16.13 GST registrants must submit returns to Revenue Canada on a monthly, quarterly or annual basis, depending on their sales volumes. GST charged to customers and input tax credits claimed on purchases are both recorded on the return. If the total GST charged by a registrant exceeds the input tax credits claimed, the registrant is expected to remit the excess to Revenue Canada. This is known as a "debit return." If input tax credits claimed exceed the GST charged, the registrant seeks a refund from Revenue Canada. This is known as a "credit return."
16.14 Depending on the nature of their operations and their circumstances, registrants may also be eligible for various rebates. For example, municipalities are eligible for a rebate designed to lower the effective rate of GST applied to them. Rebate entitlements may be included in GST returns and, if they are large enough to offset the net GST otherwise payable, they will produce a credit return. In other cases, rebate entitlements are claimed on a separate form.
16.15 In May 1999, there were approximately 2,411,000 active GST registrants. Revenue Canada administers the GST for the approximately 1,925,000 registrants outside Quebec. Under a 1991 agreement between the Government of Canada and the Government of Quebec, the ministère du Revenu du Québec administers the GST on behalf of Revenue Canada for approximately 486,000 registrants in Quebec. The ministère uses Revenue Canada's guidelines and policies but produces its own audit strategies and administrative procedures. The GST administered by the ministère is recorded in Revenue Canada's databases.
16.16 The Harmonized Sales Tax (HST) collected in three of the Atlantic provinces is an integral part of the GST program. For simplicity, in this chapter the term GST also includes HST.
16.18 Until recently, Revenue Canada's Summerside Tax Centre was responsible for processing all GST returns and rebate forms sent in by non-Quebec registrants. Revenue Canada has begun to decentralize a portion of its GST processing activities to tax centres across the country. Rebate forms will continue to be processed in Summerside.
16.20 Automated validity checks select about 30 percent of the credit returns and rebate forms. These items are automatically referred to tax services offices across Canada according to the registrant's location (see Exhibit 16.2). There, a departmental officer brings up on a computer screen information about the selected items. The officer examines each item, and considers such factors as industry type, prior filing history and prior audit results. The officer then decides whether to pay the amount claimed (called "waiving" the item) or to conduct a more in-depth examination. About 97 percent of the items are waived. Generally, for the approximately 3 percent of items requiring further examination, an officer contacts the registrant to ask for more information and/or documentation. If satisfied, the officer approves the item for payment; if not satisfied, the officer adjusts or rejects the item.
16.22 Post-payment audits are handled by the tax services offices (see Exhibit 16.2). Relying heavily on their knowledge of the registrant population, external sources, leads from other areas of the Department - for example, pre-payment audit or income tax - and analysis of departmental databases, teams of post-payment auditors select registrants for examination. They normally select registrants who are expected to represent a high risk of non-compliance.
16.23 In 1997-98, the GST pre-payment and post-payment audit programs resulted in assessments of about $520 million. The net revenue assessed was approximately $624 per direct audit hour.
16.26 GST registrants have to file returns monthly, quarterly or annually, depending on their sales volumes. Returns are generally due toward the end of a month, resulting in 12 peak volumes a year, with higher peaks for the quarterly and annual periods. To accommodate this fluctuation, the Department must be flexible when allocating resources. We found that the rate at which documents are processed varies during the year but, as noted in Exhibit 16.4, about 90 percent of the documents get processed within 16 days and most of the remaining workload is processed in the next 45 days. The percentage of transactions processed within 16 days has generally increased.
16.27 Managers and employees in Summerside and at headquarters have worked together to streamline several workflows. Efficiency gains have been made in several areas: mail extraction processes have improved; redundant document preparation work has been eliminated; data capture methods have been improved; and error correction procedures have been enhanced.
16.28 We calculated that the unit cost of processing a standard return that has no errors fell from about $0.66 in 1995-96 to about $0.37 in 1998-99. Similarly, for returns accompanied by a payment but containing errors, the unit cost of processing fell from approximately $3.48 in 1995-96 to about $1.91 in 1998-99. Taking 1998-99 document volumes and using the 1995-96 and 1998-99 unit costs, we calculated that efficiency gains between the two years amounted to about $2 million.
16.30 About 40 performance standards were recently established for processing operations at the Summerside Tax Centre. However, only volume output information is used. Very few efficiency measures have been developed and reported. Also, standards deal mostly with units completed per hour but do not specify the expected level of quality, such as percentage of errors and percentage completed in a specified time period. Another shortcoming of some standards is that they may be aggregated to a point where relevant information is lost. For example, the focus is on the number of documents in the work-in-progress file as opposed to, say, the age of the work-in-progress items at peak and non-peak times for given types of transactions.
