question and mailed it to the location that had access to the information whose staff would then respond to the taxpayer.

Early in 1995, IRS implemented a networking capability among the 10 service centers so that employees could have access to IDRS data nationwide. This networking capability is referred to as Universal IDRS. Although Universal IDRS gives IRS employees access to taxpayer account information nationwide, IDRS does not always contain complete information on a taxpayer's account. Other information needed to help the taxpayer may reside in different systems that are not linked to IDRS. For example, an IRS employee using IDRS will know that a taxpayer was sent an underreporter notice, but would not have access to the actual notice. That notice is contained in IRS' Automated Underreporter system. The notice would provide additional information, such as the amount of unreported income and information return data that may indicate, for example, the amount of dividend or interest reported by financial institutions but not by the taxpayer.

To resolve these kinds of problems, IRS eventually intends to provide its employees with access to greater amounts of on-line taxpayer data in shorter time frames than those for the current IDRS data. This capability is to be delivered when IRS implements two TSM projects-the Corporate Accounts Processing System (CAPS) and the Workload Management System (WMS). CAPS is to be the main repository of taxpayer account data, and WMS is to track and manage all open account issues for a taxpayer. These projects are scheduled to be implemented in 1999.

In the interim, IRS plans to use Integrated Case Processing (ICP) to gain access to and integrate information from each of the existing IRS functional databases that contain taxpayer information. Using a taxpayer's social security number to obtain case information, the ICP software is expected to automatically assemble the relevant information on screen, provide questions and prompts for the customer service representative, and perform calculations for updating the account. With ICP, it is envisioned that IRS customer service staff will have all relevant information from a number of important databases available to them when they talk to the taxpayer. This is key to meeting IRS' customer service goals.

IRS plans to deliver ICP in four software increments. The first software increment consists of eliminating the need for IRS employees to use multiple workstations to access data on individual taxpayers from different information systems. As of February 1996, the first increment of ICP was partially deployed at 13 of the 23 customer service centers. 10 The next ICP software increment is being designed to provide enhancements over the first increment. Some of the enhancements include consolidating the information from multiple systems onto a single standard screen and providing IRS with the capability to route calls to the most skilled IRS employee who is available at the time of a taxpayer's call. Later versions are expected to provide this same level of access to information for business taxpayers.

Managerial, Technical, and Human Resource Challenges Remain

IRS' strategy for improving customer service offers promise as it is designed to improve taxpayers' ability to get assistance from IRS and provide IRS employees access to the information they need to help taxpayers. However, IRS faces important managerial, technical, and human resource challenges to fully achieve its customer service vision. Specifically, it has to manage the transition to the customer service vision while continuing to meet the current workload for providing answers to taxpayer inquiries, managing taxpayer accounts, and collecting unpaid taxes. IRS also has to determine the scope of responsibilities for those staff employed at customer service centers and provide the requisite training for that staff. IRS also has to develop the information systems necessary to support the accomplishment of its vision, including interactive telephone systems that are easy for taxpayers to use.

IRS' COMPLIANCE GOAL MAY BE JEOPARDIZED BY CHANGING ASSUMPTIONS The third major part of IRS' vision is to increase compliance. According to IRS, compliance levels have remained at 87 percent for the last several years. IRS estimates that each percentage point increase in compliance could generate billions of dollars in revenue. In addition, IRS is faced with an inventory of collectible tax debts that, according to IRS estimates, was about $46 billion as of September 30, 1995.

IRS' goal is to increase compliance to 90 percent by 2001 through improved voluntary compliance and enforcement. However, it is unclear how IRS expects to achieve that goal, especially considering some of the changes since the goal was es

10 IRS' plans call for purchasing 16,000 ICP workstations through 2000. IRS has already purchased about 2,300 of those workstations.

tablished. Since then, for example, IRS (1) has begun reassessing its data needs and revised its plans to capture 100 percent of the data on tax returns; (2) has postponed indefinitely the Taxpayer Compliance Measurement Program (TCMP), which has been IRS' primary program for obtaining comprehensive and reliable taxpayer compliance data since the 1960s; (3) no longer anticipates being able to do up-front matching of tax returns and information returns, at least until sometime after 2000; and (4) has abandoned its assumption that staff-year savings from modernization would be reinvested in front-line customer service and compliance positions. Ready Access to Good Data is Critical to Achieving IRS' Compliance Goal

Achievement of IRS' compliance goal hinges on the ability of enforcement staff to readily access good data. For example, as we discussed in recent testimony on IRS' debt collection practices, existing IRS computer systems do not provide ready access to needed information and, consequently, do not adequately support modern work processes. 11 Access to current and accurate information on tax debts is essential if IRS is to enhance the effectiveness of its collection tools and programs to optimize productivity, devise alternative collection strategies, and develop programs to prevent taxpayers from becoming delinquent in the first place.

