The decision support systems of the '80s are alive and well, masquerading as smaller, personal support systems.
The early 1980s were an exciting period for a new type of information system called decision support systems (DSS). These innovative systems were expected to aid management in making non-recurring, complex, unstructed decisions. But the excitement has diminished. Very little has been written about the use of these systems in recent years. Is the concept of decision support systems dead, buried by newer information technologies? Or are DSS now so commonplace that we take them for granted?
What are decision support systems?
Decision support systems were originally defined as interactive, computer-based systems that could be used to support complex, non-recurring decisions made by senior managers. These systems were a combination of hardware and software built to support specific top-level strategic decisions. Whether to build a new plant, or expand the company's fleet of aircraft, were considered the kinds of decisions DSS would support.
The essence of this definition has remained the same over time but the scope of application of these systems has greatly expanded. DSS are still interactive, computer-based systems that support the decision maker. They still provide relevant information to the decision maker rather than actually make the decision. But DSS are no longer used only for unstructured, one-time decisions. Their greatest use today is for spreadsheet-based support systems for semi-structured, recurring decisions such as periodic budgeting decisions. A characteristic of these systems is that they can be easily modified to analyze information based on changing circumstances.
We still think of DSS as comprising three main components: a model base, a data base, and a user interface.
The model base consists of a set of mathematical models such as linear programming and statistical models (regression, analysis of variance) that can be used to analyze raw data into useful information. A model base management system enables the DSS developer and user to retrieve, modify, add, and delete models from the model base.
The second component of DSS, the data base, includes not only data from internal sources (usually the accounting systems) but also data from external sources. These external sources include data from government, industry, and consulting sources. A data base management system allows the user to manipulate and organize data in the data base so that it can be input to the decision models.
The third component is the user interface. DSS are characterized by direct manipulation of the data and models by the user. This means that the interface must by easy to use and easy to learn. Today a variety of user interfaces is available including graphics, windows, menus, and command interfaces.
The three components of a DSS distinguish these systems from other systems found in organizations today. Transaction processing and management reporting systems focus on the storage and selective retrieval of large amounts of internal data. Very little analysis is done with the data in these systems. Accounting information systems such as computerized G/L and payroll systems are examples of transaction-processing systems.
Executive Information Systems (EIS) are considered a form of decision support system. These systems are primarily used by top management to monitor and track the organization's performance. The modelling components of these systems are specifically designed to filter and summarize specific organizational data to support management control. For example, Kraft Grocery Products Group uses an EIS built with the Commander EIS software from Comshare. The system supplies Kraft executives with highly refined and summarized marketing and production data.
Current expert systems or knowledge-based systems use IF-THEN rules, as opposed to the mathematical models found in DSS, to provide advice in narrow problem domains. For example, Canadian National Railways uses an expert system built with the expert system shell, EXSYS, to aid mechanics in the diagnosis of locomotive engine problems.
The first DSS were large institutional-type systems that required significant organizational resources to build. Examples include American Airlines Information Management System (AAIMS) and Geodata Analysis and Display System (GADS) from IBM. But as information technology has changed, so has the nature of DSS.
Most systems being built today are small, ad hoc, personal or departmental DSS. These systems are usually built by an individual decision maker or by a DSS development specialist. The modelling component of these smaller systems is typically a spreadsheet-based model or simple mathematical model such as linear programming.
Development of decision support systems
The most significant factor in the development of DSS over the last ten years has been the improvement in the technology and tools to built these systems. In the early 1980s, the technology consisted of large mainframes with dedicated terminals. This usually meant scheduling of computer use and slow response times. Today DSS are typically built using powerful, micro- and minicomputer workstations that can be dedicated to running the DSS.
DSS tools such as programming languages, graphics programs, and statistical packages are much more sophisticated and powerful than they were in the early 1980s. These tools are used to build DSS generators and specific DSS. DSS generators are software packages that usually have the model base, data base, and user interface of the DSS already built in. DSS generators allow for the development of a DSS quickly and inexpensively. Widely-used DSS generators include Lotus 1-2-3, Quattro Pro, Interactive Financial Planning System (IFPS), and 4GLs such as Nomad 2 and Focus.
While DSS tools and generators have improved during the decade, the DSS development process has remained relatively constant. The process of developing DSS has been described as "growing" DSS. Because of the unstructured nature of the supported decision task, information requirements are difficult to define at the beginning of the process. What usually happens is that an initial, basic system is developed and evaluated (a prototype or version 0). Improvements are made to the system based on that evaluation. This process, called prototyping or iterative development, continues through multiple versions until the decision maker is satisfied with the outputs from the system.
What has made the prototyping process (we now have rapid prototyping!) possible is the nature of the tools and generators used. It no longer takes many lines of computer code and special computer programming expertise to develop most small DSS. A customized quality control DSS was recently developed at the Celanese Canada Plant in Millhaven, Ontario using the dBase data base management package and Statgraphics modelling package. The system analyzes the test data from a number of production processes and provides the quality control engineers and management with information on which to base production decisions. This system continues to evolve but the initial version was built in three months. This would not have been possible without the use of the new DSS tools and generators.