16.31 Since 1993, Revenue Canada has not undertaken benchmarking activities with other large processing centres to assess the quality of its practices and technologies. The performance standards have not been compared with similar measures used in other government processing operations such as the ministère du Revenu du Québec, or in private industry. In the context of the new Canada Customs and Revenue Agency, benchmarking against similar organizations could indicate whether some service delivery options - such as contracting out - would further improve efficiency.
16.32 Revenue Canada should update its GST returns processing standards to include quality, accuracy, timeliness and unit cost, and should measure performance accordingly. Where appropriate, standards should differentiate between peak and non-peak times and among various types of transactions. Standards should presume the use of efficient processing procedures and, where possible, performance should be benchmarked against similar organizations.
Department's response: The Department fully agrees with the importance of standards for returns processing. In 1997-98 the Department began a project to significantly improve GST/HST returns processing and establish performance measures using the Balanced Scorecard approach. Existing processes were compared with best practices in other revenue lines within Revenue Canada and concepts were imported from elsewhere, in particular the manufacturing industry. These improved processes are captured in the revised GST/HST Accounting and Processing Procedures (GAPP) manual that describes how processing is now being performed. As noted in paragraph 16.54, these changes have already produced measurable improvements.
Key activities were identified as part of the development of the new GAPP, and as noted in paragraph 16.30, about 40 performance standards have been established. The next step will be to develop a Balanced Scorecard of standards, including quality, accuracy and timeliness. The first task will be to confirm the desired quality and accuracy characteristics through a survey of client needs, and benchmarking of performance levels attained by other revenue administrations and private industry. It should be noted that the move of GST/HST processing to multiple tax centres provides the benefit of closer internal benchmarking against other revenue lines and between centres.
The Department will carefully take this recommendation into consideration as it continues its development of a balanced set of performance standards for GST processing. Service standards and actual performance will be published regularly.
16.34 We found that there can be wide discrepancies in production information for a given year across systems, and over a number of years. For example, the ARMS production statistics for the months of August 1997, April 1998 and May 1999 showed no production even though the Summerside Tax Centre was operating during those months. Such discrepancies are due to inconsistencies in data input between systems at Summerside, and are not corrected because the data contained in ARMS are not reconciled with data contained in Summerside's local system. Management is aware of the discrepancies between the systems.
16.35 To establish a unit cost, we relied primarily on data supplied by management at the Summeside Tax Centre, since these data come directly from automated systems that form part of Summerside's regular reporting systems. The data do not have the large time lags that the ARMS data have.
16.36 Revenue Canada is completing the phased implementation of a Common Administrative System, which will eventually replace the Department's entire Resource Management Information System, of which ARMS is a component. It is still uncertain how many of the currently available financial reports the new system will be able to deliver immediately. The Department plans to temporarily maintain certain processing and reporting components of ARMS to ensure that the majority of currently available and useful financial reports will continue to be accessible.
16.38 Monitoring the performance of returns processing operations requires a variety of performance measures. One of these measures ought to be the amount of interest the Department has paid out each year on overdue refunds, compared with a target amount. However, the Department has not set a target for what it would consider a reasonable interest cost of administering GST.
16.39 Not all of the interest cost is attributable to the Summerside Tax Centre. As well as the number of days it takes to process the documents at the Centre, interest costs are influenced by the number of days required for the tax services offices to review refund claims rejected by the automated validity checks (see paragraphs 16.70 to 16.73) and the number of days required for Public Works and Government Services Canada to issue the cheques. Nevertheless, any reduction in the time the Summerside Tax Centre takes to process its documents would contribute to reduced interest costs and improved service to registrants.
16.40 The Department has carried out studies on how to achieve its turnaround standard for processing documents while maintaining workload at a consistent level, but it has not analyzed its processing activities specifically to find ways to reduce interest costs.
16.41 Aside from generally improving the timeliness of processing, one way to reduce interest costs would be to give priority in processing and error correction to the large refund claims; these result in high interest costs if they are not processed speedily. This would be consistent with the current practice of giving high priority in the mail extraction procedures to cheques over $25,000.
16.42 Using data supplied by Revenue Canada, we found that in 1997-98, refunds were issued to registrants on about 2 million credit returns and rebate forms. The Department estimates that about 30 percent of these refunds included interest. The actual interest expense for the year was $9.8 million, or about $16.33 per cheque on which interest was paid. In the management of items awaiting data capture and in the processing of mail received from tax services offices, we calculate that, as a rough order of magnitude, there is scope to reduce interest costs by between $800,000 and $4 million per year.
