Decision support software for tax.

The item "Decision Support Software for Tax" was written by Joe Davis, a Masters of Accountancy student with a concentration in tax, Virginia Polytechnic Institute and State University, Blacksburg, Va., and John Brozovsky, Assistant Professor of Accounting, Virginia Polytechnic Institute and State University, Blacksburg, Va. The item "Committee Addresses Concerns on Electronic Filing" was written by Joseph C. Maida, CPA, MBA, Shareholder, Nicholas C. Maida, CPA, Chartered, Trenton, N.J.

Computers have revolutionized every aspect of practitioners' lives, including the preparation and planning for income taxes. Over the years, tax software has expanded and taken new directions. Three areas of this expansion include decision support systems (DSSs), expert systems and neural networks. This article will examine these three areas as they relate to tax planning.

Decision support systems

A DSS is "an interactive, flexible and adaptable CBIS (Computer Based Information System), specially developed for supporting the solution to a particular management problem or improved decision making. It utilizes data, it provides easy user interface, and it allows for the decision maker's own insights." (See Turban, Decision Support and Expert Systems, Macmillan Publishing Co., 1993, at 87.)

DSSs provide support for decision makers mainly in semistructured and unstructured situations, i.e., in areas in which the correct decision rule is not known. (Most tax planning decisions fall in this category. This support may be provided for various levels of managerial skill and expertise, and may be provided to individuals as well as groups. DSSs support all phases of the decision-making process: intelligence, design, choice and implementation. A variety of decision-making styles and processes are supported, making DSSs adaptive over time. DSSs are easy to use and attempt to improve the effectiveness of decision-making, rather than merely its efficiency. Finally, DSSs lead to learning, which leads to new demands and refinement of the system, which in turn leads to additional learning, and so forth. Most DSSs have some of these features.

Using the definition from Turban, any computer software that supports the decision-making process can be considered to be a DSS. Decision support software in taxation can be divided into four categories.

Category 1: Simple tax preparation software. Tax preparation software packages are primarily used by individuals and tax preparers to prepare returns for individuals, businesses, estates and trusts, etc., that are filed with the Federal and various state tax authorities. While used primarily for preparation (which is not a decision support function), these programs can provide limited "what-if" calculations for future tax decisions. Typically, the planning is limited to either including or excluding an item or a group of items, and a recalculation of the new tax. The change in the tax liability is then calculated by hand as the difference between the taxes with and without the item in question. This type of software, therefore, has limited and very cumbersome decision support capabilities. Data for various alternatives must be inputted into the program one set of alternatives at a time. The returns can then be printed (or the tax liability recorded manually) and compared, thus assisting in the decision-making process. The disadvantages of this type of DSS are its limited alternatives and the extensive data entry required to compare these alternatives. Also, the final comparison must be by hand, and the various alternatives cannot be compared side by side.

Category 2: Tax preparation software with a tax planning feature in the same package. These software packages have the same capabilities as simple tax preparation software with the added feature of multiperiod tax planning. The major strength of these software packages is that, in addition to return preparation, they allow for multiple alternatives to be compared. After the data for multiple periods or alternative scenarios is inputted, the software calculates the tax for each and shows a comparison of each of the alternatives for decision-making purposes. Schedule 1, below, lists a subset of tax preparation software, its tax planning capabilities, some major features and list costs.

Category 3: Specialty decision support tax software. These software packages are true decision support packages. However, each supports a specific tax decision, such as estate and gift taxes, fixed asset calculations, interest and penalty calculations or loan refinancing analyzers. Unless a tax adviser has clients considering these specific transactions or with these specific needs, their use cannot be justified on a cost basis. Schedule 2, on page 56, lists some specialty tax software and the decisions they support.

