Audit Manual for Selection Systems_Eng.doc

Technical Assistance to Support Tax Administration Reforms (TASTAR)

MANUAL GUIDELINES

AUDIT PLANNING AND RISK BASED SELECTION SYSTEMS

NORMAN FOX, TAX ADVISOR

Table of Contents:

1—Backround and overview of total system

2—Audit strategy and planning

3—Homogeneous groups of taxpayers

Armenia groups

4—Risk based scoring systems 5—Supplemental systems

--Special projects

--Third party data

--Informant’s advice

--Related returns

--Discretionary work

--Mandatory returns

--Examples

6—Use of selection programs at Tax Inspectorates

7—Audit results and capture of data

8—Analysis of results

--Selection criteria

--Audit Department management Chapter 1: Background and overview of total system

The goal of the State Revenue Committee (SRC) is to collect the proper amount of tax from all taxpayers. All modern tax systems attempt to achieve this goal through the use of concepts that promote high levels of voluntary compliance; where taxpayers self assess their tax.

The strategies employed by the SRC, including the Audit function, should be geared toward achieving this high level of voluntary compliance. This will provide the maximum tax receipts for the least cost when the programs are implemented and working effectively.

The Audit function supports the SRC in achieving the goal of voluntary compliance by enforcing the laws of taxation. Taxpayers who file timely and correctly should be left alone and not subjected to the burden of tax audits. But, taxpayers that do not comply with the tax laws will be subject to investigation and tax return audit. If they have violated the tax laws, they will be made to pay the additional taxes owed and also could be liable for penalties and interest.

This approach should lead to higher levels of voluntary compliance and more taxpayers who will not face the burden of tax audits. However, not all taxpayers will voluntarily comply with filing accurate returns. The role of the Audit function is critical to the overall success of the SRC to achieve the goal of voluntary compliance. If taxpayers are made aware that the Audit programs are designed to detect non-compliant returns, they may be more inclined to file accurate returns that truly reflect their income and taxes.

The Audit programs are designed to make the maximum impact on all taxpayers and collect the maximum revenue that is owed to the government. It will do this by carefully selecting only a portion of all the tax returns that are filed. This is the basis of a “risk based selection program”.

The overall program will be achieved by a combination of Audit planning and a number of selection techniques to be described in the following chapters.

Chapter 2: Audit Strategy and Planning

Audit management should plan the number and type of audits that will achieve the collection of most revenue and the optimum effect on future compliance. This effort is called the “Audit Plan”. It is formulated each year, to cover a period of one year; typically, from January through December.

The annual plan should be completed prior to the beginning of the calendar year. The plan is formulated in the headquarters office, with the guidance from local tax inspectorates. The plan should be rigorous so that it encourages the offices to perform in a highly productive manner. But, it also must be realistic, recognizing the staff capabilities and environment that will be encountered.

The plan must be “balanced” between revenue collection and coverage in all classes of taxpayers.

To organize the entire population of all taxpayers in Armenia into smaller groups that can be used for planning purposes; we create “classes” or groups of returns that are similar in character. These homogeneous groups will reflect the type of return filed, the size of the business, and other unique features that differentiate them from other groups. We design selection scoring formulas for each separate group, since the data is different and the compliance issues can be different.

These groups allow the audit planners to better understand the nature of the audit, the time involved, and the potential outcomes. It further allows for a more exacting comparison of results when we compare similar audits.

In Armenia, the groups are broken down by returns; such as VAT returns, Profit Tax, and Income Tax. They are then divided by size, because the filing requirements are different for thresholds of turnover.

The Audit planners in Headquarters first consider what an ideal national plan would look like. This “optimal” plan indicates the coverage, or number of audits for each of the groups. The LTU (large taxpayer unit) will be given resources to provide substantial coverage for all of their taxpayers. Larger businesses will be given higher coverage than smaller businesses. The smallest taxpayers may produce minimal tax recoveries but they should have some coverage to achieve a balanced plan.

The annual audit plan will show the coverage percentage and the number of audits to be completed in one year, group by group. This will be broken down into each tax inspectorate, so that local management can see what is expected from them for the following year.

This plan will require that knowledge of the time expended for each type of audit will be used in calculating how many returns, group by group can be performed during a year. When we allocate the staff years to different groups of returns, the time for each audit will dictate how many audits can be expected.

