Graph standardization and management accounting at AT&T during the 1920s

Introduction

What spurs managerial accounting innovation? The source of modern-day management accounting practices can be traced to early, giant corporations whose rapid growth within a dynamic business environment required informational and institutional adjustments. Such was the case at the American Telephone and Telegraph Company (AT&T), an enterprise that dominated the US telephone industry in the early twentieth century. Due to the novelty, size and complexity of the firm and industry during this era, AT&T faced the dual challenge of maintaining rapid growth and expansion while minimizing risk. It therefore needed measures to inform and integrate diverse aspects of its enterprise – such as production planning, budgeting, personnel management and regulation. These provided AT&T with the impetus to pioneer new accounting techniques and presentation formats and standardize those practices within the firm.

This study examines a key element in management accounting standardization at AT&T – the use of graphical methods to present, communicate and evaluate information that cut across the firm’s numerous divisions, functions and geographical areas. We focus on the decade of the 1920s as this period saw the rapid rise of management accounting innovation within the firm. Our study presents another example of how developments in accounting responded to the demands of the newly emerging industrial corporations (Chapman, 1997). Extensive archival materials reveal that graphical standardization pervaded all areas of the firm, and drew upon the dominant firm culture of scientific innovation.

By the 1920s, Bell’s standardization initiatives pervaded a wide range of company activities to assist management in its responsibilities for planning and monitoring performance. ¹ The firm participated in international efforts to define standards for measuring the efficiency of telephone network performance. It also worked with the National Electric Light Association, a consortium of electric utilities and electric goods producers, to develop standards for the use of jointly operated facilities such as telephone poles (Fagen, 1975, pp.336–7). The firm played an active role in trying to win support for its “transmission unit”, later renamed decibel, as an international standard for measuring telephonic efficiency. Standardization was critical in equipment and process planning for major, new manufacturing facilities constructed in Kearney, New Jersey and Baltimore, Maryland (Fagen, 1975, pp.308–9). The firm specified uniform approaches for the distribution and installation of apparatus. It established mandatory procedures for plant maintenance, traffic management, inventory control, construction, and human resources and pension planning. It influenced office activity by specifying the design of forms, billing and collection procedures, and even types of office equipment. Health and safety practices were also transformed through the scrutiny of the firm’s “systematizers”. These initiatives afforded several advantages which enhanced the ability of top management to plan, monitor and assess operations.

Standardization became increasingly ingrained in the firm’s culture and found its way into the firm’s management accounting practices, too. Although used extensively in engineering since the last quarter of the nineteenth century, the use of graphs and charts did not become formalized within the controllership function and in managerial accounting until the 1920s. These media provided visual representation of various numeric series. Graphs strengthened management control by reducing complexity and by communicating effectively such phenomena as the direction, rate and magnitude of change, periodic patterns, points of inflexion, and gauges of relative magnitudes.

Graphic standardization provided other operational advantages. First, standard methods saved time and money, as they reduced uncertainty and confusion by limiting the range of decisions. Second, they created unambiguous reports about operational performance that were readily understood by all managers, thereby reducing costly internal information asymmetries. Third, the graphs reduced the number of acceptable alternatives, lowering data summary and analysis costs. Fourth, efficient forms design improved cognition through the simplification of business dynamics. Fifth, standardization strengthened management by ensuring that all levels of management had access to the best practices and technologies. Sixth, standardization specified norms and accepted practice for group action in resolving operational problems, thereby promoting efficiency and defining firm culture.

The following sections evaluate the experience of Bell’s comptroller’s office in standardizing graphic methods to promote managerial accounting efficiency. Initially, we clarify links to prior research. Next we discuss the factors that persuaded top management to give priority to graphic standardization in managerial accounting. The subsequent section describes the evolution of graphing techniques. The paper concludes by explaining how this episode amplifies our understanding of management accounting evolution.

Relationship to prior literature

While we examine early developments in the area, the term “management accounting” was not commonly used in the 1920s. Starting with the railroads, early industrial firms were more narrowly concerned with accounting for costs, and the term “cost accounting” began to appear in textbooks by the early twentieth century (Previts & Merino, 1998). Yet, during this period, there were pioneering publications and firm practices that led to the transition from cost accounting to management accounting as it is conceived today. At the University of Chicago, John Maurice Clark, in his 1923 book, Studies in the Economics of Overhead Costs, argued in the chapter “Different Costs for Different Purposes: An Illustrative Problem”, that: “there is no one correct usage [of cost], usage being governed by the varying needs of varying business situations and problems” (p.175). One year later, James O. McKinsey, an associate professor of accounting at Chicago (who later founded the consulting firm which bears his name) wrote a textbook entitled Managerial Accounting (Zeff, 2008). In 1921–1922, Harvard University’s Thomas H. Sanders developed and taught a course in “Industrial Accounting” which viewed accounting as a means of executive control (Zeff, 2008). Sanders distinguished “industrial accounting” from “cost accounting” (Sanders, 1923). The latter, he said, was limited in some minds “to the simple matter of adding up a few items of expense to get the total cost”, while the former signified “the science of the control of business operations through cost analysis” (Sanders, 1929).

It is clear that many of the elements of modern-day managerial accounting were present in the intellectual realm as well as within industrial giants like AT&T during the period of our study. Johnson and Kaplan (1987, p.12), for example, have noted that “By 1925, virtually all management accounting practices used today had been developed”. J.R. Edwards and E. Newell (1994), focusing primarily on the British experience, have also noted the pre-1850 origins of many managerial accounting practices. Similarly, Sidney Davidson (1963), acknowledging Clark’s contributions to relevant cost analysis, referred to later management accounting developments as “Old Wine into New Bottles”. And Fleischman and Parker (1991) have argued that the advanced nature of cost accounting at several early modern companies played a useful, though often overlooked, role in evaluating new production technologies, aiding the rise of these successful industrial enterprises.

