The 1995 Tax Filing Season: IRS Performance Indicators Provide Incomplete Information About Some Problems (Letter Report, 12/29/95, GAO/GGD-96-48). Pursuant to a congressional request, GAO reviewed the Internal Revenue Service's (IRS) performance during the 1995 tax filing season, focusing on the: (1) processing of individual income tax returns and refunds; (2) ability of taxpayers to reach IRS by telephone; and (3) new IRS computer system for processing returns. GAO found that: (1) of the 74 million refunds IRS issued as of June 16, 1995, 7 million were delayed for up to 8 weeks because of systematic checks for questionable refund claims; (2) IRS delayed refunds on returns with missing or invalid social security numbers (SSN) to identify duplicate uses of the same SSN and fraud schemes; (3) although IRS answered 11 percent more calls from taxpayers in the 1995 filing season, the chance of reaching an IRS assistor was not very good; and (4) IRS experienced numerous problems with its electronic image computer system, including extensive downtime and slow processing rates, that hampered productivity. --------------------------- Indexing Terms ----------------------------- REPORTNUM: GGD-96-48 TITLE: The 1995 Tax Filing Season: IRS Performance Indicators Provide Incomplete Information About Some Problems DATE: 12/29/95 SUBJECT: Personal income taxes Tax refunds Taxpayers Fraud Social security number Telephone communications operations Computerized information systems Questionable payments Tax returns Information processing operations IDENTIFIER: IRS 1040PC Program IRS TeleFile Program IRS Service Center Recognition/Image Processing System Earned Income Tax Credit IRS Electronic Filing System ************************************************************************** * This file contains an ASCII representation of the text of a GAO * * report. Delineations within the text indicating chapter titles, * * headings, and bullets are preserved. Major divisions and subdivisions * * of the text, such as Chapters, Sections, and Appendixes, are * * identified by double and single lines. The numbers on the right end * * of these lines indicate the position of each of the subsections in the * * document outline. These numbers do NOT correspond with the page * * numbers of the printed product. * * * * No attempt has been made to display graphic images, although figure * * captions are reproduced. Tables are included, but may not resemble * * those in the printed version. * * * * A printed copy of this report may be obtained from the GAO Document * * Distribution Facility by calling (202) 512-6000, by faxing your * * request to (301) 258-4066, or by writing to P.O. Box 6015, * * Gaithersburg, MD 20884-6015. We are unable to accept electronic orders * * for printed documents at this time. * ************************************************************************** Cover ================================================================ COVER Report to the Chairman, Subcommittee on Oversight, Committee on Ways and Means, House of Representatives December 1995 THE 1995 TAX FILING SEASON - IRS PERFORMANCE INDICATORS PROVIDE INCOMPLETE INFORMATION ABOUT SOME PROBLEMS GAO/GGD-96-48 The 1995 Filing Season (268674) Abbreviations =============================================================== ABBREV DDI - Direct Deposit Indicator DPS - Document Processing System EIC - Earned Income Credit FMS - Financial Management Service FTD - federal tax deposit IRP - Information Returns Program IRS - Internal Revenue Service RPS - Remittance Processing System SCRIPS - Service Center Recognition/Image Processing System SSN - Social Security number Letter =============================================================== LETTER B-259952 December 29, 1995 The Honorable Nancy L. Johnson Chairman, Subcommittee on Oversight Committee on Ways and Means House of Representatives Dear Chairman Johnson: This report responds to your request that we assess the Internal Revenue Service's (IRS) performance during the 1995 tax filing season. Specifically, we discuss (1) the processing of individual income tax returns and refunds, including information on IRS' efforts to combat refund fraud; (2) the ability of taxpayers to reach IRS by telephone; and (3) the performance of a new computer system for processing returns. We provided information on the interim results of our work in a February 1995 testimony\1 before the Subcommittee. -------------------- \1 Tax Administration: IRS' Fiscal Year 1996 Budget Request and the 1995 Filing Season (GAO/T-GGD-95-77, Feb. 27, 1995). RESULTS IN BRIEF ------------------------------------------------------------ Letter :1 IRS indicators show that its 1995 filing season goals were generally met. For example, IRS received more individual income tax returns in 1995 than in 1994; answered 11 percent more telephone calls than expected; issued refunds on average faster than its goal of 40 days; and accurately filled 97 percent of taxpayers' orders for forms and publications. However, those indicators do not provide a complete assessment of the filing season. There were several serious problems not obvious from the indicators: (1) IRS' efforts to combat fraud generated much adverse publicity that might have been alleviated if IRS had better forewarned taxpayers of potential refund delays; (2) our tests and IRS data showed that taxpayers continued to have serious problems trying to reach IRS by telephone; and (3) a new document imaging system did not perform as expected, leading to increased returns processing costs and lower-than-expected productivity. Of the 74 million refunds IRS issued as of June 16, 1995, over 7 million (10 percent) were delayed for up to 8 weeks because of systematic checks for questionable refund claims that IRS initiated in 1995. IRS delayed refunds on returns that were received with missing or invalid Social Security numbers (SSN). IRS also delayed refunds on many returns that had no SSN problems, with an emphasis on returns claiming the Earned Income Credit (EIC), to allow staff time to identify duplicate uses of the same SSN and fraud schemes. Although IRS found many problem returns, it also delayed the refunds of millions of taxpayers whose SSNs were found to be valid. According to one of IRS' indicators, refunds from returns filed on paper were received by taxpayers within an average of 36 days from the date the return was signed--the same average number of days IRS reported achieving in 1994. However, that result is misleading because IRS excluded from its computation in 1995 all of the delayed refunds, even if the refund was eventually determined to be valid. In addition, while we support IRS' decision to delay refunds to ensure their validity, we believe that some of the adverse publicity that ensued might have been avoided if IRS had done a better job of forewarning taxpayers. Although IRS answered 11 percent more calls from taxpayers with tax law, account, or procedural questions than it estimated it could answer during the 1995 filing season, that indicator masks the fact that IRS received many more calls than it was able to answer and that a taxpayer's chances of reaching an IRS assistor during the 1995 filing season were not very good. We tested the telephone system's accessibility by making 2,821 calls during two 2-week periods in the 1995 filing season. We succeeded in getting through to an IRS assistor only 249 times (9 percent). While most of IRS' computer systems performed without significant problems during the 1995 filing season, a new system designed to scan and produce an electronic image of certain tax returns and other documents did not perform to the level that IRS had expected. Throughout the 1995 filing season, IRS' processing centers reported numerous problems with the system. Extensive downtime and slower-than-expected processing rates limited system effectiveness and hampered productivity. OBJECTIVE, SCOPE, AND METHODOLOGY ------------------------------------------------------------ Letter :2 Our objective was to assess IRS' performance during the 1995 filing season. To achieve our objective, we interviewed IRS National Office officials and IRS officials in the Atlanta, Austin, Cincinnati, Fresno, Kansas City, Memphis, and Ogden service centers responsible for the various activities we assessed; tested the accessibility of IRS' toll-free telephone assistance and forms-ordering telephone lines by placing calls from Atlanta, Chicago, Cincinnati, Kansas City, New York, San Francisco, and Washington, D.C. (appendix III contains more information on our test methodology); analyzed filing season-related data from various IRS sources, including data on telephone accessibility, return filings, return processing errors, refund fraud, and the results of steps IRS took in 1995 to address the fraud problem; reviewed IRS publications, notices, and forms to determine what taxpayers were told about potential refund delays; reviewed reports on computer system performance and attended weekly meetings on computer system performance held by IRS' National Office Command Center; and reviewed relevant IRS internal audit reports. We did our work from January