Category 4: Tax planning software. These software packages are used only for tax planning or "what-if" analysis, and provide multiple period comparisons using multiple alternatives of tax data with no tax preparation features. Thus, they provide support for determining the tax effects of decisions without getting bogged down in the format required for filing various returns. Instead, they are concerned with presenting data in a format that can be understood and used by the decision maker in his tax planning. Schedule 3, on page 56, lists the tax planning software that was reviewed.

Expert systems

Expert systems are computerized advisory programs that attempt to imitate the reasoning processes and knowledge of experts in solving specific types of problems. They attempt to make or give advice concerning an unstructured or semistructured decision normally made by a human with special expertise and reach the same conclusions as the best human experts. Expert systems are becoming increasingly popular in many disciplines because of their ability to enhance productivity in decision-making in the less structured areas in which they are used.

Human experts possess several characteristics that help them to solve problems quickly. Human experts can learn from experience, change their point of view to suit a particular situation and are able to transfer their knowledge from one area of expertise to another. They are also able to judge the reliability of their own conclusions and know when they are unable to come to a conclusion and communicate their findings to other people.

Expert systems differ from conventional processing systems in several ways. Conventional systems are organized in a sequential manner. The program executes from top to bottom and the logic of the system is built using individual statements that are executed sequentially. Expert systems distribute their logic over numerous independent condition/action rules. As these rules are invoked, the results are stored in global working memory. The next step in the process is totally dependent on which rule has been invoked.

The article, "Expert Systems," by Black, Carroll and Rex (The Tax Adviser, Jan. 90, at 3), extolled the potential for expert systems in tax planning. The benefits listed in the article were: expert systems allow a firm to use the knowledge acquired by its human tax experts more readily and spread the use of this knowledge to all members of the firm. The systems allow practitioners at all levels of the firm to work at higher levels of performance. The systems also encourage the learning and training process, and a higher level of quality and consistency within the firm.

As we enter 1995, in reviewing the decision support software available on the market, there is little software that can be considered an expert system. There is no clear reason for this; however, there are several possibilities. First, the tax code is very complex and is constantly changing. The velocity of changes directly increases the cost of building and reduces the estimated useful life of an expert system. Second, despite articles extolling the value of expert systems, there appears to be a general lack of belief among tax practitioners as to their benefits. Third, CD-ROM tax libraries (such as CCH and RIA) may limit the current need for expert systems. These tax databases provide ready access to information on most topics of interest to tax advisers and are usable by even the most novice tax adviser. Fourth, the tax impact and the business concerns of the proposed transactions may be so integrated that a tax only expert system is not adequate to support the planning decision. Regardless of the reasons, expert systems have not developed at the rate anticipated five years ago.

One bright spot in the use of expert systems is the newly introduced CCH Tax Assistant products that are DSS software built using a decision tree approach to tax planning. This software is a DOS-based productivity tool designed to help tax advisers solve complex tax problems by focusing on relevant tax issues, asking questions and giving answers, and generating forms and documents that can be filed with the IRS, including election statements, research memos and client letters. In addition, this software performs calculations not performed by tax preparation software packages and allows the user to perform "what-if" planning scenarios. While Tax Assistant is not a full expert system, it is a step in the right direction.

Neural networks

The neural network is another computerized aid to decision-making that has found significant application in several areas, most notably medicine. A neural network is composed of elements that perform in a manner analogous to the most elementary functions of the biological neurons in animal brains. It is designed to learn to solve a problem based on a pattern of previous cases. A neural network is trained by providing it with case histories of past experience. Artificial neural networks built with computers attempt to mimic the biological neural networks of humans by tying patterns of artificial neuron firings to the action or advice desired from the neural networks. The disadvantage of this type of network as a tax DSS is that many situations do not have a previously developed pattern. This pattern must be developed before the neural network can function. In addition, the tax system is subject to frequent major changes that would require that the neural network relearn the system each time. This would necessitate constant relearning by the tax neural network system. It would also require a continuous stream of case histories conducted outside of the network, in order for the network to learn the system. This continuous retraining is expensive, thereby making neural networks an impractical system for tax decision support systems.