The evolution of management accounting practices has been studied both by economists and historians. Information economists view it as a tool for overcoming uncertainty in decision processes (Demski, 1980; Demski & Feltham, 1976). In the closely related agency literature, accounting provides the information for contracts that connect economic agents. Accounting informs principals about their agent’s activities, thus facilitating the crystallization of mutual understanding about contract outcomes (Demski & Feltham, 1978; Baiman, 1982). Alchian and Demsetz (1972) further argue that firms arise because of the benefits of team production. In their view, accounting information becomes vital because of its ability to disclose to principals the outcomes of agent activity. While information economics and agency theory have attempted to provide mathematical rigor and equilibrium solutions to understanding management accounting arrangements, their usefulness in understanding information in complex organizations such as AT&T is hampered by restrictive assumptions like utility and profit maximizing behavior (Kaplan, 1984). Moreover, agency theory assumes institutional structures as given, resulting in a static equilibrium analysis. It does not consider the role of knowledge and innovation as value drivers over time. This approach is therefore ill-suited to understanding the dynamic complexity of the evolution of management accounting practices such as those at AT&T during the 1920s.

One index of the growing use of graphs as an adjunct to accounting is reflected in the entries of the American Institute of Accountants’ (forerunner of the modern American Institute of Certified Public Accountants) annual bibliographic publication, The Accountant’s Index. The first edition of this appeared in 1920 and had the ambitious goal of including the entirety of academic and professional publications dealing with accounting to date. The initial volume listed 130 articles relating to “graphic methods” which were mostly published after 1900 (American Institute of Accountants, 1921, Vol.2, pp.853–5).

Business historians have also contributed to our understanding of management accounting evolution by focusing on the nexus between knowledge and organization. Alfred D. Chandler (1977) in The Visible Hand explained how the development of America’s first big business, railroads, would have been impossible without the early cost accounting promoted by Albert Fink and others. For Chandler, the growth of industry necessitated new managerial accounting information. H. Thomas Johnson and Robert Kaplan in Relevance Lost stressed how the problem of joint cost allocation and the difficulties of acquiring economically relevant information impeded managerial decision making at large, complex business organizations. This implies that information systems contributed to firm growth (Johnson & Kaplan, 1987).

Yates (2000) argues that the great spurt in managerial accounting innovation at the turn of the twentieth century resulted from the need to exert managerial control in an era of growth and uncertainty. She notes that as firms grew vertically and horizontally, the need for managerial control to coordinate activities also grew. This gave birth to a field which she calls systematic management (a term which encompasses the scientific management movement). The “systematizers” produced extensive documentation which they disseminated both up and down the firm’s hierarchy. This great volume of information led to problems of information overload and communication. This is turn led to the quandary of how to reduce this overload while emphasizing important facts. Consistent with this perspective, AT&T turned to graphs as an especially effective way to handle this problem.

Standardization of graphs made possible the extraction of information and its communication to a management which was removed from daily operating activities. Standardization also contributed to the process of “deskilling”, the transfer of knowledge from the skilled worker to management (Braverman, 1974), which became a hallmark of scientific management (Hoxie, 1921, p.124). Critical accounting researchers such as Loft (1986) have stressed the labor disciplining aspect of standardization and deskilling; those with a more nuanced view hypothesize that deskilling allows management to control workers by depriving them of the market power of their skills (Boyns & Edwards, 1997). Our research analyzes the role of graphic presentations in simplifying complex information as a pragmatic solution to complexity and not as part of an effort to deskill labor.

Our article examines the same general time-frame investigated by these researchers; however we do not enter into the debate on whether growth required or was enabled by better information. Instead, we observe that large corporations were acquiring vast amounts of information about their businesses, and the sheer size and complexity of the firms brought difficulty in distilling meaning from the voluminous accounting information generated. The accountants at AT&T met this challenge by adopting the tools of graphical analysis, the development and use of which is the focus of our article. Our definition of managerial accounting, then, encompasses more than the classification of account balances and ratios or formulas using this data. We include within managerial accounting the wide body of ideas, tools, and techniques available to management to control and improve business and the modes by which they are presented and shared (Boyns & Edwards, 1997).

Margaret Levenstein (1998) in Accounting for Growth demonstrated how the information provided by the cost accounting system at Dow Chemical Company changed as the firm made the transition to research-driven growth. Similarly, management accounting researchers have examined the development of cost constructs and their role in improving the business decision process (Garner, 1954; Solomons, 1968; Chatfield, 1974; Johnson, 1972, 1980, 1983; Kaplan, 1984; Johnson & Kaplan, 1987; Fleischmann & Tyson, 1996, 1998; Previts & Merino, 1998; MacDonald & Richardson, 2002; Boyns & Edwards, 2004). Some of these authors stress that the accounting profession was not an important source of cost accounting innovation, but that these innovations developed within large and complex firms as a response to business strategies (Johnson & Kaplan, 1987; Boyns & Edwards 1996; Foreman, 2001). However, despite the importance of graphs as a form of comprehending, presenting and sharing accounting information, little research exists on the growing role of graphics by managerial accountants during the early evolution of management accounting (Davison & Warren, 2009). ² Our study contributes to this body of managerial accounting literature by examining graphical artifacts developed and used within AT&T as important early innovations in managerial accounting.

The modern use of graphs as a tool to manipulate financial reports in an attempt to manage external perceptions of economic performance has been investigated by a few researchers (Johnson et al., 1980; Beattie & Jones, 1992, 2000, 2002, 2008). ³ Similarly, Jones (2011) provides evidence of impression management of graphs in social and environmental reporting by contemporary firms. In examining the role of graphical representations as a way to manipulate external information dissemination, these studies assume the existence of numerous formats from which management selects. In contrast, our study examines the development of graphs for internal communication and for the compilation of a firm’s stock of knowledge. While we agree that the creation of graphs rests on a selection process in which management determines the information to be presented, we do not focus on the selection and use of graphs to manipulate users’ perceptions; rather, we examine the creation of standardized graphs within a firm as a response to information overload and an uncertain environment. It is not a selection process which aims to manipulate the perceptions of others, but one which attempts to uncover important relationships while creating a shared learning base.

The use of graphical reports in management accounting contexts has also been examined in the experimental literature. Descantis and Jarvenpaa (1989) provide support for their contention that graphical reporting formats can improve the accuracy of forecast judgments. Research has also largely demonstrated the superiority or incremental usefulness of graphics in complex multidimensional decision contexts (Benbasat & Dexter, 1986; Wright 1995). More recently, Cardinaels (2008) argues that decision makers with a low level of cost accounting knowledge benefit greatly in the quality of their decision making when they use graphical formats.