through September 1995 in accordance with generally accepted government auditing standards. We requested comments on a draft of this report from the Commissioner of Internal Revenue or her designee. On November 13, 1995, several IRS officials, including the Assistant Commissioner for Taxpayer Services, Director of Investigations (Tax Refund Fraud), Electronic Filing Executive, Director of Tax Forms and Publications, and Senior Program Analyst (Submission Processing), provided us with oral comments. Another IRS official, the Chief of Taxpayer Services, provided additional comments on November 30. IRS' comments are summarized and evaluated on page 18 and incorporated in this report where appropriate. FILING SEASON GOALS WERE GENERALLY MET, BUT SOME SERIOUS PROBLEMS OCCURRED ------------------------------------------------------------ Letter :3 IRS uses various indicators to measure its filing season performance. Because IRS' most important job during the filing season is to process tax returns, two important workload indicators are (1) the number of individual income tax returns received in total and (2) the number of returns received through alternative filing methods that IRS developed to help make the returns processing function more efficient. According to IRS data, more individual income tax returns were received in 1995 than in 1994, but the number received through the alternative filing methods decreased. Among other IRS filing season indicators are those related to workload, such as the percent of scheduled tax assistance calls answered; timeliness, such as the number of days needed to process returns or issue refunds; and quality, such as the accuracy of IRS' answers to taxpayer questions and IRS' processing of returns and refunds. Those indicators show that IRS met or exceeded most of its performance goals for the 1995 filing season. According to IRS data, for example, (1) IRS' telephone assistors answered 11 percent more calls than IRS anticipated and provided accurate answers to 91 percent of taxpayers' tax law questions; (2) 97 percent of taxpayers' orders for tax forms and publications were filled accurately; (3) on average, refunds on paper returns were processed and issued within 36 days; and (4) service centers met deadlines for processing tax payments submitted with returns. What IRS' indicators do not reveal are the difficulties IRS experienced in the 1995 filing season. There were several serious problems not obvious from the indicators: (1) IRS' efforts to combat refund fraud took millions of taxpayers by surprise when closer scrutiny of their returns resulted in refunds being delayed; (2) most taxpayers who called IRS to ask questions could not get through; and (3) IRS' implementation of a new tax return processing system fell far short of expectations. NUMBER OF RETURNS FILED INCREASED, BUT USE OF ALTERNATIVE FILING METHODS DECLINED ---------------------------------------------------------- Letter :3.1 With one exception, as shown in figure 1, the number of individual income tax returns filed has increased every year since fiscal year 1987. Figure 1: Number of Individual Income Tax Returns Filed in Fiscal Years 1987 Through 1995 (See figure in printed edition.) Source: Data for fiscal years 1987 through 1994 are from IRS annual reports. Data for 1995 are from IRS' Management Information System for Top Level Executives. While there was an increase in the overall number of returns filed in 1995, the number received through alternative filing methods declined. IRS offers three alternatives--electronic filing, TeleFile, and 1040PC--to the traditional filing of paper returns.\2 As shown in table 1, the use of electronic filing and 1040PC decreased in 1995 compared with 1994, while the use of TeleFile increased. Table 1 Number of Individual Income Tax Returns Received Through Alternative Filing Methods in 1994 and 1995 Actual Actual Percent filed filed change Type of Return in 1994\a in 1995\a from 1994 ------------------------------ ------------ ---------- ------------ Electronic 13,510,000 11,142,000 -17.5 TeleFile 519,000 680,000 31.0 1040PC 4,183,000 2,902,000 -30.6 Total 18,212,000 14,724,000 -19.2 ---------------------------------------------------------------------- \a Data are current as of September 30, 1994, and September 29, 1995. Numbers are rounded to the nearest thousand. Source: IRS' Management Information System for Top Level Executives. IRS attributes the drop in electronic filing to the various steps it took in 1995 to deal with refund fraud. One of those steps was to eliminate the direct deposit indicator (DDI).\3 Because the elimination of the DDI increased the risk of the loan, lenders cut their maximum loan amount and raised their fees. Some potential electronic filers may have decided to file on paper when they found they were unable to get a refund anticipation loan or were unwilling to pay the additional fee. Other steps IRS took to deal with fraud, some of which may have also contributed to the decline in electronic filing, are discussed later. The decline in 1040PCs resulted from a private tax preparation firm, which was the largest user of 1040PCs, dropping out of the program. For the 1995 filing season, IRS required that preparers provide taxpayers with some type of descriptive printout or legend that explained each line on the taxpayer's 1040PC return. The purpose of the legend was to provide better supporting documentation than was previously available to the taxpayers and was to be used as an aid in doing things such as preparing state returns and completing financial aid forms. According to an official of the tax preparation firm that dropped out of the program, the firm chose to stop participating rather than incur the extra cost associated with providing the legend. The growth in the third alternative, TeleFile, was due in part to its availability to more taxpayers. In 1995, TeleFile was available to certain taxpayers in 10 states--3 more states than in 1994.\4 Some of the growth might also be due to improved accessibility. As we discussed in our report on the 1994 filing season,\5 IRS experienced an overload of the TeleFile system in 1994, and taxpayer accessibility might have been higher had the system been able to handle the number of calls. For the 1995 filing season, IRS took several steps that increased accessibility. For example, IRS increased the number of telephone lines from 144 to 336 and stopped testing the use of voice signatures, which shortened the length of calls. In August 1995, IRS' Internal Audit reported that the number of busy signals received by taxpayers trying to use TeleFile in 1995 decreased dramatically from 1994 and that 87 percent of the taxpayers using TeleFile in 1995 were able to access the system on the first attempt. As a possible result of that improved accessibility, the number of TeleFile filers in each of the seven states that were involved in the program in 1994 increased in 1995. -------------------- \2 Under electronic filing, returns are transmitted over communications lines to an IRS service center, where they are automatically edited and processed. Under TeleFile, certain taxpayers who are eligible to file a Form 1040EZ are allowed to file using a toll-free number on touch-tone telephones. Under the 1040PC method, a filer uses personal computer software that produces tax returns in an answer-sheet format. The 1040PC shows the tax return line number and the data (dollar amount, name, etc.) on that line. Only lines on which the taxpayer has made an entry are included on the 1040PC. \3 The DDI signaled that IRS would not reduce the taxpayer's refund to pay another federal debt of the taxpayer. Financial institutions used the DDI as a basis for making refund anticipation loans. \4 IRS plans to make TeleFile available nationwide in 1996. \5 Tax Administration: Continuing Problems Affect Otherwise Successful 1994 Filing Season (GAO/GGD-95-5, Oct. 7, 1994). OTHER INDICATORS SHOW THAT IRS GENERALLY ACHIEVED ITS PERFORMANCE GOALS ---------------------------------------------------------- Letter :3.2 As shown in table 2, IRS met or exceeded almost all of its other performance goals for 1995. We did not assess the overall appropriateness of those goals. However, as discussed in the next section, the indicators for refund timeliness, number of tax assistance calls answered, and returns processing cycle time masked serious problems that occurred in 1995. Table 2 IRS' Performance Goals for 1994 and 1995 and Related Accomplishments Indicator Goal Accomplishment Goal Accomplishment -------------- -------------- -------------- -------------- ---------------- Accuracy of Process 94.4 95.28 percent Process 93.0 93.42 percent returns percent were processed percent were processed processed by accurately accurately accurately accurately Code and Edit\a staff\b Accuracy of Process 94.1 95.84 percent Process 94.0 93.93 percent returns percent were processed percent were processed processed by accurately accurately accurately accurately data transcribers\b Service center Process 9,475 9,557 returns Process 9,800 10,837 returns returns returns per were processed returns per were