Conclusion

As we enter the last half of the 1990s, the dominant form of tax decision support software is limited to comparing alternatives in a spreadsheet format. While there is significant specialty decision support tax software available, at this time, the primary tax DSS is tax planning software. Because of the velocity of changes in the tax law, neither expert systems nor neural network techniques look promising for development as decision support software. The possible exception that provides hope for the development of expert systems is CCH's Tax Assistant. At least for now, multicolumn comparison of alternatives is the primary tax decision support software for the 1990s. Note: The software listed is limited to those that could be discovered in a search of advertisements in various tax and computer journals. These lists most probably are incomplete; the Tax Computer Applications Committee appeals to readers to bring to its attention any software overlooked. The committee anticipates providing a more detailed review of the existing tax decision support software in a future article and would like to include any software not discussed in this one. Please direct correspondence to: Dr. Sam A. Hicks Virginia Tech 3007 Pamplin Hall Blacksburg, VA 24061-0101

Committee Addresses Concerns on Electronic Filing

At its December meeting, the Tax Division's Tax Computer Applications Committee discussed its concerns about the future of electronic filing (ELF) of Federal income tax returns. W. Lloyd Strickland, a member of the Tax Division's Tax Practice and Procedures Committee, outlined five major concerns with electronic filing.

1. Workload compression.

2. Lack of forms; 5% of the forms are not accepted for ELF.

3. Abuses of the ELF system.

4. Increased cost to tax advisers and taxpayers.

5. Increasing education of tax advisers on the ELF program.

At the plenary session, Peggy C. Rule, the IRS's Executive for Electronic Filing Strategy, gave her view of the future of ELF. She noted that by the year 2001, the IRS expects to have 62% of all returns filed through the ELF program. However, a major concern is the control of fraud as the program expands. To that end, the IRS has increased the number of security checks, including verification of all taxpayer identification numbers before processing an ELF return. Ms. Rule indicated that the Service will step up its visitation of offices engaged in ELF for compliance with Rev. Proc. 94-63.

On a positive note, Ms. Rule indicated that ELF return data would be more accessible to taxpayers and their advisers after filing. Currently, only 30% to 40% of paper return data is actually entered in the IRS's computer system. ELF returns will have all data fields on-line for future retrieval. The committee will continue to monitor the technological issues as electronic filing expands.

RELATED ARTICLE: Tax Preparation Software with Planning Capabilities

[TABULAR DATA OMITTED]

RELATED ARTICLE: Schedule 2: Specialty Tax Software

A Plus Tax:                                           Cost
Preparation support of charitable trust returns      $1,150
Preparation support for tax-exempt organizations
  and private foundations                                595

Brentmark:
Pension and excise tax planner                           249
Charitable financial planner                             249
Financial planning tools in investment, inflation,
  real estate, insurance, net worth, financial goals,
  budgeting                                              249
IRS interest and penalties calculator                    129
Estimated tax penalties calculator                        79

Denver Tax:
C Corp vs. S Corp analyzer                                89
IRA distribution calculator                              129
Loan refinance analyzer                                   89
Individual and corporate NOL program                     149
Lease vs. buy analyzer                                   109
IRS interest and penalties calculator                     99

RELATED ARTICLE: Schedule 3: Tax Planning Software(*)

                   Alternative     Updates        States   Cost
BNA Tax Planner        10       As law changes      50     $890
A Plus Tax              5       Annually            44      700
BTB Tax Planner         9       As law changes       6       89
Hyper Tax               2       Summer update      (**)     179
Tax Crystal Ball        6       3 times a year       25     495
Tax Minimiser          15       3-4 times a year   (**)     395

(*) All of the software use an interactive method of input.
(**) State rates must be programmed in.

Editor's note: Dr. Hicks, Mr. Love and Mr. Maida are members of the AICPA Tax Division's Tax Computer Applications Committee.