Although technology plays a key role in helping an organization collect good data and make it readily accessible to employees, it is critical that the organization first determine what data it needs. IRS has not yet identified all of the data that enforcement staff need to do their job.

IRS currently captures about 40 percent of the data provided by taxpayers on their individual income tax returns. IRS' intent, as part of modernization, was to capture, either through electronic submission or imaging, 100 percent of the data. However, as part of the TSM reassessment effort discussed earlier, IRS has decided that it will continue capturing about 40 percent of the individual income tax return data for at least the next 5 years, with the intent of moving to 100 percent data capture sometime after that. If IRS is going to continue capturing 40 percent of the tax return data, it is critical that it capture the right 40 percent. IRS does not now know if it is capturing the right data. According to IRS officials, efforts are underway to validate IRS' business needs and to do a line-by-line review of the individual income tax return (Form 1040) to determine what is and is not needed.

It is also important that any data IRS captures, whether 40 percent or 100 percent of the universe, be easily accessed by staff who need it. In that regard, IRS officials told us that enforcement staff are not able to readily access the data that IRS is now capturing. Like Customer Service, IRS' enforcement functions should benefit from the eventual replacement of the current master files with CAPS and WMS.

Data are also critical to IRS' new approach for researching ways to improve compliance. IRS has traditionally responded to noncompliance through audits and other enforcement efforts. Over time, IRS concluded that enforcement was essential to pursue intentional noncompliance but that improved taxpayer assistance and education, rather than enforcement, might be more appropriate for correcting unintentional noncompliance. With this in mind and concerned about noncompliance levels, IRS created a compliance research and analysis approach in 1993, with the intent of identifying noncompliant market segments and appropriate enforcement and nonenforcement efforts to address that noncompliance. IRS' major research tool is to be the Compliance Research Information System (CRIS). Plans call for CRIS to be an integrated network of databases containing a sample of internal, external, and multi-year data, accessible to national and district office personnel to support analyses of voluntary compliance levels. CRIS is expected to enable IRS to develop working hypotheses on the means to increase voluntary compliance, test hypotheses, evaluate the results, and make decisions on how to implement the new strategies. IRS may not have objective compliance data available when needed for its research efforts. In October 1995, IRS indefinitely postponed TCMP due to budget and taxpayer burden concerns. TCMP has been IRS' primary program for obtaining comprehensive and reliable taxpayer compliance data since the 1960s. IRS has not done a TCMP of individual income tax returns since 1988. With the postponement of TCMP, IRS lacks current measures on compliance and does not have the data it needs to determine which market segments to research on ways to correct noncompliance. As we discussed in an April 1996 report to the Commissioner of Inter

11 Tax Administration: IRS Tax Debt Collection Practices (GAO/T-GGD-96-112, Apr. 25, 1996).

nal Revenue, although IRS plans to mitigate the data losses resulting from the postponement of TCMP, it has no specific proposal on how to accomplish this.12

IRS' original vision assumed that compliance efforts would be enhanced by more timely issue identification and resolution, facilitated in part by accelerating the matching of tax return data with data provided by third parties, such as banks and employers. The ultimate goal was to achieve up-front matching whereby data are received and processed soon enough to allow matching with the tax return while the return is being processed and before any refund is issued.

IRS has been accelerating the matching process, but the first notice to taxpayers advising them of any discrepancy is still not sent out until about a year after the tax return was filed. According to IRS officials, IRS eventually wants to be able to send out notices in the same year the return was filed. Besides increasing the likelihood of contacting the taxpayer and resolving the case, sending the notice the same year the tax return is filed might help taxpayers avoid the same mistake on the following year's return. It is not clear when IRS will be able to do up-front matching.