Whether graphs are used as information management tools or to improve internal decision processes, a key benefit is that their use can reduce a firm’s cost of capital. At AT&T, graphical tools lowered cost of capital by decreasing information asymmetry between central management and the operating entities. Providing useful data in public debates over rates and other corporate policies was also a method of increasing transparency with the investing public, and decreasing political cost, again with the aim of ultimately reducing financing costs.

Statistics and graphics are methodologies used to summarize, communicate and analyze complex phenomena (Feinberg, 1979). Graphical representations can be classified as a genre of organizational communication produced by a firm’s internal reporting systems (Yates, 1989; Yates & Orlikowski, 1992). Each genre addresses a recurrent business situation by employing a communication style consistent in both form and substance. Genres can develop either within a particular firm or across an industry. At AT&T in the 1920s, then, we have a case study in the creation, standardization, and dispersion of a particular communication genre – graphical presentation. We use this as a lens to explore the institutionalization of knowledge in the successful evolution of modern industrial organizations.

In explaining comparative firm performance in high-tech industries, Alfred D. Chandler (1992) focuses on internal learning capacities as a prerequisite for exploiting first mover advantages from new scientific knowledge. In the field of management research, Robert M. Grant (1996) believes that a firm’s successful knowledge integration requires the development of common modes of interaction – rules, routines, and sequencing – all of which depend on a common language and other modes of inter-group expression. A high degree of common language thus facilitates the smooth transmission of information and knowledge through complex organizational settings. Michael Dennis (1987) conceived of statistics and accounting information as heuristic devices employed by firms to internally leverage their substantial investment in knowledge.

In the same vein, John Burns and Michael Scapens (2000), focusing on the evolution of managerial accounting, see the process of innovation as being bounded by the influence of “rules” (formal guidelines), “routines” (actual practices) and “institutions” (underlying assumptions about inter-group dynamics). Innovation researchers would classify Burns and Scapens’ investigations into the genesis of managerial accounting practices as studies of administrative innovation.

Innovation occurs in the design of a product as well as its delivery. Innovation can be radical and groundbreaking, overturning existing industry standards. It can be incremental, reflecting adaptations and developments that improve an earlier innovation or move it into a different arena. Though innovation is commonly thought of as being related to technology, it can also refer to changes and improvements in administrative or managerial practices. To succeed in a new industry or with a new product, firms need to adapt their organizational practices; but, by their very nature, administrative innovations usually lag behind technological innovations (Damanpour & Evan, 1984; Dunk, 1989). As a firm constantly pushing technological barriers, AT&T’s challenge of managing businesses where proven measurement standards did not exist impelled the firm to innovate administratively. AT&T’s experience seems to confirm Dunk’s suggestion that firms which are able to minimize the time span between technological innovation and the development of complementary managerial accounting tools will gain a competitive advantage because they will be able to take advantage most fully of the technical innovations (Dunk, 1989).

Our study of graphical representation of management accounting information at AT&T is a contextual analysis of a complex organization during a dynamic decade. While Chandar and Miranti (2009) have examined AT&T’s institutionalization of knowledge in an integrated learning base, that earlier study focuses on the development of statistical approaches in managerial accounting with a view to providing linkages among business forecasting, production planning and budgeting. The current research, in contrast, stresses the development of methodologies as a means to present and communicate information more broadly, encompassing various functional areas of the firm such as production, marketing, finance, personnel management, etc., as well as parties outside the firm such as regulators. Historical methodologies such as ours complement insights gleaned from stylized economic models and fill a niche indicated by Kaplan (1984, p.407). Kaplan suggests that research on “the roles of knowledge, technology, and innovation, so critical for the survival of contemporary firms, have yet to be examined by any contemporary theorists from economics or managerial accounting”. We seek to fill this void as we contribute to an understanding of management accounting in the control and coordination of a complex, decentralized firm. We show how AT&T devised graphical techniques for the creation and dissemination of information pertaining to a variety of areas within the firm. The company, through detailed statistical bulletins, also classified graphical approaches, provided detailed principles of graphical presentation and disseminated these practices firm-wide. Such standardization of information within the firm was necessary for managing firm complexity in an era of rapid expansion and economic uncertainty.

Impetus to standardize visual reporting techniques at AT&T

Founded in 1877, AT&T exploited its ownership of the Alexander Bell patents to enjoy a monopoly in the telecommunications market until 1894 when these rights expired (Garnet, 1985). A period of intense competition followed, but over the next two decades the firm was effectively able to re-establish its monopoly in many of the nation’s most lucrative urban centers. In 1907, Bell had just over half (50.3 per cent) of the national telephone market of 6.1 million phones. ⁴ Bell was intent on increasing its share of a growing market by expanding physical facilities and through acquisitions. The costs associated with this strategy were huge; the total telephone plant on the books in 1907 amounted to $503 million and grew to $742 million in five years, representing an annualized growth rate exceeding 8 per cent. This growth rate was superseded by over a 10 per cent annual increase in Bell’s customers: Bell’s market share of the nine million phones in service was 55.8 per cent by 1912.

The Bell System’s dominance became so apparent that the administration of President Woodrow Wilson intervened in 1913 to force the company under the “Kingsbury Commitment” to sell its interest in Western Union in an anti-trust effort to boost competition (Danielian, 1939; John, 1998). Though the commitment further called for Bell either to cease growth through acquisition or face anti-trust litigation, in fact, growth continued. By 1920, Bell phones totaling nearly 8.3 million represented 61.5 per cent of the market. If connecting and subscribing phones are included, the Bell System served 94 per cent of the market. This growth strategy succeeded because the firm balanced the operating companies’ ability to respond flexibly to local needs while still coordinating national interconnections. Table 1 illustrates the rapid growth of the firm.

OPEN IN VIEWER

Table 1.

Growth of the Bell System, 1900–1930

Year	Total assets	Total revenues	Miles of wire	No. of employees (excludes Western Electric & Bell Labs)	No. of AT&T stockholders
1900	230,226	46,386	1,962	37,067	23,469*
1910	753,324	165,613	11,642	120,311	40,381
1920	1,634,250	458,141	25,377	231,316	139,448
1930	5,000,196	1,151,965	76,248	324,343	567,694

Notes: Information from AT&T annual reports; numbers in 000s, except for employees and stockholders. *Number of shareholders at 31 December 1907, the earliest count of shareholders mentioned in the annual report.