16.43 Improving the processing of documents can reduce not only interest but also other costs, and can improve service to registrants. For example, after a set date following the end of the filing period, the GST system automatically generates notices to registrants who have not filed their returns. Even if returns have been received, if they are not processed within this timeframe the system automatically generates the notices and incurs additional expenses for printing and mailing. This also results in increased salary costs for client service officers at the Summerside Tax Centre and at tax services offices (who must respond to phone inquiries from registrants) and error correction staff at Summerside (who must deal with second copies of returns submitted by registrants).
16.44 While reducing interest costs is a valid objective, it must be balanced against the potential monetary loss associated with the payment of improper refund claims if adequate time is not allowed for examination of the claims.
16.45 Revenue Canada should monitor and control the amount of its portion of interest charges, and should work with Public Works and Government Services Canada to find ways to better manage interest costs and improve service to registrants.
Department's response: The Department agrees with the need to monitor and control interest costs. Interest costs are incurred when all stages of the process - returns processing, pre-payment audit and cheque production - take longer than the time provided for in the Excise Tax Act. Because interest costs are the result of processing delays and not their cause, the Department believes that processing time should be the primary performance indicator and thus the focus of attention for process improvements. Concentrating directly on the reduction of interest costs could unfairly skew the focus of processing; for example, the rapid handling of large-business returns would become more important than small-business returns. The Department will therefore control the level of interest costs by focussing on improved timeliness and quality for all business clients so that a fair allocation of the benefits is achieved.
The Department continues to seek process improvements that will reduce the initial processing time, and the time taken to produce the cheque. Improvements in these two areas will allow more adequate time for consideration of the validity of the claim. Taken together, these improvements will reduce the risk of improper payments and the total time taken to process returns. The results of returns processing improvements to date are already evident, as reported in 16.54, and work will continue in this area. The Department also continues to work with Public Works and Government Services Canada (PWGSC) to reduce the time taken to process payments, and significant progress has already been recorded. For example, Revenue Canada now provides PWGSC with daily payment files, and cheques and direct deposits are produced within three business days. In the near future, PWGSC will offer electronic data interchange payments that will provide next-day deposit into the client's bank account.
16.47 We did not audit the entire realignment process. However, we understand that Revenue Canada decided to reduce the number of tax centres that were processing income tax returns because the introduction of electronic filing of personal income tax returns and other initiatives had left an excess of processing capacity.
16.48 Revenue Canada officials informed us that the rationale for partially decentralizing GST processing when it realigned income tax processing was management's aim to provide in future a single tax form combining all types of instalments and a GST return, and a single statement of account for all taxes. Decentralization would also allow each taxpayer to deal with only one tax centre. The Department also wished to ensure that each tax centre, except for Ottawa, would maintain employment at approximately the current level.
16.49 The Department has told us that it did not perform a cost-benefit analysis of the available alternatives for decentralizing GST returns processing. It wished to put in place a fairly uniform organizational structure among tax centres. This would facilitate the management of all tax streams.
16.50 We expect that there will be some negative consequences of decentralizing GST returns processing. These consequences include initial start-up costs such as purchasing additional equipment and training staff. Also, the processing of GST rebate forms is to remain at Summerside. This means that registrants who file regular GST returns and rebate forms will have each document processed at a different location. This could cause difficulties in matching documents, increased costs to the registrants and, if there are delays, increased interest costs on overdue refunds and rebates.
16.51 Decentralizing GST processing means that functions will be performed at several sites across the country, with lower volumes than at Summerside. Moreover, should electronic filing of GST be introduced, the volume of work would decrease further. This raises the possibility that unit costs of processing may increase. Departmental officials informed us that, to deal with the possibility, they conducted a thorough review of Summerside's procedures for processing GST returns and developed a plan to simplify the activity in a way that could be transplanted to other tax centres. This meant, for example, finding substitutes for certain local area network-based computer applications that at present can be run only at Summerside.
16.52 A key aspect of simplifying the process was to pave the way for "lean production" techniques that can achieve low unit cost in an environment characterized by small volumes of a variety of transactions. This will require each tax centre to develop innovative processing techniques to deal with unique situations it may face and, where appropriate, to share such "best practices" with other tax centres.
16.53 In November 1998, the Department established a GST pilot project at Surrey Tax Centre. This project was not intended to determine whether or not decentralization ought to be undertaken. The project was intended solely to iron out the practical difficulties of transplanting processing operations.
16.54 To determine the likely impact of decentralization on processing costs, we compared the unit cost of two activities - data capture and error correction - in the Summerside Tax Centre for the period April 1998 to October 1998 (before the pilot project commenced) with the unit costs at both Summerside and Surrey during the pilot for the period April 1999 to June 1999. Our comparison indicates that hourly production rates were higher in both tax centres during the pilot than Summerside had achieved previously. This improved performance may be due to Revenue Canada's efforts to simplify processing.