processed processing staff year per staff year staff year per staff year productivity\c Returns 11 days Various types 11 days Various types of processing of 1040s 1040s ranged cycle time\d ranged between between 5 and 9 5 and 7 days days Accuracy of Process 98.0 98.6 percent Process 97.0 99.5 percent refunds on percent were processed percent were processed paper returns accurately accurately accurately accurately Timeliness of Issue within Issued within Issue within Issued within an refunds on an average of an average of an average of average of 36 paper 40 days 36 days 40 days days returns\e Timeliness of Remittances Remittances Remittances Remittances processing tax received 4/15 received 4/ received 4/17 received 4/17/ payments through 5/1/ 15/94 through through 5/2/ 95 through 5/2/ submitted with 94 were to be 5/1/94 were 95 were to be 95 were returns\f deposited no deposited by deposited no deposited by 5/ later than 5/ 5/3/94 later than 5/ 3/95 2/94 3/95 Tax assistance Answer 98 Answered 35.5 Answer 100 Answered 39.2 calls percent of the million calls percent of the million calls scheduled to 34.9 million (102 percent 35.4 million (111 percent of be answered\g calls of schedule) calls schedule) scheduled to scheduled to be answered be answered Accuracy of Answer 89 90 percent Answer 90 91 percent were tax law percent were answered percent answered assistance accurately accurately accurately accurately Forms ordering Answer 98 Answered 7.5 Answer 100 Answered 7.2 calls percent of the million calls percent of the million calls scheduled to 8.7 million (86.8 percent 7.3 million (98.3 percent of be answered\g calls of schedule) calls schedule) scheduled to scheduled to be answered be answered Accuracy of Process 96.5 96.2 percent Process 96.5 97.2 percent processing percent were processed percent were processed form orders accurately accurately accurately accurately -------------------------------------------------------------------------------- \a Code and Edit staff prepare returns for computer entry by, among other things, ensuring that all data are present and legible. \b According to IRS, these two indicators for 1995 are not comparable to the indicators in 1994 because they included forms 1040EZ (as well as forms 1040 and 1040A in 1994 but not in 1995. Forms 1040EZ were excluded in 1995 because they were to be processed via a new computer system that would replace the work done by Code and Edit staff and data transcribers. \c Returns processing productivity is based on the number of weighted returns processed, which includes all returns (whether processed manually, through scanning equipment, or electronically). The different types of returns are weighted to account for their differing processing impacts. For example, a paper Form 1040 has a higher weighting factor than a paper Form 1040EZ, which in turn has a higher weighting factor than electronically processed returns. \d Cycle time is the average number of days it takes service centers to process returns. \e This indicator is based on a sample of paper returns and is calculated starting from the signature date on the return. As discussed on page 13, refunds that were delayed due to IRS' closer scrutiny of SSNs and EIC claims were not included in this calculation. \f For 1994, 9 out of 10 service centers met the May 2, 1994, completion date. For 1995, the cutoff date of April 17 was used because April 15 fell on a Saturday. For information on a test of an alternative method of processing tax payments, known as lockboxes, see appendix I. \g The "calls scheduled to be answered" indicator is the number of telephone calls IRS believes its call sites will be able to answer with available resources. It does not reflect the number of calls IRS expects to receive. This indicator is further discussed on pages 13 and 14. Source: IRS data. BETTER ADVANCE WARNING MIGHT HAVE EASED ADVERSE REACTION OVER REFUND DELAYS ---------------------------------------------------------- Letter :3.3 As shown in tables 3 and 4, the number of returns identified by IRS as containing fraudulent refund claims and the amount of identified fraudulent refunds that were issued before IRS could stop them have increased significantly since 1990. Table 3 Number of Detected Fraudulent Refunds in Calendar Years 1990 Through 1995 Detected Total Detected fraudulent fraudulent fraudulent electronic returns Year paper returns returns detected\b ------------------------- ----------------- ----------------- --------------- 1990 5,302 411 5,713 1991 5,422 5,746 11,168 1992 12,244 12,725 24,969 1993 51,883 25,957 77,840 1994 44,137 33,644 77,781 1995\a 31,830 27,411 59,241 ================================================================================ Total\b 150,818 105,894 256,712 -------------------------------------------------------------------------------- \a Data for 1995 are through September 30, 1995. \b Totals may not add due to rounding. Source: IRS data. Table 4 Amount of Detected, Deleted, and Issued Fraudulent Refunds in Calendar Years 1990 Through 1995\b (Dollars in millions) Amount Amount Amount Amount Amount Amount of of Amount of of Amount of of Amount fraudul fraudul of fraudul fraudul of fraudul fraudul of ent ent fraudul ent ent fraudul ent ent fraudul refunds refunds ent refunds refunds ent refunds refunds ent Ye detecte deleted refunds detecte deleted refunds detecte deleted refunds ar d \a issued d \a issued d \a issued -- ------- ------- ------- ------- ------- ------- ------- ------- ------- 19 $15.9 $14.8 $1.1 $1.2 $0.5 $0.7 $17.1 $15.3 $1.8 90 19 32.3 30.7 1.6 10.7 2.6 8.1 42.9 33.3 9.6 91 19 33.2 30.9 2.3 33.6 22.5 11.1 66.8 53.4 13.4 92 19 82.8 73.0 9.8 54.0 29.1 24.9 136.8 102.1 34.7 93 19 90.7\c 81.5 9.2 69.8 35.9 33.9 160.5 117.4 43.1 94 19 66.5\d 56.5 10.0 58.3 21.5 36.8 124.8 78.0 46.8 9 5 \ b To $321.4 $287.4 $34.0 $227.6 $112.1 $115.5 $548.9 $399.5 $149.4 t a l \ e ----------------------------------------------------------------------------------- \a A deleted fraudulent refund is one that IRS has stopped before the refund is paid out. \b Data for 1995 are through September 30, 1995. \c This figure excludes two returns claiming refunds totaling $347 million. \d This figure excludes seven returns claiming refunds totaling $315 million. \e Totals may not add due to rounding. Source: IRS data. Because of concerns raised in several of our past products\6 and in congressional hearings about those increases, IRS placed more emphasis on reducing fraud in 1995. In addition to eliminating the DDI, discussed earlier, those steps included closer scrutiny of SSNs and of refunds involving the EIC--problem areas that IRS had identified in the past. IRS' efforts generated much adverse publicity when over 7 million taxpayers had their refunds delayed for many weeks. Although IRS' decision seems prudent because of the level of possible fraud involved, it seems that IRS could have prevented some of the adverse reaction to those delays if it had done a better job of forewarning taxpayers. On a related matter, the methodology IRS used to measure refund timeliness in 1995 was flawed, in our opinion, because it excluded those refunds that were delayed. Inclusion of those refunds most likely would have increased the average refund time beyond the 36 days reported by IRS. -------------------- \6 Tax Administration: IRS Can Improve Controls Over Electronic Filing Fraud (GAO/GGD-93-27, Dec. 30, 1992); Tax Administration: Increased Fraud and Poor Taxpayer Access to IRS Cloud 1993 Filing Season (GAO/GGD-94-65, Dec. 22, 1993); Tax Administration: Electronic Filing Fraud (GAO/T-GGD-94-89, Feb. 10, 1994); IRS Automation: Controlling Electronic Filing Fraud and Improper Access to Taxpayer Data (GAO/T-AIMD/GGD-94-183, July 19, 1994); and Earned Income Credit: Targeting to the Working Poor (GAO/GGD-95-122BR, Mar. 31, 1995). IRS DELAYED MILLIONS OF REFUNDS IN 1995 ---------------------------------------------------------- Letter :3.4 In 1995, to better ensure the appropriateness of refund claims, IRS increased its efforts to verify the accuracy of SSNs on tax returns.\7 When IRS received a paper return with a missing SSN or an invalid SSN (i.e., one that does not match the Social Security Administration's records), it delayed the refund and, depending on the circumstances, contacted the taxpayer in an attempt to resolve the problem. IRS delayed refunds for up to 8 weeks on other returns (both paper and electronic), even if the returns had no missing or invalid SSNs, to allow staff time to identify duplicate uses of the same SSN and fraud schemes. Because most of the refund fraud cases IRS identified in the past involved the EIC (about 90 percent of the cases identified in 1994, for example), IRS concentrated these efforts on returns claiming the EIC. Because the delay only applied to that part of the refund attributable to the EIC, some taxpayers received two checks--one for the non-EIC part of their refund and a second, several weeks later, for the rest, assuming IRS determined that the EIC claim was valid. IRS added filters to the electronic filing system to prevent returns with missing or invalid SSNs or with SSNs that already had been used by another taxpayer from being filed electronically. As of the end of May 1995, IRS had (1) notified about 3 million taxpayers whose returns had missing or invalid SSNs that their refunds were being delayed, (2) delayed another 4 million refunds to allow time to check for duplicate SSN use and fraudulent returns, and (3) sent out about 4 million reject notices from the electronic filing system because it had identified a missing, invalid, or duplicate SSN.