916): 114-124.
The thesis of this article is that proper disclosure will result in public pressure to control spending. This case study is a description of the transition in the Ohio budgetary system in 1914 and 1915.

1924

Warshaw, H. T., “Inventory Valuation and the Business Cycle,” Harvard Business Review (October, 1924): 27-34.
The objective of this paper was to argue the merit of the base stock inventory method. Warshaw compares the income performance of the National Lead Company with that of ten other companies from 1915 through 1922.

1927

Le Rossignot, J. E., “Demands of Business for Trained Men Met by Schools and Universities,” The American Accountant, XII, 11 (December, 1927): 13-15.
The objective was to describe business education at the University of Nebraska. The data on graduates was gathered by questionnaire.
Taylor, J. B., “Some Phases of North Dakota’s Experiment in Flour Mill Opera-tion,” Accounting Review, II, 2 (June, 1927): 129-130.
Taylor’s objective was to evaluate the state operation of a flour mill. He describes the legislative and financing history of the North Dakota venture in this case study. Also, he presents the 1926 balance sheet for the milling company and related numbers and discusses the cause of the unfavorable financial position.

1928

Nissley, W. W., “Progress of the Bureau for Placements,” Accounting Review, III, 1 (January, 1928): 36-37.
Nissley’s objective was to describe the work of the Bureau (AIA) and answer criticism of accounting professors. Data was obtained by questionnaire sent to AIA members admitted during 1917 through 1928 and from the records of placements by the Bureau.

1929

Atkins, Paul M., “University Instruction in Industrial Cost Accounting,” Account-ing Review, IV, 1 (January, 1929): 23-32.
Atkin’s objective was to describe a course in cost accounting that met the needs of both advanced undergraduate and graduate students. Empirics were used to describe existing courses. He reviewed 295 college catalogues, 88 of which listed cost accounting courses. Questionnaires were sent to the eighty-eight colleges listing cost accounting courses and thirty-four responses were received. Catalog descriptions were used for non-respondents.
Burrell, O. K., “An Experiment in Student and Teacher Ratings,” Accounting Review, IV, 3 (September, 1929): 194-197.
This project had the joint objectives of the development of an instrument to evaluate student aptitude for accounting and to evaluate faculty. The sample consisted of 225 freshmen taking the first accounting course in 1927 and 350 in 1928.

Hornberger, D. J., “Accounting for No-Par Stock Issues,”‘ The Accounting Review, IV, 4 (December, 1929): 213-217.
Hornberger’s objective was to describe how corporations account for no-par stock and to determine the rational for the treatments. He examined financial statements for the treatment of 1,213 stock issues of 614 utilities and 599 industrials. An undisclosed number of managers were then interviewed to determine why they used a particular method.

Van Vlissinger, Arthur, Jr., “A Survey on the Effect of Departmental Consolidation on Cost-Office and Factory,” N.A.C.A. Bulletin (August 15, 1929): 1451-1465.
Van Vlissinger’s objective was to discuss the effect of organizational structure change and office systems redesign on costs. The data upon which the article was based was one thousand questionnaire responses from of fices large enough to have a management problem and an unspecified number of interviews with accounting and budget officers. The survey results and anecdotes are reported in the paper.

Yoakum, Clarence S., “Business and the Young Accountant,” Michigan Business Studies, 11, 3 (May, 1929): 1-42.
Yoakum’s objective was to provide information on career opportunities and difficulties in accounting. Data were obtained from a questionnaire sent to 250 persons taking three or more accounting courses at the University of Michigan from 1916 through 1926.

1930

Briggs, L. L., “Accounting in Collegiate Schools of Business,” Accounting Review, V, 2 (June, 1930): 175-181.
Briggs’ objective was to describe accounting taught in AACSB accounting programs. He obviously used program and course catalogues from member schools as his data source.

“Cost of Handling Merchandise Returns is Subject of Study,” The American Accountant (March, 1930): 110-111.
The objective of this brief article was to review an Ohio State report of case studies of department store treatment of sales returns and to present significant statistics from the report.

New York State Society of Certified Public Accountants, Special Committee on Accountants Responsibility for Verification of Inventories, “Results of Survey of Accountants Responsibility for Inventory,” The American Accountant, XV, 4 (April, 1930): 162-165.

This was the report of the Committee which surveyed some unspecified number of accountants. The report, a discussion of the data collected with no position taken, contains no indication of procedures or methodology.

1931

“Operating Ratios of Transient Hotels at Varying Rates of Occupancy,” The American Accountant (April, 1931): 125.
This review of a report prepared by the Department of Hotel Adminis-tration, Cornell was probably prepared by one of the professional editors of The American Accountant. Average ratios and standard error of estimates are presented and the meaning of the standard error is discussed.
Wasserman, Max J., “Accounting Practice in France During the Period of Monetary Inflation (1919-1927),” The Accounting Review, VI, 1 (March, 1931): 1-32.
Wasserman’s objective was to describe the French proposals (theories) for inflation adjusted accounting following World War I. The bulk of the article is a description of these theories. The empirical portion identifies the

French price changes during this period and is used to demonstrate the sustained nature of inflation in France from 1918 through 1927.