Applications of DSS in management accounting
Decision support systems, both large, institutional systems and small, ad hoc systems, are now a common tool of the management accountant. Their applications are many and varied. Budget and cost/variance analysis, discriminant analysis, break-even analysis, tax computation and analysis, depreciation methods and analysis, and capital budgeting are just some of the management accounting tasks DSS support. The following examples demonstrate the potential of these systems to support management accounting decisions.
The Empire Pencil Co. has implemented a DSS for manufacturing cost planning operations.[*] This system is an institutional system that allows managers in three departments to determine the maximum raw material costs to pay while retaining the required gross margin and competitive selling prices. The strength of the system is the manufacturing cost model that anticipates the effects of variable prices. The basic model aids managers in choosing the appropriate mix of manufacturing cost elements based on material price, scrap, capacity, and departmental efficiencies. The system took one year to develop and was built using a variety of DSS generators including Lotus 1-2-3.
A second example of DSS to support manufacturing costs decisions is the new systems to support the process of activity-based costing (ABC). These systems facilitate the development and use of a computer-based activity/cost model. A firm's cost data is input into the model and analyzed. The resulting output provides managers with the detailed information about how actual costs have been distributed and how they might be better allocated.
One such ABC decision support system is called "Net Profit." This DSS models the activity flows, allows dynamic costing, and attaches those costs to the flows. This DSS is typical of the newer, institutional DSSs that typically use mainframe-based data and micro- or mini-computer models. A number of Toronto area firms use this DSS to assist in making decisions about their manufacturing costs.
The third example is a DSS used by a small, independent food services company in Kingston, Ontario. The company operates a catering business, two restaurants, and a bakery. The DSS supports the daily (and weekly) planning and control decisions that are essential to small businesses of this type. The DSS is constructed using a number of spreadsheet models written in Lotus 1-2-3. Selected data is inputted from files from the firm's ACCPAC accounting system.
The DSS monitors actual performance of labor and materials, compares it with the budget, and then performs standard variance analysis and produces variance reports and screens. Formulas for variance have been developed over the last few years to reflect actual experience. Management primarily uses the variances by periods and/or food types. These variances have become essential information for the continued operations of the firm.
This example is typical of many accounting DSS in small businesses in Canada. Data is either entered interactively as needed or imported from files from an accounting system or other data base system. The model base consists of simple spreadsheet-based models for calculating variances, determining breakeven points, or projecting cash flows.
Current trends in DSS
Decision support systems have come a long way since their initial use in the 1970s (see figure). The technology "push" of new, powerful personal computers and sophisticated, yet easy-to-use software development tools has revolutionized the development and use of these systems. The diffusion of this technology has enabled management accountants to build and use their own DSS. It is now relatively rare for a specific DSS to be built by a large team of computer specialists. Institutional DSS are still being built but they are relatively rare.
Thus, the economics of developing DSS have changed. In the late '70s and early '80s, it was becoming increasingly difficult to justify the expense of building a DSS that would be used to support only one decision. The organizational resources to build such systems were usually substantial. But today with low cost, yet powerful DSS generators, the expense of developing useful DSS is substantially reduced. Therefore, many more, smaller DSS are in use. Most of these systems are initially built to support one specific decision (such as a capital budgeting decision) but are then modified to support other decisions. The bottom line is that there has been a dramatic shift to personal, ad hoc DSS and prepackaged DSS from large, institutional DSS.
The whole notion of decision support systems has also been expanded. These systems no longer need to be individual support systems. New systems called "Group Decision Support Systems" or GDSS are now being marketed commercially. These systems support teams of decision makers.
An example is TeamFocus from IBM. This GDSS has a number of tools to assist management teams in generating new initiatives, analyzing alternatives, and developing plans of action. For example, one tool in TeamFocus is called Alternative Evaluator. This tool allows each team member in a meeting to rate a number of alternatives against a set of criteria. The scores are instantly combined into a weighted team matrix that is then analyzed by the group. The tool helps the team achieve consensus on which alternative should be chosen.
Decision support systems are also being integrated into other types of information systems. Data base management systems are being combined with modelling systems to analyze selected data in the data base (such as the Celanese Canada example mentioned earlier). Executive information systems generators now include sophisticated modelling techniques to allow greater variety in the way information is shown to management. Finally, DSS are beginning to be used as "front ends" for expert systems. Raw data is analyzed using the DSS and the output forms the basis for "facts" used by the expert system. For example, the expert systems shell, EXSYS, can input data from a Lotus 1-2-3 DSS to be used in specific production rules.
In summary, the message is that decision support systems are alive and well. But they have changed substantially from the early days of large, institutional DSS. They are now much more pervasive as smaller, personal support systems. They are now supporting teams of decision makers and are also found as components of other information systems. It is clear, however, that DSS are still essential tools of management accounting. As the management accountant's role has evolved from information provider to information analyst and decision maker, effective use of decision support systems will continue to add value to his or her contribution. [Tabular Data Omitted]
[*]W. C. Austin and H. B. Eom, "Decision Support that Anticipates the Effects of Variable Prices," Financial & Accounting Systems, Vol. 6, Winter 1991.
Brent Gallupe, PhD, CMA, ISP, is associate professor of management accounting systems at Queen's University, Kingston, Ontario, and director of Queen's Executive Decision Centre.…