In 1987, the AICPA published a management advisory service special report entitled, An Introduction to Artificial Intelligence and Expert Systems. The goal of the report was to present information about expert systems, including: basic components, opportunities for the accounting profession, ways accounting firms might use expert systems, issues to be considered, and prospects for the future.

One section of the report described applications suitable for future expert system development. The report suggested that expert systems would be developed to aid in making decisions concerning the correct accounting treatment for complex accounting transactions, such as leases, foreign currency exchange, acquisitions, pensions, and income taxes.

Benefits of Expert System-Based

Decision Aids

Normally, a CPA in industry who becomes involved with a complex accounting issue for the first time must research current accounting pronouncements. An expert system-based decision aid captures all available information in its knowledge base, and that knowledge would be readily available for use.

Another advantage of an expert system-based decision aid is that such tools promote consistent and uniform performance of accounting tasks. Even if a firm or company is occasionally exposed to a complex lease transaction or a business combination, this does not ensure that similar transactions will be dealt with in the same manner each time. With an expert system-based decision aid, this would not happen, since a lease that should be capitalized or a business combination that should be accounted for as a pooling of interests will be consistently identified.

Another benefit available is to use the expert system to train accountants for special or unique tasks.

Expert System Shell Software

Developing an expert system-based decision tool is not as difficult as most people imagine. This is because of the existence of expert system "shell" software packages that allow users without considerable computer experience to develop their own personalized expert systems. A shell is an "empty" application consisting only of interfaces and an inference engine, without any built-in expert knowledge in the knowledge base. A special routine in the software allows users to input the shell's knowledge base. Figure 1 lists several popular expert system shell software packages currently available.

Table : FIGURE 1

EXPERTSYSTEMSOFTWAREPACKAGES

Product(*)SystemSpecificationsListPrice

Goldworks5MBextendedmemory,$7,500

GoldHillComputers10MBharddisk

26 Landsdowne St. Cambridge, MA 02139 (617) 621-3300

MicroExpertand640KRAM$69

KnowledgePro640KRAM$495

Knowledge Garden Inc. 473A Malden Bridge Rd. Nassau, NY 12123 (518) 766- 3000

GURUFirstStepand640KRAM,5MBharddisk$895

GURUSolveur640KRAM,5MBharddisk$4,995

mdbs Inc. P.O. Box 5268 Lafayette, IN 47903 (800) 344-5832

VP-Expert 384K RAM $249

Paperback Software International 2830 Ninth St. Berkeley, CA 94710 (415) 644-2116

(*) Some firms offer more than one expert system shell software package.

One expert system shell is VP-Expert, developed by Paperback Software International of Berkeley, California. The VP-Expert includes an "Induce" command that automatically creates a knowledge base from a table contained in a text, database or worksheet file. Optional development windows allow the user to observe the behind-the-scenes path of the inference engine as it navigates the knowledge base to solve problems. VP-Expert makes use of simple English rule construction and provides the ability to explain its action during a consultation. An interface can create "if/then" rules directly from decision tables in external files. Confidence factors (from zero to 100) can be assigned to each rule and response. More information on VP-Expert is available in a book, VP-Expert for Business Applications, published by Holden-Day, Inc., of Oakland, California.

Examples of Expert System-Based

Decision Aids

The AICPA has suggested that expert system-based decision aids will be developed to solve various accounting problems. At least two such decision aids have been constructed. One determines the accounting treatment for leases by the lessee. (1) The other determines the accounting treatment for business combinations

Lease Expert System. In a study by Boer and Livnat, an expert system was constructed to determine the accounting treatment for leases by the lessee. The system was used as a teaching aid by the researchers. The students in one accounting class used the expert system and textbook material, while another class used only the textbook material in solving problems pertaining to leases. The students that used the expert system made substantially fewer errors in classifying leases as either capital or operating leases than did the students who used only textbook material.