\8 -------------------- \7 For information on problems IRS had in implementing its efforts to detect and deter refund fraud, see appendix II. \8 Because an electronic return can be rejected for more than one reason, the number of reject notices may be greater than the number of returns rejected. TAXPAYERS NOT ALERTED TO POSSIBLE REFUND DELAYS IN CERTAIN CIRCUMSTANCES -------------------------------------------------------- Letter :3.4.1 IRS warned taxpayers that their refunds could be delayed if they submitted a return with a missing or incorrect SSN. On the cover of the instructions accompanying Form 1040, for example, IRS warned taxpayers to check their SSNs and explained that "incorrect or missing SSNs for you, your spouse, or dependents may delay your refund." It then referred the reader elsewhere in the instructions for details on how to get an SSN. IRS also issued several public service announcements to alert taxpayers to the need for correct SSNs. However, IRS did not do very much to warn taxpayers that their refunds might also be delayed even if their SSNs were correct. The only warning in the Form 1040 tax package or Publication 17 (Your Federal Income Tax)--the two IRS documents that most taxpayers would rely on for such information--was a statement in both documents that alerted potential electronic filers that "some refunds may be temporarily delayed as a result of compliance reviews" to ensure that the returns are accurate. Taxpayers who did not intend to file electronically--about 90 percent of the filers--were not told anything. Also, by advising only potential electronic filers of possible "compliance reviews," IRS might have given the impression that electronically filed returns are more subject to audit than paper returns--not the kind of message that would help expand the use of electronic filing. Conversely, IRS prominently displayed, in both the Form 1040 tax package and Publication 17, its customer-service standards for 1995. One of those standards says, "If you file a complete and accurate tax return and you are due a refund, your refund will be issued within 40 days if you file a paper return or within 21 days if you file electronically." Thus, not only were most taxpayers not told that their refunds might be delayed even if they filed a valid return, but they were led to believe the opposite by IRS' customer-service standard. The refund delays generated much adverse reaction. Numerous news articles during the filing season cited criticism from taxpayers, executives of tax preparation services, an industry lobbying organization, and members of Congress commenting on the problems they observed during the 1995 filing season. In July 1995, IRS' Internal Audit reported that it had advised management in December 1994 of its concerns about IRS' decision not to publicize the potential delay of EIC refunds. Internal Audit said that IRS "could have jeopardized the public's trust and confidence" and that "those who had already filed may have felt confused, misled, disillusioned, and perhaps angry." Internal Audit also said that advance publicity about delaying refunds might have also deterred some unscrupulous filers. We can understand IRS not wanting to disclose the details of its plans, but we fail to see how any harm would have been caused by simply alerting taxpayers to the possibility that their refunds might be delayed even if there were no problems with their SSNs. IRS' MEASURE OF REFUND TIMELINESS DID NOT REFLECT THE IMPACT OF IRS' FRAUD CHECKS -------------------------------------------------------- Letter :3.4.2 IRS' customer-service standard for issuing a refunds from returns filed on paper is 40 days. To track its success in meeting that standard, one of IRS' filing season indicators is "refund timeliness." To measure refund timeliness, IRS takes several samples of paper returns involving refunds and computes the elapsed time from the date the taxpayer signed the return to the date the taxpayer would have received the refund, allowing 2 days after issuance for the refund to reach the taxpayer. IRS' results for the 1995 filing season indicated that refunds on paper returns were issued in an average of 36 days--the same as in 1994 and 4 days quicker than IRS' goal. That result is misleading, however, because IRS excluded from the computation the over 7 million refunds that were delayed because of IRS' fraud checks. Because IRS' customer-service standard is predicated on the filing of a complete and accurate return, we agree that IRS should have excluded from its computation those refunds that were delayed because of missing or invalid SSNs (about 3 million of the 7 million delayed refunds). However, IRS did not identify any problems with the SSNs associated with about 4 million delayed refunds, and those refunds were eventually issued. Thus, consistent with IRS' standard, those refunds should have been included in the computation of refund timeliness. Using IRS data on the number of refunds in its refund timeliness samples and the number of refunds excluded from the samples--assuming that each of the excluded refunds was delayed 8 weeks, thus taking 56 more days to issue than the 36-day average--we determined that inclusion of the excluded refunds would have increased the average to 38 days. Such a result, in our opinion, would have shown more correctly a drop in performance from the 36-day average achieved in 1994. ACCESSIBILITY TO IRS' TOLL-FREE TELEPHONE LINES CONTINUES TO BE A SERIOUS PROBLEM ---------------------------------------------------------- Letter :3.5 An important indicator of filing season performance is how easily taxpayers who have questions or who want to order forms and publications are able to contact an IRS assistor on the telephone. IRS assesses its performance in that area by estimating the number of calls it expects to answer during the filing season (known as "scheduled calls") and comparing that number with the number of calls it actually answered. For the 1995 filing season, IRS answered 111 percent of the scheduled calls to its toll-free tax assistance telephone line and 98 percent of the scheduled calls to its toll-free forms ordering line. Because IRS' indicator is based on the number of calls IRS expects to answer rather than the number it expects to receive, the indicator masks the serious problems taxpayers have encountered in the past and encountered again in 1995 in trying to reach IRS by telephone. In reports on past filing seasons, we discussed the difficulty taxpayers had in reaching IRS by telephone (i.e., the "accessibility" of IRS' telephone systems).\9 Although IRS answers millions of calls each year, even more calls go unanswered. Many taxpayers receive busy signals, are kept on hold for a long time, or simply give up. Between January 1 and April 15, 1995, IRS received 236 million calls for tax assistance but was able to answer only 19 million of those calls. Our most recent report on telephone assistance accessibility offers several recommendations to improve IRS' ability to answer more taxpayer calls.\10 To determine whether accessibility was a problem during the 1995 filing season, we conducted two tests. One test was to determine the accessibility of the toll-free assistance for taxpayers who have questions about their accounts, the tax law, or IRS procedures. The second test was to determine the accessibility of the toll-free system that IRS tells taxpayers to call if they want copies of tax forms and publications. Our test methodology is described in appendix III along with (1) details on the results of our tests and (2) our computations of accessibility using more global IRS data. Results of both tests indicated that again this filing season taxpayers had significant problems reaching IRS by telephone. For example, of 2,821 calls we made to IRS' toll-free assistance number, we succeeded in reaching an assistor 249 times--a 9-percent accessibility rate. Although our test of the form ordering system produced better results--a 50-percent accessibility rate--there was still much room for improvement. As in past years, our measure of accessibility is based on the percent of incoming calls answered. We recognize that the number of calls coming in does not equal the number of taxpayers seeking assistance because many taxpayers are probably calling several times in an attempt to reach an assistor. We have been working with representatives from the Department of the Treasury and IRS to develop a better way to measure IRS' performance in terms of the number of taxpayers, but those efforts have not been completed. -------------------- \9 Tax Administration: IRS' 1992 Filing Season Was Successful but Not Without Problems (GAO/GGD-92-132, Sept. 15, 1992); Tax Administration: Increased Fraud and Poor Taxpayer Access to IRS Cloud 1993 Filing Season (GAO/GGD-94-65, Dec. 22, 1993); and Tax Administration: Continuing Problems Affect Otherwise Successful 1994 Filing Season (GAO/GGD-95-5, Oct. 7, 1994). \10 Telephone Assistance: Adopting Practices Used by Others Would Help IRS Serve More Taxpayers (GAO/GGD-95-86, Apr. 12, 1995). DOCUMENT IMAGING SYSTEM FAILED TO MEET EXPECTATIONS ---------------------------------------------------------- Letter :3.6 With one significant exception, the computer systems IRS used to process returns and remittances in 1995 generally performed without major problems. The exception was a new document imaging system that IRS used in 1995 to process several forms, including individual income tax returns filed on Form 1040EZ. To process tax returns more efficiently and economically, IRS intends to move from a system that relies on labor-intensive data transcription to one that relies on electronic data capture. Electronic filing and TeleFile are steps in that direction. For returns filed on paper, IRS plans to achieve its objective through document imaging. The Service Center Recognition/Image Processing System (SCRIPS) is the first of two planned document imaging systems.