The planning process uses the audit of one year of one type of return as the basis for planning. If the auditor completes just one year for one type of return; the tally will be for one audit of one year for that group. If the auditor completes two years for a return type, the tally will be for two audits. If the auditor completes audits on both VAT and Profit Tax returns during a comprehensive audit, the tally will reflect that audits are completed on returns on all the groups that are audited.

To assure that accurate time data is collected it is necessary to collect the time by group and by each year audited. This data will be used in future planning and for analysis of results. A later chapter will describe in more detail the use of audit results data.

When headquarters management breaks down the national plan into separate plans for each TI (tax inspectorate) it should consult with local management about the goals. Local knowledge will help to refine the plan, based on the skill level of the staff, the practicality of the plan, and any local conditions to consider in making a final plan for the office.

The number of filed returns in each office will vary, as well as the type of returns. Also, the number of auditors in the office will dictate the number of audits that can be achieved during the following year. These statistics will be available to planners to ensure that realistic numbers are used in plans.

The local plan will be a goal to accomplish. It is not expected that every office will meet every goal for every group. Some offices may over accomplish the goal for some groups. Some offices may fall short of goals for some groups. Audit management and planners will review the accomplishments at periodic intervals to see how the plan is progressing. Adjustments can be made if justified.

Chapter 3: Homogeneous groups for Armenia It is beneficial for numerous reasons to break the total population of taxpayers into smaller homogeneous groups. This is done for planning purposes, tracking audit results, and to develop more powerful scoring systems to select the riskiest returns within each group.

In Armenia, there are separate returns for Profit Tax, Income Tax, and VAT tax. For Profit tax, there are simplified returns and regular returns. For VAT, some taxpayers file monthly and some file quarterly. These separate returns are processed and the data elements are entered into the computer data base. The data base fields are used as potential “variables” in the scoring formulas. Since the fields vary from one group to another; it is essential that the scoring formula be designed specifically for the group.

All the returns filed will be scored; the higher the score, the greater the risk and the higher the potential audit result. Returns within a group are compared to each other within that group to see which have the highest score. These are the returns, and taxpayers, that are to be selected for audit.

The bigger taxpayers with more turnover, more taxes to be paid, and more complexities, deserve to be given more coverage, since there is greater risk. A basic concept in tax planning is to design the audit plan to give higher coverage to the returns in the groups with “bigger” taxpayers. The taxpayers in the LTU will receive the highest coverage. Small taxpayers with much less tax to pay and thus lower risk, will be given lower coverage.

Tax return groups for Armenia are as follows:

Physical Persons:

--Presumptive tax

--Non-business

--Sole Proprietors

Legal Entities:

--Presumptive tax

--Less than 58.3 million turnover on Profit tax declaration (no VAT requirement). File simplified return.

-- Greater than 58.3 million turnover (VAT requirement). This group also includes those taxpayers under 58.3 million turnover who do file VAT voluntarily and also file the regular Profit tax return.

VAT:

Quarterly filers (under 100 million turnover)

Monthly filers (over 100 million turnover)

Chapter 4: Risk Based Scoring system

There will be a number of separate techniques available and used to select returns for audit. This chapter will focus on the general scoring system. Other selection systems are called supplemental systems. They also are designed to be risk based. However, they will use separate criteria, or parameters, or methods to target taxpayers or returns. They may be selected for audit based on the supplemental system targeting, regardless of the score from the general scoring system.

The scoring system will score every return that is filed just after filing. The return will first be slotted into the group that it belongs to and the computer will use the formula for that particular group to score the return. All the elements used in the formula will be in the computer data base. It does not depend on any outside data that is not part of the SRC computer data base.

Scores will be from 0 up to 999. The higher the score the greater the risk. The scores are only used to compare returns within the same group. Scores for returns in other groups will also be scored from 0 to 999, but they depend on a different formula and cannot be compared. There is no reason to compare scores between groups. The plan will indicate the coverage and number of returns to be audited, group by group. Audit management should assure that additional returns placed into the audit stream come from the highest scores, or from a supplemental system.

For a more detailed description of how the formulas are compiled see the separate documents: a) How to Build a Selection Formula and b) Three Step Process to Build a Formula.

The actual criteria, and formulas and the algorithms used are confidential. They are to be revealed on a need to know basis only. The list of persons privy to the formulas should be as small as possible. Anyone who reveals any part of any formula to anyone not entitled to know the information will be prosecuted.