Changes in Staffing Assumptions Make Achievement of IRS' Compliance Goal More Difficult

Data and the technology that provides it are critical, but so are the people who are tasked with using the data. IRS already has tens of thousands of staff who work in areas, such as taxpayer service, examination, and collection, that can affect compliance levels. Until recently, IRS had assumed that staff savings resulting from modernization (such as the savings anticipated in the returns processing function) would be reinvested to provide more of those front-line staff, with a corresponding increase in revenues. That is no longer the case, at least not to the extent originally anticipated. According to IRS officials, one of the assumptions surrounding its recent reassessment of TSM was that IRS would be smaller and could not rely on reinvesting TSM savings.

We have not quantified the implications of this change in staffing assumptions. It is clear, however, that IRS' success in increasing compliance is directly related to the number of staff involved in compliance-related activities and that any significant change in staffing could significantly affect IRS' ability to achieve 90-percent compliance goal. IRS could mitigate that effect, at least somewhat, by making sure that is has the right mix of staff.

Staff mix is an issue, for example, in IRS' collection function. While the private sector emphasizes the use of telephone collection calls, IRS allocates a significant portion of its collection resources to field offices where revenue officers visit taxpayers. IRS has initiated programs and made procedural changes to speed up its collection process, but historically IRS has been reluctant to reallocate resources from the field to earlier, more productive, collection activities. IRS' fiscal year 1997 budget request states that, although traditional enforcement positions (which include revenue officers) "comprise the lion's share of IRS' enforcement efforts, they also represent on the margin the least efficient use of IRS resources." In that regard, the budget request provides for an increase in staff for IRS' telephone collection activities and a decrease in revenue officers-a shift toward the kind of mix that we have advocated in the past.

Another way to mitigate the effect of fewer-than-expected staff is to improve staff productivity. In that regard, one of IRS' efforts to improve compliance involves the automation of certain tasks done by enforcement staff in IRS' district offices. These tasks, like many in IRS, have for years involved the manual processing of paper, which has resulted in enforcement staff spending significant amounts of time on routine administrative duties. IRS has been implementing systems that are designed to ease this burden and help make enforcement staff more productive.

The Integrated Collection System (ICS) is a computer-based information system that is intended to automate some of the labor-intensive tasks performed by revenue officers. Although this effort is not a major technological advancement, it should enable revenue officers to spend their time more productively. According to IRS, implementing this system in two pilot districts resulted in increased collections, faster case closing, and less time spent on each case. The system is currently operating in six districts, and IRS plans to roll it out in three more districts this year. According to IRS, further implementation depends on future funding and final measurements of productivity.

IRS is also developing an automated inventory delivery system that is intended to direct accounts, based on internally developed criteria, to the particular collection

12 Tax Administration: Alternative Strategies to Obtain Compliance Data (GAO/GGD-96–89, Apr. 26, 1996).

stage where they can be processed most efficiently and expeditiously. This system, which IRS plans to test in July 1996, is intended to move accounts through the collection process faster and cheaper than under the current system.

Another field system, the Totally Integrated Examination System (TIES), provides automated tools to help IRS' Examination staff capture data about their audits and compute tax liability, interest, and penalties-tasks that previously had to be done manually. TIES has been implemented in seven IRS offices. However, IRS recently decided not to continue funding TIES through TSM. It is our understanding that any future funding will be done outside of TSM.

According to IRS, the features of ICS and TIES will eventually be incorporated into the Integrated Case Processing (ICP) system, which will also provide IRS' compliance function with automated tools for case assignment and tracking. However, as a result of the recent reassessment of TSM, IRS has decided to delay that integration until after the year 2000.

TSM INVESTMENTS REMAIN AT RISK

In March 1996, we told the Subcommittee on Oversight, House Committee on Ways and Means, that additional investments in TSM are at risk given current managerial and technical weaknesses.13 Those were weaknesses that we discussed in our July 1995 report on TSM.14 The Department of the Treasury is expected to report to the Senate and House Appropriations Committees on IRS' progress in dealing with those weaknesses soon.

One of the managerial weaknesses discussed in our July 1995 report that has significant programmatic implications was a lack of integration of IRS' reengineering efforts and TSM projects. Specifically, we said that IRS' reengineering efforts were not tied to its TSM projects and that IRS lacked a comprehensive plan and schedule defining how and when to integrate these business reengineering efforts with ongoing TSM projects.