Much of this growth occurred under the leadership of Theodore N. Vail who began his tenure as AT&T’s president in 1907. Under his “universal service” plan, Vail sought to broaden public access to a telephone service that had previously been limited to the wealthiest individuals and companies. Vail envisioned universal service as requiring business rationalization, with centralization and standardization as essential elements of this process. In order to reduce costs to achieve his goal, the telephone executive tightened controls over the firm’s widespread and loosely monitored operations by mandating greater standardization of system technology and management practice. The development of uniform management accounting reporting regimes proved difficult in a loosely knit holding company structure with a high degree of administrative autonomy among its many decentralized subsidiaries. This problem largely stemmed from the early decision to build the company through exclusive regional franchises and later through corporate acquisitions from an aggressive program of horizontal expansion.

From Vail’s perspective, it was vital for planning and control that the information available to central management mirrored that used in peripheral operations. Such consistency was critical in pursuing its two-prong strategy of growth originally propounded in the 1880s. First, the firm focused on developing economies of scale by concentrating on building local service in fast-growing, traffic-dense cities. Second, it sought to exploit economies of scope by linking urban centers through a nationwide, long-distance telephone network.

Standardization of reporting also responded to the demands of both state and federal agencies which began to encroach on corporate prerogatives during the early decades of the twentieth century. Although such oversight had initially centered on urban utility commissions such as the one that monitored telephone growth in Chicago, this began to change with the rapid rise of state regulatory boards in the period 1906–1915. By 1915, there were 45 state commissions in operation (John, 1998). These agencies exercised authority over rate setting in their jurisdictions and other critical corporate policies. Moreover, they increasingly demanded that telephone companies present evidence of economical (reasonable cost) and efficient (rapid system response and good transmission quality) service to justify the continuance of their valuable public franchises. On the national level, the Bell System first came under federal authority in the form of the Interstate Commerce Commission (ICC). ⁵ The ICC required the Bell System to file uniform financial and statistical reports (reports whose design the firm helped to define).

From the beginning of his tenure, Vail had anticipated and welcomed federal regulation, stating “there is no serious objection to such control, provided that it is intelligent, considerate, thorough, and just” (AT&T, 1907, p.18). Vail’s embrace of regulatory oversight was in part an attempt to forestall a breakup of the Bell System, but it also reflected operational realities (Danielian, 1939). Regulation provided an important rationale supporting standardization, a process that could help weave together an integrated national system from a sprawling enterprise.

Centralized administrative controls were needed to bring order to Bell’s far-flung components. The System’s toll lines reached over 70,000 locations throughout the continental United States, and in 1913 there were 8,133,017 stations, of which roughly a third were operated by independent or associated firms and not by the Bell System directly (AT&T, 1913). Regulation, though constituting external interference in the firm’s affairs, was useful to Vail because it added force to management’s standardization efforts. The regulators’ value system favored a large-scale telephone network over local alternatives. This aligned regulators’ interests with those of Bell’s management against parochial regional concerns. Moreover, standardization protected the overall system from suffering if there was weakness in any one section (Lipartito, 1989). The spur to managerial accounting innovation at AT&T provided by regulators appears unusual; neither Boyns and Edwards’ (1996, 1997) examinations of cost accounting innovation in nineteenth-century England nor Foreman’s (2001) exploration of the spread of scientific management techniques in the Australian defense industry of the early 1900s found government regulation a significant force in accounting innovation. ⁶

Another factor in AT&T’s development of graphical methodologies was scientific management, which suggested that business efficiency could be attained through rationalizing the firm’s activities. Information collection and dissemination were central to scientific management. References to scientific management at AT&T were most frequently found at Western Electric, the manufacturing arm of the Bell System, ⁷ but its influence was not limited to manufacturing. For instance, Gantt charts, developed by scientific management’s Henry Gantt in the mid-1910s, were used by AT&T in 1919 to track changes in labor costs (AT&T, 1919).

Bell leadership’s knowledge of scientific management included awareness that it could be employed in ways harmful to the firm. In a Supreme Court case which pitted regulators against the railroads, the Eastern Rate Case (1910–1911), Louis Brandeis had used scientific management principles to refute railroad costs and concurrent claims for higher rates. ⁸ Bell management recognized that it needed to document costs – to show that they were accurate and not excessive – to defend the rate structure against criticism by regulators. ⁹ This gave the firm additional impetus to master and extend the tools of scientific management.

Standardization of graphical reporting at AT&T in the 1920s

Standardization of graphical representations was part of the effort to improve the application of managerial accounting at the Bell System. Graphic standardization occurred at the same time as the firm was developing other new tools of economic analysis, such as the “business activity index” (Chandar & Miranti, 2009). Although graphical depictions had long been used for engineering and business purposes, the firm had not established rules about acceptable forms and designs. This doubtless reflected the lack of consensus among world experts concerning best practices. In 1855, for example, the International Congress of Statistics, with such experts as Adolphe Quetelet of Belgium’s National Observatory, Ernst Engel of the Prussian State Statistical Bureau, Alfred Marshall and Francis Y. Edgeworth of Cambridge University and Willard C. Brinton of the American Society of Mechanical Engineers, could not agree on standardization. Yet, although a unified theory did not emerge, the conclave increased the sensitivity of the problem for both statisticians and business leaders. ¹⁰

Interest in graphing took a stronger hold in the Bell System’s corporate accounting department with the formation of a specialized statistics division in 1909 under the leadership of Malcolm Churchill Rorty. Although the scope of graphical analysis increased during Rorty’s tenure (lasting until 1917), the firm did not adopt formal methodological policies at this time. This changed in the 1920s with reorganization of the statistics division by its new director (and eventual firm president), Walter Sherman Gifford. A gifted financial analyst and system designer, Gifford was well prepared for the task of statistical reorganization because of his experience in World War I as Chief Statistician for the War Industries Board (headed by Bernard Baruch) and his membership on the Council of National Defense. His goal was to improve the quality and comprehensibility of vital information and to reduce transaction costs by increasing the efficiency of graph preparation through standard practices. Gifford increased the division’s reporting responsibilities and expanded the staff to over 100 professionals.

It is essential to recognize that AT&T was a firm founded by engineers familiar with mathematical problem-solving. The statistics department, as overseen by Gifford, represented management’s attempt to employ mathematical and graphical methods to understand invisible business processes. At this time, the firm employed statistics and graphs as tools of discovery and efficiency. Presented with a complex business environment, the firm wanted to discover causal relationships between input and performance amid the volume of accounting information – a form of data mining. Graphs helped uncover meaningful relationships among diverse data. Graphical representations became a time-saving, efficiency enhancing device as they quickly converted numerical sequences into meaning, in much the same way that the ubiquitous pocket calculator would provide calculative shortcuts over a half-century later.