16.56 The pre-payment audit program has been in existence since the GST began. It is intended to prevent the inappropriate payment of refunds claimed on credit returns or rebate forms. Refund claims are subjected to a series of automated validity checks. A claim that fails one or more checks is referred automatically to the appropriate tax services office. There, a departmental official reviews the claim and decides either to waive it (in which case the claim will be paid) or to have it subjected to an in-depth review and possibly a pre-payment audit.
16.58 During our visits to tax services offices, officials told us that pre-payment audits are a significant source of leads for future GST and income tax audits. Our sample of post-payment audits (see paragraph 16.77) indicated that the largest identifiable source of these audits was the pre-payment audit program.
16.59 As an indication of how effective pre-payment audits are at detecting non-compliance, consider these results. In 1997-98, pre-payment audits prevented inappropriate payments at the rate of $1,518 per direct audit hour. These included large refund claims that had been requested by some well-known and established businesses, and that were found to be inappropriate and payment prevented after only a few hours of work. The $1,518 per hour represents a net cash saving: since the pre-payment audits prevent payments from being made, there is no potential for bad debts.
16.60 We are pleased to note that Revenue Canada is increasing its pre-payment audit activities. Resources allocated to the program in 1999-2000 have increased by 9 percent over 1997-98. In addition, the Department has initiated an enhanced compliance initiative, the main focus of which is to identify and address potentially abusive refund claims.
16.62 A high percentage of credit returns and rebate forms continue to be rejected by the automated validity checks. Approximately 30 percent of the credit returns are rejected by the automated validity checks (see Exhibit 16.6). This indicates that the checks are not particularly discriminating in their initial targeting of inappropriate refund claims. If the checks were refined to better target the high-risk credit returns and rebate forms, and if the tax services offices could better determine which rejected returns were most likely to be in error, the offices could likely reduce the time spent reviewing the rejected returns. The saved time could be used to audit additional returns or to perform more detailed audit work on the ones rejected.
16.63 In 1994 we noted that little analysis had been done on the usefulness of each automated validity check. Since then, Revenue Canada has made some refinements to the checks. Some were made in response to legislative changes, such as the introduction of the HST. Others were made to correct errors in the computer logic or to improve the cost effectiveness of specific checks. However, there has still been little analysis of the relative cost effectiveness of each check for the purpose of reducing the number of refund claims being rejected by the system. The percentage of rejected returns is approximately the same as in 1992-93.
16.64 The automated validity checks used most often are those that determine whether the amount claimed by a registrant on a return (or the cumulative amount claimed on a series of returns) exceeds a pre-determined threshold amount. Increasing these thresholds for the registrants with the lowest risk of non-compliance might significantly reduce the number of refund claims being rejected by the checks. Other ways to make the current checks more effective include:
16.67 One way that Revenue Canada could identify incorrect debit returns at an earlier date would be to develop automated validity checks for these returns. Her Majesty's Customs and Excise in the United Kingdom is currently using a limited number of such checks to select debit returns for review and audit, and is planning to extend the use of these checks in the future.
16.68 Another way for Revenue Canada to identify incorrect debit returns at an earlier date would be to expand its Computer Assisted Audit Selection (CAAS) system (see paragraphs 16.82 to 16.86) to select debit returns for review and audit at the same time as it conducts pre-payment audits of credit returns. Revenue Canada officials have informed us that the ministère du Revenu du Québec already has such a program in place.
16.69 Revenue Canada should enhance its ongoing program to review and revise the automated validity checks. It should also consider reviewing and auditing debit returns at the same time as it conducts pre-payment audits of credit returns.
Department's response: The Department agrees with the recommendation to enhance its ongoing program to review and revise the automated validity checks. As part of this process, the Department will carefully examine the Auditor General's suggestions in paragraphs 16.64 and 16.65. Additionally, the Department intends to modify the automated validity checks to include the appropriate risk issues identified in the Computer Assisted Audit Selection (CAAS) system. Furthermore, recognizing that one of the key ways in which to prevent fraudulent refund claims is to ensure that at the time of registration only legitimate businesses are registered for GST, departmental staff are working on a number of initiatives in this area.
With respect to debit returns, it should be noted that the Department already assesses all debit returns for risk using CAAS and the Department advocates the continuation of its work to further improve the CAAS system for debit returns. This notwithstanding, the Department will examine ways to conduct earlier reviews and audits of registrants identified as high risk. Senior departmental officials will be meeting with officials from other jurisdictions to share audit strategies and approaches for both credit and debit returns.