The reengineering efforts we referred to in July 1995 were put on hold pending the outcome of IRS' reassessment of TSM. As a result of the recent reassessment of TSM, IRS decided to reengineer "the tax settlement process". IRS has defined that process as beginning at the point taxpayers collect information necessary for the filing of tax returns and ending when the current year tax account is satisfied or enforcement action is initiated. IRS has identified 18 high-level processes for this time period. One of those processes focuses on IRS' tax return processing activity that we mentioned earlier.

We question IRS' ability to make sound investment decisions on TSM until reengineering of important processes, such as tax return processing, are sufficiently completed. Reengineering could result in new business requirements that are not addressed by planned TSM projects or that make those projects obsolete. For example, if IRS decides that it is cost-effective to outsource paper tax return processing, it will not need the scanning and imaging technologies that DPS is being designed to provide.

In closing, Mr. Chairman, our main point is that until clearly defined business requirements drive TSM projects, there is no assurance that TSM projects will achieve the desired objectives and result in improved operations. IRS must clearly define its business needs and determine the most cost-effective means for meeting those needs to ensure that it makes effective use of funds provided for information technology projects.

That concludes my statement. We welcome any questions that you may have. Chairman STEVENS. Do you have any comments to make at this point, Dr. Stillman.

Ms. STILLMAN. No, I do not, Senator.

Chairman STEVENS. I keep asking myself why we are here. It seems that there has been a lack of willingness of the IRS to really accept outside advice; is that a fair statement?

Ms. STILLMAN. You are correct there is a history of IRS getting but ignoring very good advice. They show that they have very good taste in advisors-they have gone to the National Research Coun

13 Tax Administration: IRS' Fiscal Year 1996 and 1997 Budget Issues and the 1996 Filing Season (GAO/T-GGD-96-99, Mar. 28, 1996).

14 Tax Systems Modernization: Management and Technical Weaknesses Must Be Corrected If Modernization Is To Succeed (GAO/AIMD-95-156, July 26, 1995).

cil, we have given them advice, they have some very good contractors-but they do have a history of not taking that advice.

Chairman STEVENS. And at one point you say you question IRS' ability to make sound investment decisions in TSM until reengineering of important processes such as tax return processing are sufficiently completed. Reengineering could result in new business requirements that are not addressed by planned TSM projects or that make those projects obsolete. I was taken by that. It is almost like we are dealing with two railroad trains going the same direction, but they cannot get on the same track. What do we need to do to get them on the same track, Ms. Willis?

Ms. WILLIS. Well, Chairman Stevens, I think what we need to do is to start thinking about reengineering and systems modernization in the same sentence rather than viewing them as two separate activities. I think if there is one point I would like to emphasize from our statement, and it is a point we have made in the past, technology is a tool that helps you achieve your business vision. It should not drive your business vision, and as you look at how your processes are going to operate in the future, the opportunities provided by technology need to be considered, but it is just one set of options. And what you need to do is make sure that the options you select are the most cost effective, and in all cases that may not be a single type of technology or a particular technology.

And at this point IRS is in the position of needing to look at its reengineering efforts, deciding where those reengineering efforts fit into its business vision, how it will support the achievement of that vision in very practical, direct terms, and then look at where technology provides opportunities for efficiently accomplishing its mission. Technology should not be moving on down the track on its

own.

Chairman STEVENS. Well, if I understand your statement right, electronic filing has not been as good in 1995 or 1996 as it was in 1994, compliance is stagnant, and really the modernization program itself is on hold. Now, have you made suggestions as to how to deal with-let us just take the question of the filings-I am learning a new word-submissions now. Were the estimates for the number of people that would use the electronic method excessive? Were they unachievable? What happened in that field? Why is it falling behind 1994?

Ms. WILLIS. Well, part of the reason why it is behind 1994 was the impact of certain filters and controls that were put in place in 1995 to reduce the amount of filing fraud within the system. That is not bad, and I do not mean to imply that the fact that electronic filings went down in 1995 in any sense is something that we should say is a result of things that were necessarily bad. I think more important is the longer term view that if IRS continues on its current trend and growth of the number of electronic filings, it will fall far short of its goals and it will still continue to get many more paper returns than it originally anticipated.

How we are going to deal with those paper returns not only in terms of processing them but also in terms of acessing the data from them for downstream customer service and compliance uses is very, very important. IRS has never really had a strategy that clearly laid out how they were going to get to their goal for elec