AT&T relied upon firm-wide conferences, begun in the 1880s, to standardize practices. Originally considered informal exchanges between the statisticians and the consumers of their information from operating departments, these conferences evolved into a platform for Bell’s management to lay out standard routines and procedures that the operating subsidiaries were compelled to adopt (Lipartito, 1989). ¹¹ In 1921, at the General Accounting Conference (GAC) the parent company specified both the types of requisite information and reporting formats (Belcher, 1921; Crunden, 1921; Richardson 1921). The emphasis on graphic construction was continued two years later, at the next set of conferences (Richardson, 1923).

Other corporate media also supported graphic standardization. The statistical department provided guidance about its new practices through articles in its technical bulletin. The Bell System Technical Journal published articles such as “A Method of Graphical Analysis” (Bateman, 1922), and “Some Applications of Statistical Methods to the Analysis of Physical and Engineering Data” (Shewhart, 1924). More general periodicals including Long Lines and the Bell Telephone Quarterly produced overviews of standardization efforts throughout the Bell System (Osborne, 1929a, 1929b), and “Making Facts and Figures Talk Plain English” (Grimes, 1924) introduced graphing concepts and techniques to a lay audience. These communications were an important extension of Vail’s “universal service” objective by supporting shared knowledge and an integrated learning base across the firm.

The repetition of common topics shows how important standardization was throughout the Bell System. The smallest details of operations were managed. ¹² Bell’s management recognized that in such a large firm, geographically dispersed and with many different divisions, separate cultures (with associated practices and measurement norms) could arise. If this happened, conflicts of interest could develop which would cripple the firm. Dissimilar reporting practices would create internal informational asymmetries and perceptions of uncertainty that could undermine central management’s efforts to coordinate and control the interdependent elements of the giant enterprise. Standardized reporting tools, the creation of common institutions and a shared sense of identity across the firm militated against this.

In its use of graphical presentation, the Bell System was a leader, but not alone. Joann Yates (1985) has documented the beginning use of graphics as an analytical tool at DuPont to 1919. Goodyear Tire and Rubber Co.’s accounting department was also experimenting with graphs at this time (Bloor, 1921). In documenting its preferred graphical methodologies, AT&T may have been influenced by the efforts of the American Statistical Association (ASA). The ASA published its “Preliminary Report on Standards for Graphical Presentation” in 1915 with the aim of benefitting mankind by increasing the “speed and accuracy with which complex information can be conveyed” (ASA, 1915). AT&T’s management perhaps was not as concerned about benefitting mankind, but it certainly saw graphics as a better way to manage its business.

AT&T’s management faced the problem of developing knowledge to better understand its business while hurdling over scale, scope and complexity. The firm operated across the continental United States through many entities, providing different lines of service to its diverse customer base. Expansion and upgrades continually required the firm to raise funds for capital outlays. Quantification of operating data helped promote efficient summarization, but numerical series could be difficult to understand. Graphs increased the comprehensibility of statistical data. The flexibility of graphic design increased understanding through the integration of information from multiple factors.

Standardization of graphical practices required internal modes of communicating best practices, achieved through firm-wide conferences and the publication of statistical bulletins. AT&T’s Statistical Bulletin No. 11 (AT&T, 1928) served as a primer to promote graphical regularization and more effective design. This and other bulletins were the primary mechanism for educating employees firm-wide; they provided general principles of graph construction as well as detailed examples of their applications. The bulletins also pointed to possible misuses or abuses of graphical representation and cautioned preparers and users about their suitability in specific contexts. One general rubric emphasized simplicity and visual appeal in order to enhance comprehensibility. While recognizing the value of single data series for the depiction of seasonal fluctuations, the bulletin stressed showing several related curves on the same grid to build up informational richness and reveal correlations. Some of the rules seemed very narrow, such as the requirement that a zero value always appears on the horizontal axis of a graph. For the most part, however, design advice pertained to more general issues. This was apparent in the definition of the broad classes and applications of generic graphs acceptable for corporate purposes.

Bulletin No. 11 provided an overall framework of acceptable graphic methods. It illustrated the applications and interrelationships among variables. Exhibit 1 identifies the three major categories of graphs and their interrelationships: (I) time-trend charts; (II) frequency charts; and (III) bar charts. In the remainder of this section, we discuss these three main categories and their variants. We show how these graphical techniques were applied in managerial accounting as gleaned from presentations at various internal conferences conducted by the firm in the 1920s. These included the General Accounting Conferences, the Plant and Engineering Conferences, Traffic Conferences, Personnel Conferences, State Managers’ Conferences, and regional conferences such as the Southern Bell Telephone and Telegraph Company Plant Conferences. We then discuss standardization with respect to the physical construction of various graphs as disseminated through Statistical Bulletin No. 5.

OPEN IN VIEWER Exhibit 1.

While these bulletins were created for internal users, the graphical work of the statistics department was not confined to the firm; it was disseminated to the broader business world. Bell engineer H.F. Dodge gave several papers on graphic presentation to the American Society of Mechanical Engineers (Dodge, 1930, 1931) where he discussed problems associated with graphs and solutions recommended by his experience. Though he never presented the methodologies as standards developed by AT&T, he did reference the statistics department as the source of much of his material. Doubtless, employees transferring to other firms from AT&T also brought the graphical methods they had learned. Thus, through formal and informal channels, the Bell System’s graphical methods moved towards becoming business norms. Several graphical methods that the firm developed are discussed below.

Time-trend charts

Time-trend charts were favored heavily to depict comparisons such as those between planned and actual performance in financial budgets (called “provisional estimates” in the Bell System). ¹³ A second time-trend variant depicted how one time series exerted force on another related activity such as changes in new plants per station against total plant investment per station. A third category contrasts current and previous performance for a given series, such as the number of toll calls per day contrasted with a device such as a “superimposed curve” to reveal past patterns. These analyses assessed enterprise performance and compared actual to budgets. For the fast-growing telephone business, installation of stations (telephones) was the main revenue driver, thus tracking this growth was vital for both projecting trends and capital budgeting. An example is seen in Exhibit 2 from an engineering conference with comparative information on the number of stations and net station gains (Gherardi, 1929).