16.71 We noted that tax services offices are having trouble keeping up with the workload generated by the automated validity checks - both at the initial review stage and later, when the work is assigned to auditors. In 1997-98, according to Revenue Canada officials, 72 percent of the pre-payment audits took longer than 30 days to complete and accounted for 13 percent of the amounts paid to registrants. In a sample of 27 pre-payment audits that we selected from among the five tax services offices that we visited, we noted delays in completing 17 of them. Nine took more than 180 days to complete. We determined that the audit delays were caused primarily by the registrant in 7 of the 17 cases and in 10 cases by Revenue Canada.
16.72 Delays caused by the registrant consisted primarily of the time it took the registrant to provide requested information. Delays caused by Revenue Canada consisted primarily of the interval between rejection of the refund claim by an automated validity check and commencement of the auditor's work. Tax services officials informed us that these delays were due mainly to the workload created by the large volume of refund claims.
16.73 If a pre-payment audit were to ultimately prevent a refund cheque from being issued, the fact that it had taken many months to complete would not add to Revenue Canada's interest costs. However, of the 17 delayed audits in our sample, 15 ultimately resulted in a cheque being issued to the registrant. In total, the registrants' claims were reduced by 38 percent. While this is more than the 17 percent reduction for all 1997-98 credit returns (see Exhibit 16.6), it means that the registrants received interest for the period of the delay on 62 percent of the amount claimed.
16.75 Staffing matters are compounding this risk. Officials at three of the five tax services offices that we visited informed us that the units responsible for reviewing the refund claims were experiencing high staff turnover. At two offices, we were informed that there was a move toward using lower-level staff to review the claims. If this were to occur, decisions could be made by staff who lack sufficient knowledge of the office's registrant population to be able to identify the highest-risk credit returns and rebate forms. Again, the reviewer might perform either too little or too much work on a particular item.
16.76 Revenue Canada should intensify its review of the procedures used in the tax services offices to review and audit refund claims. This would include reviewing how the offices control their workload, and the number, levels and experience of audit staff performing the various functions.
Department's response: The Department is in agreement with this recommendation and has already completed some work in this regard, and has initiated further study and review activities as suggested by the Auditor General.
16.80 Revenue Canada cannot tell the extent to which the various audit selection methods are being used. To help determine the most effective selection methods to use for a particular sector or in a particular region, Revenue Canada needs to know how well various methods have worked in the past. However, the Department does not have detailed information on how registrants have been selected for post-payment audit. The audit selection codes currently captured in the GST mainframe and in the Audit Laptop System are not recorded consistently and are sometimes not specific enough to identify the method of selection. Furthermore, during the period covered by our audit, not all offices were using the selection codes set up in the Department's Audit Information Management System (AIMS) to complement the Audit Laptop System. We understand that the tax services offices have since begun using the audit selection codes in AIMS.
16.81 One tax services office that we visited had routinely used narrative to complement the audit selection codes, so we were generally able to determine the selection method. In the other four offices, we were unable to determine the selection method for over 40 percent of our sample items.
16.82 The effectiveness of one of Revenue Canada's audit selection methods is hampered by some data deficiencies and a need for ongoing training. When Revenue Canada's Computer Assisted Audit Selection (CAAS) system is combined with the Audit Report Generator On-line (ARGO), the system allows auditors to focus on specific industry sectors, geographic locations and audit issues. The system contains GST and income tax data, some provincial sales tax data, and data on importers. Auditors can use this information to identify registrants who do not fit the usual profile of the industry, or whose GST and income tax data appear to be inconsistent.
16.83 While we consider CAAS/ARGO to have potential for selecting registrants with a high risk of non-compliance, Revenue Canada auditors had used it to select only 7 percent of the post-payment audits in our sample. Revenue Canada officials at headquarters and some auditors tasked with using the tool cited certain deficiencies in it. For example, information on gross revenues and other pertinent business data for each registrant are not up-to-date. Also, there are no data on export sales. (As these sales are not taxed, they can significantly distort ratios used to highlight potential non-compliance.) Revenue Canada officials identified these and other deficiencies as early as 1996 but, despite their efforts, have not yet been able to resolve them. At present, they are considering adding the reporting of export sales to the GST returns.
16.84 The effectiveness of CAAS/ARGO is also hampered by missing or incorrect standard industry classification codes used to group registrants by industry sector. The use of CAAS/ARGO to select registrants in a particular industry sector would not capture registrants with missing or incorrect codes. Despite recent improvements, in May 1999 over 10 percent of the registrants in Revenue Canada's database still did not have standard industry classification codes. In our post-payment audit sample, 9 percent did not have codes and, in our view, more than 10 percent of the other registrants had codes that were incorrect.