\11 Under IRS' new organizational structure, to be implemented over the next several years, paper tax returns are to be processed in only 5 of the 10 existing service centers. Those five sites are to be known as submission processing centers. Because imaging is the process IRS intends to use to capture data from all paper returns in the future, SCRIPS was installed in only the five service centers that are to be submission processing centers. Each of the five centers experienced hardware and software problems with SCRIPS. Those problems included hardware problems that kept documents from feeding properly into the scanner and software problems that affected SCRIPS ability to accurately capture name and address information. Two of the five centers completely stopped 1040EZ processing on SCRIPS, and the other three centers stopped processing for extended periods of time. Those stoppages caused IRS to redirect some 1040EZ processing workload back to its manual data entry system. In total, IRS was able to process only about 56 percent of the expected 8.6 million forms 1040EZ on SCRIPS that it had planned to process. As a result of the problems with SCRIPS, IRS has not yet realized the system's intended benefits. For instance, IRS had expected that increased processing rates would result in lower labor costs. However, IRS processed fewer forms 1040EZ per hour on SCRIPS in 1995 than it did in 1994 on the old system SCRIPS replaced. Thus, SCRIPS has not yet achieved any savings in labor costs associated with processing forms 1040EZ. In addition, IRS has postponed plans to redistribute additional workload to SCRIPS and to introduce the final form scheduled for SCRIPS. Appendix IV has additional information on the effects of problems with SCRIPS. Despite the many problems that limited SCRIPS effectiveness, IRS' "processing cycle time" indicator, which measures the average number of days it takes service centers to process returns, showed that service centers processed returns faster in 1995 than IRS expected. More specifically, IRS' data showed that the 10 service centers, in total, processed individual income tax returns in 1995 within a range of 5 to 9 days depending on the type of form (1040, 1040A, or 1040EZ) and the processing systems used (manual data entry or SCRIPS). That compares favorably with IRS' processing cycle time goal of 11 days. However, that comparison is misleading because IRS' 11-day goal was much higher than the 5- to 7-day cycle times the service centers had achieved in 1994. Comparing IRS' 1995 cycle times to its 1994 cycle times rather than to its goal for 1995 shows that the cycle times in 1995 worsened in many cases. In 1994, for example, none of the 10 service centers averaged longer than 9 days to process any type of individual income tax return. In 1995, six centers took longer than 9 days, including four of the five centers that had SCRIPS. Throughout the filing season, IRS officials worked with the SCRIPS contractor to remedy the hardware and software problems. At the conclusion of the filing season, they met to assess the cause of these problems and determine the actions needed to be taken before the next filing season. Among the actions being considered are upgrades to key components of the system that are intended to improve processing rates. We will continue to monitor IRS' efforts to address SCRIPS problems and the effect of these efforts on IRS' readiness for the 1996 filing season. -------------------- \11 SCRIPS is to be followed by the Document Processing System (DPS), which is intended to replace most of IRS' current labor intensive data transcription operations. DPS is being designed to image data from all types of returns as well as correspondence. Both SCRIPS and DPS are designed to create a digital image of a paper-based document and convert selected data into machine readable form, known as optical character recognition. CONCLUSIONS ------------------------------------------------------------ Letter :4 Although IRS' indicators point to a successful 1995 filing season, there were several problems that are not obvious from those indicators. IRS' assertion that it issued refunds on paper returns in 1995 as quickly as it did in 1994 (i.e., within an average of 36 days), masks the fact that in 1995, unlike 1994, millions of taxpayers had valid refunds delayed for up to 8 weeks. IRS chose to exclude those refunds in computing the refund timeliness indicator, even if IRS found no problem with the refund and eventually issued it, making the indicator an inaccurate measure of timeliness in 1995. Also, while we agree that IRS needs to ensure the validity of refund and EIC claims, we believe that IRS could have avoided some of the adverse reaction caused by the refund delays if it had done a better job alerting taxpayers that even refunds on accurate returns might be delayed. A related source of potential taxpayer confusion was the apparent conflict between IRS' promise, via its customer-service standards, to issue a refund within a certain number of days if the taxpayer filed a complete and accurate return and IRS' decision to delay certain refunds well beyond the promised time frame while it verified that the returns were complete and accurate. Likewise, IRS' ability to answer more calls than it estimated it could answer means little to the many taxpayers whose calls to IRS went unanswered or who gave up in frustration after receiving numerous busy signals. By focusing on the number of calls IRS expects to answer rather than the number of calls actually coming in or the number of taxpayers trying to reach IRS, the telephone assistance indicator provides a distorted picture of the accessibility of IRS' telephone service. IRS is working to develop a better measure of accessibility. Such a measure, once developed, would be a more meaningful indicator of IRS' telephone service during the filing season than the percent of scheduled calls indicator now used. Even though IRS reported success in meeting its returns- processing time frames, it did not achieve that success by following its plan to use the new SCRIPS equipment. IRS was only able to achieve its overall goals by rescheduling some workload back to its old manual data entry system. IRS has efforts under way to correct the SCRIPS problems. If those problems cannot be resolved, the scheduling of other forms on SCRIPS will be delayed even longer, resulting in further lost benefits the system was intended to provide. RECOMMENDATIONS TO THE COMMISSIONER OF INTERNAL REVENUE ------------------------------------------------------------ Letter :5 If IRS plans to continue validating SSNs and delaying refunds in 1996, we recommend that it adjusts its methodology for assessing refund timeliness to include delayed refunds associated with validly filed returns. Also after IRS develops a measure of taxpayer assistance accessibility that focuses on the number of incoming calls and/or the number of taxpayers calling for assistance, we recommend that it includes that measure among its key filing season performance indicators. AGENCY COMMENTS AND OUR EVALUATION ------------------------------------------------------------ Letter :6 We requested comments on a draft of this report from the Commissioner of Internal Revenue or her designated representative. Responsible IRS officials, including the Assistant Commissioner for Taxpayer Services, Director of Investigations (Tax Refund Fraud), Electronic Filing Executive, Director of Tax Forms and Publications, and Senior Program Analyst (Submission Processing), provided IRS' comments in a November 13, 1995, meeting. The Chief of Taxpayer Services provided additional comments on November 30. IRS also provided a few factual clarifications that we have incorporated in this report where appropriate. The Chief of Taxpayer Services noted that IRS has emphasized the importance of having accurate SSNs on