OPEN IN VIEWER Exhibit 2.

Because of the risk to the firm associated with the borrowing needed to construct new facilities, it was critical to monitor actual installations closely to the budget; significant deviations would have important implications for firm finance. The chart also presents a visual comparison against the previous trend of high growth (1924–27). Such comparisons signal whether past trends are continuing or whether the firm may have reached an inflexion point.

Time-trend analysis also used data smoothing techniques to eliminate random or seasonal fluctuations that might mask important long-term trends (AT&T, 1921). Smoothing constrained the adverse effects of statistical noise, or unusual fluctuations, due to extraordinary items that impeded visualization of central tendencies. Another smoothing mechanism involved the cumulative curve chart that portrayed successive cumulative totals in a given data item from a specific date. This also smoothed out unusual fluctuations and permitted a better picture of the general trend.

Relationships between variables provided useful management accounting information and were represented by a double-vertical-scale chart or the secondary-scale chart. These blended two series in a single chart. The interrelationships between the two component elements were shown by using a common attribute such as dollars, or as a percentage of a common base measure. Exhibit 3 relates “Net Telephone Earnings” with a band of curves depicting 3 per cent to 8 per cent of “Average Plant Assets in Service” to reflect the range of values of telephone earnings as a percentage of plant assets. So, “Net Telephone Earnings” could be read both in dollars and as a percentage of “Average Plant Assets in Service”. In constructing these types of charts, Statistical Bulletin 11 warned that the interpretation of the charts could be influenced by the scale used. The charts may also incorporate a logarithmic scale to convey change in rates rather than amounts.

OPEN IN VIEWER Exhibit 3.

Exhibit 3 has special importance for both financial and regulatory purposes. The graph basically allows comparison of annual net telephone revenues against returns on total assets developed for a base period. This places annual earnings against a range of returns that the firm could reasonably expect to achieve on its average asset base. It is a visual indication of management’s ability to optimize utilization of assets. Beyond investors, such a display could be useful for addressing regulators’ questions about the reasonableness of returns based upon prevailing rate structures.

Surface charts were time-trend charts in which the space between a continuous series curve and the horizontal axis was shaded to form a two-dimensional surface. Bulletin 11 notes that, “this shading emphasizes the relation to the base line and creates a stronger impression of quantity than would the curve alone” (AT&T, 1928, p.9). The visual impact provided by the “bold relief” that the shading provides was emphasized. A zero line was necessary in these cases to avoid communicating “an entirely false impression of the magnitude and relative change of the series” (AT&T, 1928, p.9).

In a strata chart, value layers were stacked on top of each other. Each layer was measured from the top of the layer immediately below. Exhibit 4 shows a strata chart that decomposed plant additions for the Bell Operating Companies and its component parts used in reviewing production performance at the Plant and Engineering Conference in 1926 (Burcher, 1926). Strata charts could also be structured using a percentage scale. For instance, revenue could be broken down into its proportional components.

OPEN IN VIEWER Exhibit 4.

From a business perspective, Exhibit 4 shows changing patterns in main categories of costs constituting what the firm called “outside plant”, a category that would be expected to vary directly with changes in the intensity of system utilization. Outside plant included telephone poles, underground cable, aerial cable, and toll aerial wire. The chart design effectively ties past performance to future expectations. The actual spending trend for these four asset groups are extended from 1925 to 1926. Importantly, though prepared by analysts in 1925, this graph projects a significant downturn in investment needs by 1929.

Column charts employed vertical bars or columns to place in sharp contrast the changes in sequential values on discrete data points. A zero line was important in the design of these charts because “the eye seeks the zero line as the base of comparison much more surely in the case of columns than in the case of curves” (AT&T, 1928, p.12).

In general, the column chart was not subject to misinterpretation because its component sections were directly proportional to their values. Composite-column charts and their variants yielded better results than strata charts in representing time-series trends for components elements. Exhibit 5 is a deviation-column chart used to present the annual revenue effects of rate changes for exchange, toll, and the total for the years 1916–25 along with the budgeted amounts for 1926 at the Plant and Engineering Conference in Shawnee (Wilson, 1926).

OPEN IN VIEWER Exhibit 5.

From a business perspective, Exhibit 5 communicates the vital significance of the rate changes imposed by regulatory boards which affected the firm’s economic well-being. Besides breaking down the net impact of rate changes, this Exhibit further subdivides results between exchange or the local calling rates, and toll, or the rates for connections outside the exchange. Each segment is further shaded to indicate the dollar effect of both rate increases and decreases, as well as their overall net impact by year. Under “program”, the chart also projects the likely results of both impending and contemplated rate changes for the next fiscal year.

Bell statisticians also experimented with combined forms that blended graphing to increase comprehension. One such example was a chart that overlay total physical property (bar chart) with funded debt (line chart). The combined chart should be rooted in simple graphical forms and not deviate from “the fundamental principles of presentation” (AT&T, 1928, p.16). Suggested applications included comparing actual and budgeted expenses in which monthly and cumulative figures shown on the same grid are distinguished by the use of different graphic forms.

When superimposed graphic forms were confusing, Bell statisticians resorted to parallel presentation of separate grids. As illustrated in Exhibit 6, Southern Bell provided parallel presentations relating to the number of its employees eligible to receive pensions and its pension outlays over time (Sharp, 1926). Exhibit 6 also incorporates superimposed charts to provide additional information about pensioner service and disability.

OPEN IN VIEWER Exhibit 6.

Exhibit 6 communicates the growing deferred pension liability of the Bell System between 1913 and 1925. The pension plan, first launched in 1913, was considered a critical benefit that could help reduce the high cost of employee turnover. The liability mounted steadily during this period because of higher wages, growing workforce numbers, and improved retirement benefits. The graphs provided information about both the firm’s total annual pension costs and the population of retirees among southern subsidiaries. Comparisons are provided between the average pension payments for retirements and disabilities between the southern district and the entire Bell System.