16.85 Further, based on interviews conducted in the five tax services offices that we visited, it appears that the effectiveness of CAAS/ARGO is hampered by the lack of training received by some employees tasked with using the system. Revenue Canada officials informed us that high staff turnover may be contributing to this.
16.86 Revenue Canada should continue to improve its Computer Assisted Audit Selection and Audit Report Generator On-line systems. This would include dealing with the identified deficiencies, ensuring that all necessary data (such as standard industry classification codes) are available, and providing additional training to staff who are expected to use the system.
Department's response: Although Revenue Canada's Computer Assisted Audit Selection and Audit Report Generator On-line systems for GST are relatively new, we have been seeking and will continue to seek ways to improve them. This ongoing process includes incorporating new data from various sources and increasing the completeness and accuracy of currently available data.
We recognize the importance of data on export sales and will continue our efforts to obtain such data. We also recognize the importance of standard industry classification codes and are working with officials from Statistics Canada to address this.
Although we have been providing ongoing training on the use of these systems, we recognize the need for continuous training and will increase our efforts in this area.
16.87 Random audits are needed. One way to identify the sectors with the highest rates of non-compliance would be to periodically select a random sample of registrants in which all sectors have an equal chance of being represented. This sample could also be used to measure changes in levels of non-compliance over time. After many delays, Revenue Canada has a project under way to select a random sample from one industry sector. Samples were sent to the tax services offices in February 1999, and the first audit results are due in December 1999.
16.89 While good selection techniques will help to ensure the selection of registrants with the highest risk of non-compliance, they must be followed by good planning to ensure the efficient and effective use of audit resources.
16.90 Documentation of planning is limited. To assist in the planning process, Revenue Canada's Audit Laptop System contains suggested planning procedures. The procedures are not designed to deal with many industry sectors or complex issues. Given these limitations, Revenue Canada requires its large file auditors to augment the planning procedures in the Audit Laptop System with written audit plans. However, of the eight large file audits that we believe needed an audit plan, only four had a written plan.
16.91 Revenue Canada officials have informed us that the Department also requires its auditors of small and medium enterprises to augment the planning procedures in the Audit Laptop System with written audit plans. However, most of the audit files in our sample did not contain these plans.
16.92 Revenue Canada's Quality Assurance Division also concluded that more documentation of planning would be beneficial. It also noted that better training - both through courses and on the job - would help to improve the planning process.
16.94 We noted areas where all tax services offices were performing well. For example, on most audits:
16.96 We also noted that some tax services offices have enhanced the standard audit support tools. For example:
16.98 Many audits took many months to complete. A national report of GST audits in progress on 22 March 1999 showed that 32 percent of the audits had been under way for more than 180 days. These accounted for 70 percent of the cumulative GST audit hours. This means that audits begun in the past six months accounted for only 30 percent of the audit hours.
16.99 In our own audit sample, 67 of the 141 post-payment audits had taken more than 180 days, and 14 had lasted more than two years. The average duration of the 67 audits was 495 days - about 16 months.
16.100 The extensive length of these audits was the result of delays caused by either Revenue Canada or the registrant or both. Delays by Revenue Canada were due primarily to conflicting priorities of the auditors involved. Delays caused by registrants were primarily due to a failure to respond to requests for an initial meeting or to provide requested documentation on a timely basis.
16.101 Some of the delays caused by Revenue Canada could have been avoided by better audit planning and increased monitoring of the work in progress. In one tax services office, we were told that audits had not been monitored because team leaders were being rotated every few months. Detailed monitoring was performed only after permanent team leaders had been assigned. Another office told us that for various reasons there had been no detailed monitoring, so many audit files had been outstanding for more than 180 days. We were also informed that to make the workload more manageable, a number of old files had been reviewed and a decision made to discontinue work on those expected to generate low or no assessments.
16.102 Registrant delays can add significantly to the cost of the audit, and are often beyond Revenue Canada's control. It can penalize registrants who do not provide requested data promptly. While we understand that the Department does impose these penalties, and some files in our audit sample indicated that it had threatened to impose them, it had not done so in any of our sample items.
16.103 Some GST audits were not efficient. Audits that take many hours but yield only small assessments are not necessarily poor audits. Very large, complex registrants may take hundreds of hours to audit, and may turn out to be in compliance with the Excise Tax Act. On the other hand, a pattern of such audits could be a symptom that they were poorly done.
16.104 To determine whether audits that take many hours but yield only small assessments are good or poor audits, we reviewed the 25 audits in our sample that had taken more than 100 hours and had yielded a revenue recovery of less than $100 per hour. We found that in over 75 percent of the cases, the time spent appeared to be excessive in relation to the size and complexity of the registrant and the scope of the audit work. This not only represents an inefficient use of resources but also adds to the burden on the registrants being audited.