tax returns filed in 1995 by including a message on the cover of all tax packages and through many public service announcements. Our report acknowledges that fact. However, our concern is with the lack of sufficient warning to taxpayers that their refunds might still be delayed even if they had accurate SSNs on their tax returns. The Chief acknowledged that taxpayers who filed complete and accurate returns also had their refunds delayed to allow IRS additional time to verify the claims before issuing the refunds, and he said that IRS regretted any inconvenience. Officials at the November 13 meeting mentioned that there was a lot of discussion within IRS, before the 1995 filing season, about how much IRS should divulge about its plans. They also noted that by the time IRS had finalized its plans for 1995 it would have been too late to make any changes to the tax packages and Publication 17, which had already been printed. They said that even if IRS had decided to tell taxpayers more, it would have been too costly to reprint those documents. IRS said that it plans to continue validating SSNs and delaying refunds in 1996 but has revised its SSN-validation procedures and criteria. Thus, it expects that taxpayers with valid SSNs will have only a small chance of having their refunds delayed in 1996. Because of those changes, IRS saw no need to revise its methodology for assessing refund timeliness. We agree that IRS would not have to revise its methodology if those changes have the expected result of limiting the extent to which valid refunds are delayed. The officials acknowledged, however, that if that result is not achieved, the methodology would have to be adjusted. We will be monitoring the impact of IRS' revised procedures during our assessment of the 1996 filing season. The Chief of Taxpayer Services noted that IRS has been working with us to develop appropriate measures and had proposed that the accessibility of its toll-free telephone service be measured in three ways: (1) the percentage of individual callers served; (2) the number of attempts made by successful callers, expressed in the form of a range; and (3) the disposition of all calls, whether they were answered, received a busy signal, or were abandoned. Appendix III of this report includes a discussion of IRS data on accessibility using those three measures. The Chief said that IRS would continue working with us to finalize these measures and that, given those continuing discussions, IRS felt that our recommendation was premature. We disagree. IRS has already developed measures, as indicated above, and those measures represent reasonable indicators of the accessibility of IRS' toll-free telephone service. Our continuing discussions with IRS are not centered on the measures themselves but on the reliability of the data used for those measures. Our recommendation merely seeks a commitment from IRS that one or more of those measures, once finalized, be included among IRS' key filing season performance indicators. We do not believe it is premature to seek that commitment. Our draft report also included two proposed recommendations that were intended to provide taxpayers with better information on potential refund delays in 1996. We proposed that if IRS planned to continue validating SSNs and delaying refunds in 1996, it (1) clearly alerts taxpayers, in the 1040 tax package and Publication 17, to the possibility that their refunds will be delayed even if there are no problems with the SSNs provided on their returns and (2) reconciles the inconsistency between those refund delays and IRS' customer-service standard. In commenting on the proposed recommendations, IRS said that the problem we identified in 1995 with respect to adequately alerting taxpayers should not recur in 1996 because of the aforementioned changes to IRS' SSN-validation procedures and criteria. IRS has however, revised its customer-service standard on refunds by including a caveat to alert taxpayers that their refunds may be delayed if their returns are selected for further review. The revised standard has been included in the tax packages and Publication 17 for tax year 1995 (those that taxpayers will use in preparing returns to be filed in 1996). Assuming that IRS is correct in believing that its revised procedures will cause few taxpayers with valid SSNs to have their refunds delayed, we believe that further action is unnecessary. Accordingly, we have deleted the two proposed recommendations from our final report. ---------------------------------------------------------- Letter :6.1 We are sending copies of this report to various congressional committees, the Secretary of the Treasury, the Commissioner of Internal Revenue, the Director of the Office of Management and Budget, and other interested parties. Major contributors to this report are listed in appendix V. Please contact me on (202) 512-9110 if you have any questions. Sincerely yours, Lynda D. Willis Director, Tax Policy and Administration Issues PROCESSING TAX PAYMENTS THROUGH LOCKBOXES =========================================================== Appendix I IRS envisions that by the year 2001, 90 percent of tax payment processing will be done by lockboxes. Under this concept, which is already being used for some types of tax payments, taxpayers are to mail payments to a lockbox, which is a postal rental box serviced by a commercial bank. The bank processes the payments and transfers the funds to a federal government account. The payment and payer information is then recorded on a computer tape and forwarded to IRS where the tape is to be used to update taxpayers' accounts on IRS' master file. IRS conducted two lockbox tests during the 1995 filing season to assess taxpayers' willingness to use different procedures for mailing tax payments associated with their returns. For each test, IRS sent special Form 1040 packages to specific taxpayers. These packages included (1) mailing instructions that were different for each of the two tests and (2) a payment voucher that could be scanned by optical character recognition equipment. One test package contained one return envelope with two different tear-off address labels--one label addressed to the lockbox was to be used for a return with a tax balance due, while the other label addressed to the service center was to be used for a return with a zero balance or with a refund due to the taxpayer. Taxpayers with balance-due returns were instructed to include the return, payment, and voucher in one envelope and to affix the label addressed to the lockbox. The bank that serviced the lockbox separated the return from the payment, deposited the payment, recorded the payment information on a computer tape, and forwarded the return and the computer tape to IRS for processing. The other test package used two envelopes--one addressed to the service center, the other addressed to the lockbox. All taxpayers were instructed to send their returns in the envelope addressed to the service center. Taxpayers who owed a balance were to use the second envelope to send their payments and vouchers to the lockbox. The bank processed the payment and voucher as described above. As of mid-June 1995, IRS had not yet received the management information needed to evaluate the two lockbox tests. However, IRS had already made decisions to (1) continue testing the two-label method in certain tax packages for the 1996 filing season, (2) include a voucher inside every 1996 Form 1040 tax package (except 1040A and 1040EZ),\12 (3) instruct practitioners to send all returns with remittances, no matter what 1040 tax form they are associated with, to the lockbox, and (4) implement the two-envelope method in all 1040 packages (with the possible exception of Forms 1040A and 1040EZ) starting with the 1997 filing season. According to an IRS official, the purposes of including a standard voucher in tax packages not included in the lockbox test are (1) to familiarize taxpayers with the use of a voucher and (2) to lighten the workload being processed through the old remittance processing system (RPS) at service centers. IRS plans to use a newer system, RPSII, to scan the scannable vouchers sent to the service centers from the test tax packages. According to an IRS official, the two-envelope method will not be used for the 1996 filing season because IRS cannot easily determine if the return inside the envelope is one that involves a refund. That determination is important because IRS gives priority to refund returns to help ensure that the return gets processed and the refund gets issued before the government has to pay interest on the refund. In the past, a service center knew a return did not involve a refund if it opened the envelope and found a check inside. Under the two-envelope system, the service center only receives the tax return and thus has no quick way to isolate those returns involving payments from those involving refunds. For the 1997 filing season, IRS is considering redesigning the tax forms to help service centers more easily identify the type of return received. According to information obtained from IRS, the use of lockboxes to process remittances associated