1932

*FitzPatrick, Paul J., “A Comparison of Ratios of Successful Industrial Enterprises with Those of Failed Firms,” Certified Public Accountant (October, 1932): 598-605; (November, 1932): 656-662; (December, 1932): 727-731.

FitzPatrick’s objective was to compare ratios of successful firms with unsuccessful firms. He obtained financial data from Moody’s and Poor’s Industrial Manuals for three years. Neither firms nor periods were specified.

Kohler, Eric, “Editorial: The Public Utility Holding Company,” The Accounting Review, VII, 4 (December, 1932): 301-305.
The purpose of this editorial is to identify some of the unacceptable business and accounting practices of large public holding companies and to recommend action for the correction of the abuses. Kohler reviews the history of unacceptable practices of a large public utility investment-holding company to illustrate his points.

1934

Atwood, Paul W., “The Manufacturer Looks at His Cost of Distribution,” The Accounting Review, IX, 1 (March, 1934): 23-28.
The objective of this article was to discuss a survey of distribution costs of manufacturers (312) firms conducted by the Association of National Ad-vertisers.

1941

*Clendenin, J. C, “How 118 Major Corporations Account for Bond Discount,” The Journal of Accountancy (July, 1941): 37-44.
Clendenin’s objective was to identify the accounting methods used for discounts on retired bonds and the motives for the selection of the methods. Data was obtained from 180 questionnaires for which 118 responses were received.

Research and Technical Service Department, N.A.C.A., “Accounting for Excess Labor Costs and Overhead Under Conditions of Increased Production,” N.A.C.A. Bulletin, XXII, 24 (August 15, 1941): 1551-1570.

The objective was to determine how companies were accounting for increased labor costs (overtime, training, etc.) and overhead in non-defense activities that resulted from war induced demand for all goods. Data for 263 companies was obtained from questionnaires distributed at N.A.C.A, meetings to chapter members “best qualified to supply the information desired.”

1942

Devine, Carl T., Inventory Valuation and Periodic Income, New York: The Ronald Press Company (1942).
Devine’s objective was to investigate the “effects of various inventory valuation methods on the reported income stream.” (p. 111) His real data were the inventory numbers and related information 1933 through 1938 reported by Wilson [1939] for an unidentified firm that processed cotton.

Magor, Donald M., “Job Evaluation for Salaried Positions,” N.A.C.A. Bulletin, 23 (June 15, 1942): 1354-1365.
This case study describes the development and implementation of a salary plan for positions within the York Ice Machinery Corporation.

1947

*Borth, Daniel, Jr., “Published Financial Statements of Banks,” The Accounting Review, XXII, 3 (January, 1947): 282-294.
Borth examined the 1946 statements of twenty-five Chicago banks and made qualitative judgments based on reporting standards current at the time.
*King, Robert W., “Effect of Inventory Valuation Methods on Profit,” The Ac-counting Review, XXII, 1 (January, 1947): 45-53.
King’s objective was to illustrate the effect of FIFO and LIFO. Most of the paper is directed towards specifying IRS requirements for inventory valuations. The empirics appear as if they were an afterthought and contribute nothing to the paper.
*Lemke, B. C, “The Treatment of Unamortized Discount and Expense Applicable to Bonds Refunded Before Maturity,” The Accounting Review, XXII, 4 (October, 1947): 375-384.

This is a normative discussion of the treatment of bond discount and expense upon refunding with a survey of the methods used and the amounts involved. Data (treatment of discount and expense, assets, liabilities, and the amount of discount and expense) for thirty-four refundings by thirty industrial firms and thirty-one refundings of thirty utilities were displayed.

*Littleton, A. C, “What Auditors Do,” The Accounting Review, XXII, 1 (January, 1947): 80-81.

Litdeton’s objective was to determine what auditors do by determining the frequency of verb usage in audit literature. A content analysis of the work of five unidentified authors (1,720 pages of auditing literature) was performed and the frequency and distribution by author of commonly used verbs was reported.
*Staff of the Journal of Accountancy, “What the Public Thinks of Financial Statements,” Journal of Accountancy (June, 1947): 487-489.
This is a brief summary of Roper’s [1948] survey with the theme that the public does not understand financial statements.

Williams, Aubry, “Trends in Accounting Terminology and Form Revealed by 1946 Corporate Reports,” The Journal of Accountancy (October, 1947); 310-317.
Williams’ objective was to describe changes in terminology and statement format used in published financial statements. No data was presented other than examples of departures from standard terminology and format.