16.105 Revenue Canada officials have informed us that post-payment audits that had taken more than 100 hours and yielded a revenue recovery of less than $100 per hour constituted only 5 percent of the audits completed in 1997-98. Nevertheless, we believe that excessive audit time is a matter that Revenue Canada needs to pursue. We found in sample items other than the 25 just noted that audit time had been excessive in relation to the size and complexity of the registrant. In addition, Revenue Canada's Quality Assurance Division also reported that it found many audits conducted in 1996 and 1997 that had taken an excessive length of time in relation to the audit scope, and that many audits went on far too long. The report concluded that better audit planning and greater management participation would reduce the time spent on audits.
16.106 One way to control the number of hours spent on an audit would be to budget at the planning stage for a certain number of hours. Ideally, budgets would be tailored to each set of circumstances to provide a better basis for managing and controlling audit performance. This is not being done. Instead, tax services offices often use national averages, developed by Revenue Canada as part of its annual budgetary exercise. These national averages fail to account for the unique characteristics of each audit. One tax services office that we visited now requires staff to sign "performance communications", stating that, among other things, they will normally complete their audits in less time than the national averages. While we understand that these documents were implemented to provide the office's management with greater control over the number of hours spent on each audit - a commendable goal - we are concerned that this approach could lead staff to unduly limit the scope of their audits.
16.107 Another way to control the time spent on an audit would be to increase active participation by team leaders. This would also help to ensure that audits do not drag on too long and are performed as planned. Seventy percent of our post-payment audit files showed no evidence of such participation.
16.108 In a recent report, Revenue Canada's Quality Assurance Division also reported that team leader involvement in audits is very limited. The report cites various contributing factors, such as group size and the heavy demands of other responsibilities. The Department is considering reducing the size of groups and redefining the roles and responsibilities of team leaders.
16.109 Revenue Canada should enhance its efforts to reduce the duration of its audits and the hours required to complete them. These efforts should include enabling team leaders to be more actively involved in their audits, and providing greater support to auditors in planning and performing their work. Efforts should also include making more use of penalties to deter delays by registrants.
Department's response: As described below, Revenue Canada has already taken action to improve overall program delivery in the areas identified in this recommendation.
16.111 We noted that the audit period for 76 of our 141 post-payment audits exceeded the period called for in the policy, in a large number of cases without justification. The Department has issued a directive to its auditors of small and medium enterprises to reinforce the policy.
16.112 Incorporating a compliance review of income tax into GST audits is not a common practice. The strategy for audits of small and medium enterprises calls for the integration of income tax and GST audits to the extent possible. In cases where only one tax is audited, a "compliance review" of the other tax is generally required. This work entails examining specific indicators of the second tax to determine the level of compliance. Compliance reviews offer increased coverage, which should lead to greater deterrence and increased revenue recovery.
16.113 Using the Department's own criteria, we concluded that an income tax compliance review checklist had been warranted in 77 of the post-payment audit files we reviewed. Fifty-two of these files contained no evidence that these checklists had been completed. Departmental officials have informed us that they recently made changes to the Audit Information Management System to provide management with the information needed to ensure that compliance reviews are done when necessary.
16.115 Senior managers in all five tax services offices that we visited stated that they do not have the ideal complement of audit staff and equipment. Two offices have a surplus of auditors at the AU-1 level (who typically audit smaller corporations) and a shortage of "generalist auditors" at the PM-2 level (who typically do both GST and income tax audits of small unincorporated businesses). Three offices have a shortage of senior auditors (AU-3 level) who audit large corporations.
16.116 Senior managers and auditors in the five tax services offices informed us that the shortage of senior auditors is made worse by high turnover, due to staff transfers within the Department as well as staff departures. Auditors often do not perceive that they have a promising career path in the GST stream, and prefer to switch to the income tax stream. This perception has its roots in differences in the job descriptions of auditors, predating the consolidation of the former departments of Taxation (which administered the income tax) and Customs and Excise (which administered the GST). For example, the job descriptions call for GST auditors at the AU-2 level to perform audits of the larger small and medium enterprises, while the income tax audits of these enterprises are done by more senior auditors, at the AU-3 level.
16.117 As for training, the generalist auditors at the PM-2 level in the five tax services offices claimed that their training was excellent. Other GST auditors informed us that they lacked training in specialized areas. In particular, some auditors told us that they felt ill-prepared to audit complex and specialized organizations, such as public sector bodies (for example, municipalities). In some offices, technical advisors have augmented the official training program. When time permits, these experts in legislative interpretation give presentations on technical topics of local interest.