with Forms 1040 in 1995 resulted in an interest cost avoidance of about $44.3 million by getting money deposited faster through the lockbox. This means that the Treasury did not have to borrow this money to pay towards certain government obligations. At the same time, according to IRS, it cost about $3.4 million to process those remittances through the lockboxes, leaving a net savings of about $40.9 million. IRS expects that the amount of interest cost avoidance will decrease each year as the lockboxes take on higher volumes of remittances thereby slowing the banks' productivity. As the program is expanded to all types of tax packages, volumes at the lockbox will increase while average dollar amount remitted will decrease. Bank costs associated with the larger volumes are also expected to increase. Treasury Financial Manual Bulletin No. 94-07, dated March 1, 1994, provides that if the interest cost avoidance of a lockbox's accelerated deposits is less than the cost charged by the lockbox, the agency (in this case, IRS) is required to pay all lockbox bank charges, other than those needed to maintain a regular bank account. Otherwise Treasury's Financial Management Service (FMS) pays the charges. Because the amount of interest cost avoidance resulting from IRS' lockbox program has exceeded the related bank charges, FMS has paid those charges. According to an IRS National Office official responsible for the lockbox program, neither IRS nor FMS expects the amount of interest cost avoidance in the future to fall below the amount of bank charges. -------------------- \12 According to an IRS official, forms 1040A and 1040EZ are being excluded because historical data indicate that only 14 percent and 1 percent respectively of taxpayers who file those forms will have a remittance. IRS EFFORTS TO DETECT AND DETER REFUND FRAUD ========================================================== Appendix II In 1995, IRS expanded its efforts to combat refund fraud. Much of what IRS did involved verifying SSNs, with an emphasis on returns claiming the EIC. IRS was looking for missing SSNs, SSNs that did not match the Social Security Administration's records, and SSNs that had already been used on another return filed in 1995. As we discussed in a June 1995 testimony\13 before the Senate Finance Committee, the expanded procedures for selecting paper returns to verify SSNs identified many problem returns, but some that should have been selected for SSN verification were not. In total, IRS identified approximately the volume of paper returns with invalid SSNs that it had expected to handle during the filing season, but volumes fluctuated widely among IRS service centers. For example, one service center received about 360 percent of its expected volume, while another received only 61 percent. As a result, service centers used somewhat different criteria for determining which taxpayers would be asked to verify SSNs and to provide additional evidence of their EIC eligibility. Computer problems also occurred during the filing season, which caused some returns not to be selected for SSN verification when they should have been. IRS also experienced some problems as it began checking for duplicate SSNs. These problems included difficulties in constructing the database to identify duplicate SSNs, poorly organized computer listings that enforcement personnel found difficult to use, and cumbersome procedures for coordinating the work of different IRS service centers. IRS is analyzing the results of the 1995 initiative and plans to make changes for 1996. Further automation of the process is a primary goal. We were not able to assess the success of IRS' initiatives. At the time we completed our audit work, information was not yet available on such things as the number of (1) duplicate SSNs identified and resolved by IRS, (2) EIC claims adjusted or withdrawn after IRS questioned a taxpayer about an SSN, or (3) erroneous SSNs corrected as a result of IRS' efforts. Some information was available, however, that sheds light on the results of IRS' efforts. According to IRS: As a result of the 6-to-8 week delay on EIC refunds, IRS was able to stop an additional $6 million in fraudulent refund claims that, in past years, would have been issued before IRS had detected the fraud. IRS had received 18.9 million EIC claims as of the end of September 1995, compared with 14.8 million claims at the same time in 1994. All of that increase was due to a legislative change that made persons without qualifying children eligible for the credit in 1995. IRS had expected to receive about 20 million claims in 1995, including about 5.3 million from persons without qualifying children. EIC claims in 1995 totaled about $20.9 billion as of September 30 compared with about $15.2 billion as of October 1, 1994. Only about 12 percent of that increase was attributed to claims from taxpayers with no qualifying children. As a result of IRS' scrutiny of EIC claims, 3.2 million taxpayers received their refunds in two checks because the EIC portion of their refund was temporarily delayed. IRS tracked 400 returns that had been rejected by the electronic filing system, and found, among other things, that 113 (28 percent) of the individuals involved subsequently filed on paper, using the same SSN that had been rejected by the electronic filing system, and were issued a refund. -------------------- \13 Earned Income Credit: Noncompliance and Potential Eligibility Revisions (GAO/T-GGD-95-179, June 8, 1995). TELEPHONE ACCESSIBILITY ========================================================= Appendix III To assess the ability of taxpayers to reach IRS by telephone to ask a question about the tax law or their accounts or to order forms or publications, we conducted two tests--one of IRS' toll-free telephone assistance system and the other of IRS' toll-free form-ordering system. To conduct the tests, we placed telephone calls at various times during each workday from January 30 through February 11 and from April 3 through April 15, 1995. We made our calls from seven metropolitan areas--Atlanta; Chicago; Cincinnati; Kansas City; New York; San Francisco; and Washington, D.C. Each attempt to contact IRS consisted of up to five calls at 1-minute intervals. If we reached IRS during any of the five calls and made contact with an assistor, we considered the attempt successful. If we reached IRS during any of the five calls but were put on hold for more than 7 minutes without talking to an assistor, we abandoned the call, did not dial again, and considered the attempt unsuccessful. If we received a busy signal, we hung up, waited 1 minute, and then redialed. If after four redials (five calls in total) we had not reached IRS, we considered the attempt unsuccessful. TOLL-FREE TELEPHONE ASSISTANCE ------------------------------------------------------- Appendix III:1 We tested the accessibility of the toll-free telephone assistance system IRS tells taxpayers to call if they have a question about their account, the tax law, or IRS procedures. Of 745 attempts to contact an assistor, 249 (33 percent) were successful--87 on the first call, 55 on the second call, and 107 after 3 to 5 calls. In another 89 cases (12 percent), we got into IRS' system but were put on hold for more than 7 minutes and thus hung up before making contact with an assistor. The remaining 407 attempts (55 percent) were aborted after we received busy signals on each of our 5 dialings. Our 745 attempts to contact an assistor required a total of 2,821 calls to IRS' toll-free telephone number. Of those 2,821 calls, we succeeded in getting through to an IRS assistor 249 times--a 9-percent accessibility rate. In conducting our test, we did not ask questions of the assistors because it was not our intent to assess the accuracy of their assistance. IRS does its own test of accuracy, and we have assured ourselves in the past about the reliability of IRS' methodology. IRS' test data for 1995 showed an accuracy rate of 90.1 percent as of April 15, 1995. That compares with a rate of 89 percent for the same period in 1994. FORM-ORDERING SYSTEM ------------------------------------------------------- Appendix III:2 One way taxpayers can obtain tax forms and publications is to place an order through IRS' telephone form-ordering system. The order will then be filled by one of IRS' three forms distribution centers. To determine the level of service IRS provides to taxpayers trying to access this ordering system, we conducted another test using the same procedures used for the first test. Our results showed that the form-ordering system was much more accessible than the toll-free telephone assistance system. However, there was still much room for improvement. Of 484 attempts to contact a distribution center representative, 443 (91.5 percent) were successful--299 on the first call, 76 on the second call, and 68 after 3 to 5 calls--and 41 (8.5 percent) were aborted after five dialings. We did not abandon any calls when placed on hold because we were not held waiting for more than 7 minutes. Our 484 attempts to contact a representative required 883 calls. Of those 883 calls, we succeeded in getting