Frequency charts

Frequency charts represented the second major graph category detailed in the statistical bulletins. Unlike time-trend charts, the horizontal scale of frequency charts allowed for the portrayal of the distribution of groups or class intervals within a population. Examples of their use at AT&T include depicting the number of shares per shareholder or their age intervals; these types of graphs were consistently included in the annual reports of the Bell Telephone Securities Company. ¹⁴ The vertical scale presents the relative frequency expressed either in terms of numbers or percentages as, for example, the percent of employees falling within each age interval. While many data arrays are continuous, such as the summarization of employee age segments, a histogram or “staircase” configuration could be used for discrete representations. This method was not considered useful, however, for comparing two frequency distributions. Instead, the alternative preferred was the employment of a continuous curve formed by reducing the intervals that separate each composite class.

In constructing these charts, percentages could be substituted for actual frequency counts, as is seen in Exhibit 7. This chart shows graduated life tables, or mortality tables, for various classes of fixed assets (aerial cable, underground cable, telephone poles and private branch exchanges). The mortality curves showed the percentage of assets still in service after up to 45 years. Smoothed curves have been fitted over the data points. The graphs helped illustrate the actual life experience of long-lived assets and thus their depreciation. Depreciation was a significant operating expense that could be recovered by rates and therefore needed to be defended before regulatory agencies.

OPEN IN VIEWER Exhibit 7.

Comparative bar charts

Comparative bar charts, unlike time-trend or frequency charts, had only one numerical scale, with the lengths of the bars proportional to their comparative values. Horizontal columns were considered to be more effective than vertical because they facilitated the parallel depiction of supplemental information. Because of their single scales, bar graphs were unsuitable for time series analysis. Moreover, because the width of the bar lacked interpretive significance, Bell statisticians did not include any secondary information within these areas that might cause confusion. These charts were prized for their ability to summarize significant information and provide a comprehensive picture of operations of a large, complex business. The company used this graphical form to depict the total number of operating problems experienced in the different parts of the exchange plant in the nation’s 30 largest cities during the fourth quarter of 1925 (Burcher, 1926).

Graded bar charts were used to depict bars in either ascending or descending order. Exhibit 8 illustrates the use of these for comparing plant repair expense (Burcher, 1926). Its principal advantage derived from the fact that “the graded bar chart directs attention primarily to the top and bottom items of an arranged list and so meets an important requirement of business graphics by emphasizing outstanding situations” (AT&T, 1928, p.21). Such an arrangement also permitted, by distinctive shading, the setting off of medians and quartiles. Although frequently useful in contrasting data, Bell statisticians remained concerned about the statistical validity of such comparisons.

OPEN IN VIEWER Exhibit 8.

Because of its large size, the firm segmented its business into regional subsidiaries, all of which provided the parent company with a full array of accounting and statistical information. Exhibit 8 employs a modified bar chart to indicate the relative differences in average plant expense ratings for 1924 and 1925. As a regulated utility, such cost information was important because it evinced how well the firm’s subsidiaries were satisfying public mandates for economical service. The low-cost rating of New York reflects the high traffic density and the facilities’ utilization in heavily populated districts. In contrast, the highest operating expenses were incurred in the sparsely populated mountain states. Information about these cost factors was central to the implementation of the firm’s universal service strategy.

The bar chart also evolved into many hybrid forms. A bar-and-symbol chart allowed insertion of supplemental information on a bar chart by using a symbol such as a dot or diamond. This facilitated the making of comparisons with prior years (as in Exhibit 8), or with forecasts. Two-directional bar charts better served cases using elements derived from different scales or those in which favorable–unfavorable differences predominate. Exhibit 9 is a two-directional chart for measuring revenues, expenses and net revenues used by Southern Bell (Southern Bell Telephone and Telegraph Company, 1929).

OPEN IN VIEWER Exhibit 9.

A major focus of what eventually became known as commercial engineering at the Bell System necessitated the creation of additional revenue through the sale of advertising in telephone directories. Exhibit 9 documents the increasing profitability of this advertising during the period 1919–27. Through 1922, this service proved to be only marginally remunerative, but the sales department succeeded in materially increasing revenue and profits during the economically expansive period of 1923–27. The chart is divided into two components to illustrate graphically the trends in both gross revenues and expenses. A portion of the revenue bar was blackened to indicate the net revenue earned each year.

Graph construction

Through internal documents such as Statistical Bulletin No. 5, the Bell System also specified in great detail the methods to be employed in constructing charts, choosing grids for chart preparation, standardizing titles, scales and captions, and selecting gummed tapes and other materials to plot curves and mark off areas (AT&T, 1922). Through the special properties of the five standard grid alternatives, chart makers had some degree of flexibility in planning document layout. Bulletin 5 not only provided detailed descriptions of specific grids but also identified the ideal circumstances for their use. For instance, Grid No. 2 was considered appropriate “when it is desired to compare current figures with the corresponding figures for the previous year, and where seasonal variation is not as important as the trend of the curve” (AT&T, 1922, p.3).

The firm provided close guidance on standardizing titles, captions and scales used in chart construction. Sheets containing words that appeared most frequently in various parts of the chart were pre-prepared using plain block letters with different fonts “in order to insure the greatest legibility” and “to give proper balance between the various parts of the chart” (AT&T, 1922, p.4). The six sheets of words were grouped as Title Words (Sheet A), Title Letters, to enable formation of words not contained in Sheet A (Sheet B), Vertical Scale Captions such as thousands of dollars or number of stations (Sheet C), Curve Labels, such as total toll revenues or provisional estimate (Sheet D), numerical Vertical Scales, printed in strips which could be cut and pasted flush with the edges of the grid (Sheet E), and Time Scales representing years or months that were pre-formatted for use with specific grids. When the standardized sheets could not be used, the firm turned to draftsmen who were required to use Payzant pens of a specific size. A No. 2 Payzant pen was mandated to create title words which should be 0.45 inches in height and a No. 5 Payzant pen for vertical scale captions with letters that were 0.3 inches in height.

Prepared materials for use in constructing curve charts and areas were also itemized. The use of gum tape in four widths – ³/₃₂, ⅛, ¼ and ⅜ of an inch – was advocated, with the two narrower ones available for use in forming curves and the two wider ones for preparing column charts. It was suggested that dash lines could be formed by cutting the tape. Two cross-hatched surfaces were prepared for use in constructing area trend charts. A single cross-hatched surface could be used to form a third area when turned at right angles.