1948

*Borth, Daniel, Jr., “Adjustment of Prior Year’s Income per Depreciation and Disposal of Fixed Assets,” The Journal of Accountancy (June, 1948): 470-474. The objective was to discuss the problem of price changes and its resolution within the historical cost model. The data source and methodology were author examination of “about 150 representative published corporate financial statements [for fiscal years ending in 1946 and early 1947] … to ascertain the elements in the determination of net income for the period.” (p. 470)

*Canon, Arthur M., “Danger Signals to Accountants in ‘Net-Lease’ Financing,” The Journal of Accountancy (April, 1948): 312-320.
Cannon’s objective was to describe the materiality of sale-lease back activity and to discuss appropriate accounting. He presents data on the off-balance sheet financial lease financing of four corporations (Allied Stores Corporation, J. C. Penney Company, Safeway Stores, Inc., and Montgomery Ward & Co., Inc.) for 1946. Lease data on other firms is presented in anecdotal style. Cannon calls for statement recognition of financial lease activity.
*Roper, Elmo, A Report on What Information People Want About Policies and Financial Conditions of Corporations, Vols. I and II, New York: Controllers’ Institute Foundation (1948).

The Controllership Foundation commissioned Roper to determine what the public was interested in about financial statements and corporate affairs. Questionnaires and interviews were used and the report provides tabulation of the questionnaires and excerpts from in-depth interviews. Af ter a preliminary pilot study (754 interviews) to determine the economic distribution of stock ownership, a national cross-section of the population stratified into three economic levels was sent questionnaires and 4,259 responses were obtained. Also, 1,242 corporate employees completed questionnaires and a portion were interviewed, 59 large investors were interviewed, 61 financial analysts and bankers were interviewed, and 88 labor union officials were interviewed between January 26 and April 23, 1948.

Wood, Ben D., Arthur E. Traxler, and Warren W. Nissley, “College Accounting Testing Program,” The Accounting Review (January, 1948): 63-83.
The objective of the article was to describe the development of the AIA testing program and all aspects of the program are discussed. The sample size varied with the portion of the program, but results were available for over 25,000 exams.

1949

Child, A. J. E., “Case Studies in Internal Auditing,” Accounting Review (April, 1949): 149, 149-159.
Child describes the internal audit function within a large Canadian meat packer. The article includes sample audit programs for accounts receivable and inventories and anecdotes of the internal auditors performing special assignments and detecting defalcations.
*Jones, Ralph C, “Effect of Inflation of Capital and Profits: The Record of Nine Steel Companies,” The Journal of Accountancy (January, 1949): 9-27.
Jones’ objective was to demonstrate the impact of price change accounting on reported income of nine unidentified steel companies from 1941 through 1947.
*Included in Ball’s Index.

OTHER REFERENCES

Antil, Rick, “On Intellectual Boundaries in Accounting Research,” Accounting Horizons (June, 1989): 103-109.
Ball, Ray, “Index of Empirical Research in Accounting,” Journal of Accounting Research, 9, 1 (Spring, 1971): 1-31.
Bricker, R. J. and G. J. Previts, “The Sociology of Accountancy: A Study of Academic and Practice Community Schisms,” Accounting Horizons, 4, 1 (March, 1990): 4-14.
Bowles, H. G., Statement Presentation of Non-Current Gains and Losses,” The American Accountant, XIII, 6 (1933): 176-178.
Brown, L. D., J. C. Gardner, and M. A. Vasarhelyi, “An Analysis of the Research Contributions of Accounting Organizations and Society,” Accounting Organi-zations and Society, 12, 2, (1987): 193-204.
Brown, Lawrence D. and Miklos A. Vasarhelyi, Accounting Research Directory, New York: Markus Wiener Publishing (1985).
Cochran, William G., Sampling Technique, 3rd ed., New York: John Wiley & Sons (1977).
Danford, Fred, “Some Modern Developments in Accounting for Manufacturing Enterprises,” The Accounting Review, VIII, 2 (June, 1931): 119-121.
Flamholtz, Eric, “The Role of Academic Accounting Research: An Historical Perspective,” The Accounting Forum, 47, 1 (May, 1977): 41-51.
Lowe, E. A., A. G. Puxty, and R. C. Laughlin, “Simple Theories for Complex Processes: Accounting Policy and the Market for Myopia,” Journal of Ac-counting and Public Policy, 2 (Spring, 1983): 19-42.
Previts, Gary and Barbara Merino, A History of Accounting in America, New York: John Wiley & Sons (1979).
Scheaffer, Richard L., William Mendenhall, and Lymon Ott, Elementary Survey Sampling, 3rd ed., Boston: Duxbury Press (1986).
Shank, John K., “Academic Research in Accounting: The Impact of Accounting Research on Policy and Practice,” in Buckley, John W. (ed.), The Impact of Accounting Research on Policy and Practice, Reston, Va.: Reston International Center (1981).
Vasarhelyi, Miklos A., Da Hsien Bao, and Joel Beck, “Trends in the Evolution of Scholarly Accounting Thought: A Quantitative Examination,” The Accounting Historians Journal, 15, 1 (Spring, 1988): 45-64.
Watts, R. L. and J. L. Zimmerman, “The Demand for and Supply of Accounting Theories: The Market for Excuses,” The Accounting Review, 54 (1979): 273-305.
, Positive Accounting Theory, Englewood Cliffs, N.J.: Prentice-Hall, Inc.
(1986).
Wilson, George A., “Further Consideration of the Last-In, First-Out Basis of Inventory Valuation,” N.A.C.A. Bulletin (September 1, 1939): 1-26.