16.118 Departmental officials have informed us that they have commenced a number of initiatives to attract and retain knowledgeable staff.
16.120 Problems in the consistency of the data. In previous reports, we have commented on problems of data consistency. In this audit, we again encountered inconsistent data. Until Revenue Canada can assure itself that the data are consistent, it will not know which variances in audit results - among tax services offices and year to year - are attributable to errors and inconsistencies and which are attributable to "real" performance factors that should be followed up. These data problems will limit Revenue Canada's ability to improve its pre-payment and post-payment audit programs.
16.121 We noted the following problems:
16.123 Some aggregated data are not readily available. In order to make good audit selections, management needs information such as the number of registrants in each industry sector or in various ranges of sales revenue, prior audits of each registrant, audit hours and amounts assessed by industry sector, and adjustments to assessments made by the Appeals Branch. Although this information exists in various databases and reports, it is sometimes difficult to access and analyze. In these cases, management must rely on special ad hoc reports and registrant tape files that are updated and distributed every six months.
16.124 Revenue Canada should continue to improve the timely availability of data needed to manage the GST audit programs effectively. To that end, the Department should resolve inconsistencies in the data-capturing methods, ensure that missing data are obtained, and ensure that data can be aggregated and analyzed.
Department's response: The Department agrees with this recommendation; some work has been completed and other work is under way to address these issues.
16.126 While the ministère du Revenu du Québec uses all of Revenue Canada's systems and policies, it sets its own audit strategy. Revenue Canada officials informed us that, while they maintain open communication with the ministère on operational and administrative matters, they do not verify the completeness or accuracy of the data received from the ministère or perform detailed analyses of its audit results. Revenue Canada is close to finalizing an updated agreement with the ministère that provides for a more structured sharing of audit strategies, work plans and audit results.
16.127 As for other organizations, Revenue Canada officials informed us that they have ongoing dialogue and attend international gatherings with similar organizations in other countries with GST (value-added tax); however, they have not formally compared their compliance efforts with those countries.
16.129 The primary reason given for the lost and misplaced hard copy files was the absence of an electronic control log to track who in Revenue Canada has the GST audit files when they are transferred among various units - for example, Appeals Branch, Special Investigations, headquarters or other tax services offices. The primary explanation for the unusable diskettes was that they malfunctioned during the audit and the audit working papers contained therein could not be retrieved. Departmental officials have informed us that they are currently addressing these matters.
16.132 The timeliness with which documents are processed at the Summerside Tax Centre has improved since the Centre opened in 1993. Our examination of the timeliness of post-payment audits revealed that many audits have extended over many months and have taken many hours to complete. Pre-payment audits by their nature do not last as long as post-payment audits, but these, too, are prone to delays. Both Revenue Canada and the registrants themselves have contributed to delays in the completion of audits. We believe that further improvements are possible in the speed with which documents are processed and that additional steps can be taken to reduce the delays experienced in audits. This would result in better service to registrants and, in some situations, reduced interest charges on refunds paid after the 21-day grace period. Document processing, pre-payment audit and post-payment audit activities are linked, and the persistence of problems with timeliness suggests that performance in this network of activities has not yet been optimized.
16.133 Revenue Canada has made progress in establishing performance standards and expectations for its returns processing and audit activities, and is undertaking a major overhaul of its management information systems. At present, however, standards for returns processing tend to place insufficient emphasis on quality, accuracy, timeliness and unit cost, and the use of performance information is hampered by problems with the availability and reliability of data. In the Department's audit activities, the use of performance information is hampered by problems with the availability and consistency of data.
16.134 The Department faces a significant challenge in maintaining a sufficient complement of GST staff with appropriate knowledge, skill and experience. Staff transfers within the Department and staff departures contribute to shortages of staff in particular activities. Training for staff who must perform new functions has not always kept up with the need.
16.135 Prompted by the decision to decentralize the processing of GST returns, formerly performed only at the Summerside Tax Centre, Revenue Canada recently simplified its procedures. This follows efficiency gains made over the years in Summerside. Revenue Canada has also improved the automated validity checks it uses to select, for possible pre-payment audit, credit returns with the highest risk of non-compliance. However, we believe that further improvements are possible to make the checks more effective, and that the Department should consider reviewing and auditing debit returns at the same time as it conducts pre-payment audits of credit returns. Also, the performance of both pre-payment and post-payment audits could be improved by the consistent application of existing or strengthened procedures.
The audit took place at Revenue Canada headquarters in Ottawa and at various tax centres and tax services offices throughout the country.
Terence Brown
Thérèse Desjardins
Nisha Goyal
Patricia Smith
For information, please contact Scott Milne.