For Business Combinations. Another researcher constructed an expert system-based decision tool that determines the accounting treatment for business combinations. (2) A business combination is either a purchase or a pooling of interests. APBO 16 delineates 12 criteria that must be satisfied in order for a combination to be accounted for as a pooling of interests. All 12 criteria must be satisfied in order for a combination to be accounted for as a pooling of interests. If even one criterion is not met, the combination must be recorded as a purchase. Figure 2 shows the decision steps for the business combination expert system-based decision tool.

PHOTO : FIGURE 2

DECISION STEPS FOR THE BUSINESS COMBINATION EXPERT SYSTEM

The expert system-based decision tool for business combinations is named PURPOOL. PURPOOL was developed using VP-Expert. Using if/then rules based on the criteria contained in APBO 16, PURPOOL can assist the accountant in determining whether a business combination should be accounted for as a purchase or as a pooling of interest. (A prototype version of PURPOOL can be obtained from Dr. Steve McDuffie, Southwest Missouri State University, Department of Accounting, Springfield, MO 65804-0094.) The APBO 16 criteria and related PURPOOL rules associated with decision steps (4a) and (4b) (of Figure 2) are shown in Figure 3.

PHOTO : FIGURE 3

CRITERIA AND RELATED PURPOOL RULES

One of the more complex criterion contained in APBO 16 involves the acquisition of the combiner firm's (acquired firm) stock or net assets. This particular criterion is addressed by PURPOOL in its Rules 9 and 10. The issuer firm (acquiring firm) must either acquire at least 90% of the combiner firm's outstanding common stock or all the combiner firm's net assets. On the surface this does not appear difficult or complex. However, APBO 16 specifies that the issuer firm's common stock owned by the combiner from reduces the shares that are considered in meeting the 90% level of ownership. This complicates the situation considerably.

If one assumes the following: * Combiner Co. common stock outstanding = 20,000 shares; * Combiner Co. common stock acquired by Issuer Co. = 18,500 shares; * Issuer Co. common stock issued to finalize the combination = 9,250 shares; and * Issuer Co. common stock owned by the Combiner Co. = 500 shares.

This business combination appears to satisfy the criterion as a pooling of interests because: 20,000 shares x 90% = 18,000 shares and the issuer firm acquired 18,500 shares of the combiner firm's outstanding common stock, which represents more than 90% of the firm's outstanding common stock. However, the issuer firm's common stock owned by the combiner firm must be considered because these shares reduce the shares of stock that are used to satisfy the 90% level of ownership. The significance of the issuer shares owned by the combiner firm depends upon the stock exchange ratio between the two companies. Following a closer examination, the criterion from APBO 16 is shown to be unsatisfied and thus the combination should be accounted for as a purchase. The calculations supporting this conclusion are provided in Figure 4.

TABLE : FIGURE 4

BUSINESS COMBINATION CALCULATIONS

Expert Systems are Proliferating

Expert system-based decision tools have been developed to help solve complex accounting issues regarding leases and business combinations. Numerous other applications of expert system-based decision tools are possible, for example: 1) to determine primary and fully diluted earnings per share figures; and 2) to deal with the many complex issues associated with pension accounting.

Expert system shell software packages, such as VP-Expert, allow the accountant to efficiently design his or her own expert system-based decision tools. VP-Expert is relatively inexpensive and user-friendly. There are a number of similar software packages available.

As expert system-based decision tools are developed, they will offer opportunities to both public accounting firms and industry firms for increased productivity and better use of resources. Companies that develop or acquire decision tool of this type will probably engender a competitive advantage over other companies. These tools can be used to help solve actual accounting problems as well as to help train inexperienced accounting personnel.

Endnotes

(1). Boer, G. and J. Livnat (1990), "Using Expert Systems to Teach Complex Accounting Issues," Issues in Accounting Education, Spring, pp. 108-119.

(2). McDuffie, S. (1990), Construction and Validation of an Expert System Prototype to Determine the Accounting Treatment for Business Combinations, Dissertation, Louisiana Tech University.