through to an IRS representative 443 times--a 50-percent accessibility rate. As with the first test, our intent was to determine how easy it was to reach IRS over the telephone. We did not assess how well the distribution centers filled orders for tax forms and publications because (1) our checks in recent years showed that IRS was doing a good job of filling orders, (2) IRS contracts for its own test of distribution center performance, and (3) our prior review of the contractor's methodology resulted in changes that have improved its reliability. The contractor measures the length of time from when an order is placed until the contractor receives notification about that order (either by full or partial receipt of the material ordered or notification that the material has been back ordered). The contractor also measures accuracy by comparing the items ordered with those received. The contractor's results for the first part of the fiscal year showed that (1) it took the distribution centers an average of 16 days to fill an order, which is within IRS' stated time frame of 9 to 21 days and (2) 97.9 percent of the orders were filled correctly, which exceeded IRS' goal of 96.5 percent. IRS' DATA ON ACCESSIBILITY CONFIRMS OUR TEST RESULTS ------------------------------------------------------- Appendix III:3 We have been working with representatives from the Department of the Treasury and IRS to develop a better way to measure the accessibility of IRS' telephone service. Although there are still some issues to be resolved, such as how to best measure the number of times a caller had to dial before reaching an assistor, the data compiled by IRS for 1995 confirmed the results of our tests. IRS summarized its data as follows: "For the period January 1, 1995, to April 15, 1995, an estimated 46.9 million callers made 236.1 million call attempts to IRS for assistance. This equates to an average of 5 attempts per caller. We answered 19.2 million calls which represents 41 percent of the callers. Of the 19.2 million callers who received an answer, 50 percent were answered within approximately 1 attempt; 75 percent were answered within approximately 5 attempts." "Of the 236.1 million attempts, 19.2 million received an answer, which represents 8 percent of the total attempts. The remaining 216.9 million call attempts either received busy signals or were terminated by the callers because they did not want to wait in queue for an assistor." As shown in figure III.1, IRS' reported accessibility rate of 8 percent continued a downward trend since 1989 and was 13 percentage points below 1994. However, the 1995 accuracy rate on answers to tax law questions continued an upward trend.\14 Figure III.1: Comparison of Toll-Free Telephone Accessibility and Accuracy During the 1989 Through 1995 Filing Seasons (See figure in printed edition.) Source: IRS' Management Information System for Top Level Executives and IRS' Telephone Data Reports. -------------------- \14 In a July 1995 report, IRS' Internal Audit identified one area where telephone assistors apparently had problems--at least early in the filing season. Internal Audit made 92 test calls between January 10 and February 7 to assistors nationwide with questions about changes to EIC eligibility that were effective with income tax returns filed in 1995. In 28 (30 percent) of the 92 calls, according to Internal Audit, assistors either incorrectly advised taxpayers about their eligibility for EIC or advised them incorrectly on related tax issues. SERVICE CENTER RECOGNITION/IMAGE PROCESSING SYSTEM (SCRIPS) ========================================================== Appendix IV SCRIPS is a multimillion dollar system\15 designed to process income tax returns filed on Form 1040EZ and other IRS documents\16 by electronically scanning the document, capturing the data, and storing an image of the scanned document. SCRIPS was tested in Cincinnati in 1994 and used in five processing centers--Austin, Cincinnati, Kansas City, Memphis, and Ogden--in 1995. In conjunction with the implementation of SCRIPS, IRS consolidated the processing of IRP documents at the five SCRIPS centers and FTD coupons at four of the five SCRIPS centers. IRS continued to process forms 1040EZ at all 10 service centers but planned to consolidate 1040EZ processing in the five SCRIPS centers by 1996. IRS planned to start processing all forms 941 received at the five SCRIPS centers in July 1995 and redistribute 100 percent of the forms 941 workload from non-SCRIPS centers by 1996. IRS planned to process 76.4 million FTDs, 57.4 million IRP documents, and 8.6 million forms 1040EZ on SCRIPS during the 1995 filing season. IRS expected that SCRIPS would provide faster and more accurate document processing, lower maintenance costs, reduce manual data entry, lessen error correction, and minimize document storage requirements. But, extensive downtime and slower-than-expected processing rates during the filing season limited the effectiveness of SCRIPS. The impact of these problems was most felt in the processing of forms 1040EZ. Some centers stopped 1040EZ processing on SCRIPS completely or for extended periods of time. As a result, IRS was able to process only about 56 percent of the expected 8.6 million forms 1040EZ on SCRIPS. Although the centers were able to process the rest of the forms 1040EZ on their old systems, doing so required additional resources and costs, and some centers reported that the average time it took to process a return increased because of the SCRIPS problems. Processing center officials told us of budget overruns as a result of slower-than-expected SCRIPS processing times. IRS had scheduled 25.6 staff years for processing other-than-full-paid forms 1040EZ but used 66.5 staff years.\17 During the 1995 filing season, IRS processed 64 forms 1040EZ per hour, 28 percent slower than the 89 documents per hour processed in 1994 on the systems that SCRIPS replaced. An official at one processing center told us that as a result of the problems with SCRIPS, the center had to (1) delay furloughing seasonal staff, (2) work 2 additional weekends of overtime (about 18,000 additional overtime hours) to get returns processed within established time frames, (3) reinstall old optical character recognition equipment and add additional terminals at a cost of about $4,300, and (4) train 163 additional employees on how to use the old processing systems. IRS' Internal Audit issued a report on IRS' 1994 SCRIPS test that cited several factors that may have contributed to the problems encountered in 1995.\18 Internal Audit found that (1) SCRIPS had not been fully tested to meet output and storage requirements, (2) IRS accepted the system without conducting required acceptance and equipment testing, and (3) SCRIPS was not meeting contractual requirements for capturing Form 1040EZ and IRP data accurately. Had IRS conducted the proper testing, many of the problems encountered during the 1995 filing season might have been identified and corrected before system implementation. At the conclusion of our audit work, IRS was assessing SCRIPS performance to identify problem causes and needed corrective action. In the meantime, IRS postponed plans to process Form 941 on SCRIPS and redistribute 1040EZ workload from the five centers that do not have SCRIPS. We will be monitoring IRS' efforts to improve SCRIPS performance, especially as they affect IRS' readiness for the 1996 filing season. -------------------- \15 SCRIPS is the first of two document imaging systems intended to improve IRS' ability to capture data electronically and streamline much of its paper-processing operations. IRS' estimate of the cost to design, develop, and maintain SCRIPS through the year 2000 is $132 million. SCRIPS is to be followed by the Document Processing System, which is intended to replace most of IRS' current labor intensive data transcription operations. The Document Processing System is being designed to image data from all types of individual and business returns. \16 Other documents include federal tax deposit (FTD) coupons; Information Returns Program (IRP) documents, such as forms 1099 used by banks and other third parties to report payments of interest, dividends, etc.; and employer's quarterly federal tax returns (Form 941). \17 "Other-than-full-paid" tax returns are those returns filed by taxpayers who either were due a refund or did not pay the full amount of tax owed at the time of filing. These returns comprised about 92 percent of the forms 1040EZ filed at the five submission processing centers during the 1995 filing season. \18 Interim Evaluation of the Service Center Recognition/Image Processing System (SCRIPS) Pilot, IRS Internal Audit, Reference No. 054406, May 8, 1995. MAJOR CONTRIBUTORS TO THIS REPORT =========================================================== Appendix V GENERAL GOVERNMENT DIVISION, WASHINGTON, D.C. David J. Attianese, Assistant Director, Tax Policy and Administration Issues Monika R. Gomez, Evaluator Christopher E. Hess, Evaluator KANSAS CITY REGIONAL OFFICE Doris J. Hynes, Evaluator-in-Charge H. Yong Meador, Evaluator Marge Vallazza, Reports Analyst *** End of document. ***