Expert systems in lease accounting (abstract only)

Abstract

The authors have developed a pair of complementary expert systems to be used to analyze lease agreements and determine the required accounting treatment in accordance with the rules in the Statement of Financial Accounting Standards No. 13, “Accounting For Leases.” The user interacts with these programs by providing the values for key variables from the lease agreement in response to program prompts. The user is informed as to whether the lease is a capital or an operating lease and the reasons to support the decision. Also, for capital leases the amount to capitalize and the amortization period are reported.This application is appropriate for implementation as a rule based expert system since much of financial accounting and reporting is governed by complex sets of rules that must be followed exactly. This type of expert system can improve the consistency of application of the rules and reduce the time it takes for CPAs to verify that the rules have been correctly applied.The systems were developed using the Texas Instrument's Personal Consultant Plus running on the IBM compatible Zenith Z158 microcomputers. Personal Consultant Plus provides many features which facilitate the development of rule based expert systems. For example, rules can be entered in a simplified rule language with natural language translations available to the user. Also, the building of the user/system interface is simplified through convenient user prompt and reprompt generation capability and through the automatic provision of facilities allowing the user to ask why a certain value is required or how the system reached a conclusion.The lease accounting system consists of two parts, one for making accounting recommendations for the lessee and the other for making accounting recommendations for the lessor. The two systems are very similar, both classify the lease as a capital or an operating lease and both provide recommendations. The lessor system further classifies a capital lease as a sales or financing lease and adds appropriate recommendations depending on the type.The systems together have about 70 if-then rules. For each system the parameters “lease type” and “recommendations” are established as goals. The systems use a backwards chaining approach to try to obtain values for these goal parameters. As data is required from the user for variables whose values cannot be deduced from the rules, the system prompts the user to enter the values. At this point the user can ask why the data is required or they can ask for an additional prompt. Depending on the lease being analyzed the user may be prompted for as many as 16 values during a consultation.“lease_type” is a single valued parameter whose value must be either “capital” or “operating”. “recommendations” is a multivalued parameter whose values are text strings describing the justification and accounting recommendations. “lease_type” will be set to “capital” if any one of four main conditions are satisfied. However, even if the first of the four conditions is satisfied the system is forced to check all four so that for each satisfied condition the appropriate justification and explanation can be assigned to the “recommendations” parameter. The system also assigns accounting recommendations appropriate for the lease being analyzed to the “recommendation” parameter.When the consultation is finished the values of each goal parameter are displayed. Since “recommendations” is a multivalued parameter each of the values assigned to it are displayed. The system controls the order in which values are assigned to “recommendations” so that the result has an understandable, easy to read format.The systems are very easy for the user to use and the user can conveniently request further explanation after reading the recommendations. Although these systems could have been developed in a traditional programming language using conventional programming techniques their development using Personal Consultant Plus was very quick and the incorporation of a user friendly approach to system/user interaction was easy.