The firm also specified the steps for constructing a chart using prepared or standard materials (see Exhibit 10). These included: (1) selecting and preparing the grid appropriate for the planned presentation; (2) plotting the curve; (3) applying the curve using gummed tape, rulers or ribbon pens; and (4) applying titles, captions and making slides. The assembling of charts from standardized materials was called the “cut and paste method” and was believed to be superior to typewritten or hand constructed presentations for several reasons. It helped reduce the cost of constructing these charts in two ways: (1) by reducing preparation times; and (2) by eliminating the need for skilled draftsman. Additionally, their use increased the speed with which charts could be produced. Furthermore, the charts obtained by this process would be uniform in appearance due to standardized curves, titles, scale captions, and other attributes. Finally, this process provided greater flexibility in introducing any revisions.

OPEN IN VIEWER Exhibit 10.

Conclusion

The goal of graphical representation at AT&T accords with the contemporary definition of managerial accounting as the codification, abstraction and diffusion of information to users (Bhimani & Roberts, 2004). This is precisely the activity the AT&T accountants undertook when they deployed graphical methods throughout the firm, even if contemporary usage did not include the term managerial accounting to describe their work. The standard graphical representations they developed were part of a process whereby cost accounting became managerial accounting, accounting information used in decision making.

Managerial accounting innovation, then, occurs when there is a need to organize, synthesize, or communicate such information better. Such was indeed the case at AT&T during the 1920s as the firm experienced increasing growth and expansion accompanied by heightened economic uncertainty. Devising new modalities for the creation and use of managerial accounting information was imperative to managing the firm successfully. While the accounting function identified, recorded, tabulated and communicated information to various users, the analytical value of information was impeded by the increasing volume of information flow in the organization. Statistics and graphical representations aided accountants’ attempts to organize, synthesize and communicate information that could clarify relationships between inputs and outputs of the firm, highlight significant relationships or trends, to predict future results, and ensure that there was a degree of uniformity in the depiction and interpretation of information.

By creating and disseminating standardized graphical modalities, the firm sought to perpetuate a shared language and meaning that could be used in internal communication, management control and firm planning. Such standardized communication cut across all functions like budgeting, planning and performance evaluation, and functional areas such as production, marketing, personnel and finance. This firm-wide approach was intended to increase transparency and was a significant step to reduce internal informational asymmetry, a major challenge in the control and coordination of such a large, decentralized and complex firm.

Standardized graphical communication was also a key to institutional learning. In times of rapid growth and technological and economic uncertainty, management required new aids for decision making, leading to an explosion in managerial accounting knowledge in the early part of the twentieth century. AT&T, founded on technological innovation, turned naturally to science and mathematics to solve administrative and operational problems. Statistical tools provided insights into data analysis and control resulting from the firm’s rapid growth. As the firm rolled out new methods of measuring results and predicting future developments, management, in an attempt to create stability, codified and standardized the new practices. Standard methods were developed for deriving meaning from the reams of information on resources, external economic conditions, and operating results produced by departments, divisions and subsidiaries. Management used graphical representations to uncover and highlight important relationships among this mass of information – to make visible the invisible.

Firm-wide conferences were extensively used to disseminate knowledge about new managerial accounting techniques and graphical communication throughout the firm. While organizers of these conferences spoke about the importance of bringing advances in knowledge from operating companies to the attention of the firm overall, these conferences and related literature primarily ensured the dissemination of standardized methodologies and formats. Standardization saved time and money, and helped immensely in coordinating and evaluating geographically dispersed activities. By employing uniform goals throughout the firm, AT&T also limited future conflicts about what the goals of its business should be. The firm emphasized financial performance measures (as opposed to customer satisfaction, production throughput, or quality control) as the primary metrics to measure success. ¹⁵ The messages with respect to graphical tools were consistent; performance results were measured financially, and the standardized measurement and representation of these results was a priority.

The rapidly developing role of graphs in communicating operational results had import beyond the design of reports; it also influenced the composition of employees. In some cases, the standardized reports diminished the need to find employees with specific skill sets, as when the need for trained draughtsmen to create charts was diminished by the creation of a “standard” tool kit included as part of Statistical Bulletin 5. This is an instance of standardization leading to the “deskilling” of labor, allowing management to gain the benefits of knowledge once owned by skilled laborers. However, the adoption of new management accounting tools also required an increase in some skill levels among employees. For instance, it was recognized that use of statistical tools required accountants, who, if not experts in statistics, would not be intimidated by their use. The analysis of the firm’s operational information demanded that firm accountants expand their skills beyond familiarity with accrual accounting; to help manage the technically innovative firm, accountants would be required to innovate when evaluating managerial accounting measures such as costs and their relationship to revenue. The scope of managerial accounting was thus expanded beyond simple recording of data to providing analytical and inferential support to managerial decision-making.

Our paper also places in sharp perspective the differences in understanding the ordering of corporate information in an economic rather than historical analysis. The economic perspective, which strives to develop broad scientific principles, tends to overlook the dynamic nature of system evolution and the differential effects of environmental factors. While such theory is useful in identifying the benefits from knowledge standardization through graphic display, the development of these capacities within the firm occurred over a long time horizon and responded to many particular circumstances that affected the growth of the firm, such as changes in the nature of markets, technology and regulation. Because of this, one of the important lessons that we derive from historical analysis is to recognize that companies’ effective information systems respond to the constant flux in economic environments. Moreover, the focus on firm histories also provides a useful tool for comparative analysis to reveal the role that information policies played in shaping the success or failure of competing units in an industrial sector. The explanatory power of such industry-based cases is high because of the oligopolistic nature of many of the most dynamic sectors of contemporary economies. The study of giant, dominant organizations through historical cases can yield a rich and extensive body of evidence to support the construction of dynamic theories of economic and accounting change.

The Bell System’s experience provides insight into some of the industrial origins that encouraged the later development of cybernetics and electronic information processing. Information automation enhanced the ability of managers to confront the daunting problems of giant business scale, scope and complexity, much like the graphical approaches at firms like AT&T before the electronic age. Greater efficiency in the mining and conversion of data enhanced competitiveness by making available to managements more reliable information about their firms and the socioeconomic contexts in which they operated. In the 1920s, giant enterprises were becoming deeply involved in learning how to create more effective systems for gathering and utilizing firm-specific knowledge. From an historical standpoint, the new constructs that emerged during this process are an important focus of scholarly inquiry because they directly shaped managerial cognition. Thus, the analysis of these intellectual constructs becomes vital in understanding the perceptions of business leaders and how such a world view affected decision processes.