The score on the return will be stored in the data base. These can be used to array all returns within the group in descending score order. When more returns in the group need to be audited, returns from the top scores can be delivered to the local office audit groups.

Local office management will have the responsibility to put returns into the Audit stream. They will have the scores from the general scoring system to provide the returns with the greatest risk. They may also get some returns from other supplemental systems. They should always prioritize the workload entering the Audit stream; unless certain returns are deemed mandatory and must be expedited.

Headquarters will provide guidance to the local office on how to conduct the selection of returns. They may from time to time dictate that certain returns are given high priority and should be worked before other returns are begun.

Local management will use the annual plan for their office to guide them on which returns are required to meet the plan. This plan will spell out the goal for each separate group. Mandatory or “must work” will always take precedence, regardless of the plan. This is one reason why the goal, set for each group may not be reached.

All returns that are completed, regardless of the reason for selection, are tallied for purposes of the plan completion. When local management has discretion to select more work; they need to look at the plan, the returns completed to date and select more workload that will satisfy the plan.

Chapter 5: Supplemental systems In the previous chapter the general scoring system was described. The general scoring system makes all returns potentially available for audit selection through the score that will be posted to the computer data base. This general scoring system uses many criteria, or “variables” in computing the total score for the return filed. It is designed to give returns with higher risk a higher score.

This chapter will describe supplemental systems which are used to supplement the general scoring system. While the general scoring system is effective in looking at the entire population of returns and prioritizing them; it may not identify every return, or taxpayer that might be a non-complier.

Special, ad hoc, programs will be used to target unique groups of taxpayers that need to be considered for audit. Sometimes the group is a “pocket of non-compliance.”

We can also identify an individual person or business that should be audited, regardless of the score on the returns. This is done when specific information is uncovered which details the nature of the non compliance behavior.

Special Projects:

Special projects are designed to identify a group of taxpayers or returns that appear to be non compliant by applying a set of criteria to internal or external data. These projects will often use the SRC data base and may also use the data from external sources to match against the return information.

Special projects are typically used on an “ad hoc” basis. That is, not as a constant ongoing program, but rather on a one time basis. They may respond to a problem that has been identified, such as a tax issue which is being abused. As taxpayers discover and use new methods to avoid taxes, the SRC may use special projects to attempt to identify such taxpayers.

When the SRC has indications of tax abuse from a particular industry or the use of a particular scheme to avoid taxes, they could use specific criteria to identify returns that appear to be in violation. These projects usually work best when the computer data base can be used to quickly select returns that meet the criteria. Returns identified should be screened to assure that they appear to be good audit targets. Sometimes it may be evident from the return that the violation appears to have committed. Other times it may be difficult to really know whether a violation has been committed, or merely indicates that some level of potential exits. It is recommended that a pilot of a small number of returns be audited to determine the audit potential of the returns using the criteria. If this small group shows good audit results, then additional returns meeting the criteria can be put into the audit stream.

Data from external sources may be used to compare to the data in the SRC data base. This usually works best when the external source has computer files that can be compared to the SRC data files. This may mean some work on the part of the I T experts to make the files compatible. It also may mean that some identification code must be similar to match the files and identify the correct person, or business.

An example of such a special project may best illustrate how this works. When an individual or a business purchases a new luxury vehicle for cash it may raise the issue of whether this person or business has reported income commensurate with the ability to make such a purchase. The SRC must first determine whether data files on such purchases are available. If this data is available, then determine how the data is collected and whether it can be compared readily to the SRC data base. The matching will look for taxpayers that have reported little or no income for a number of years, and yet accumulated a large amount of cash for the purchase.

Another example is the use of business associations that maintain files on members in a particular profession, craft, or industry. Compare the external data to the SRC files and look for non-filers or very low income amounts for work that normally would be lucrative.

A special project could also be designed to use just the SRC data base. Prepare specific criteria that look for particular industry groups that have some unusual outcomes. This could be balance sheet items or changes, or other entries that appear out of normal. It could target a geographic location, or another unique feature that is available in the data base.

Third Party Data:

Previously, it was noted that some special projects may include the use of external data. This is often referred to as “third party data”. Any data that comes from sources outside of the SRC would be considered third party data. This would also include any Customs data, even though technically Customs is part of the SRC. Since Customs data is maintained in a separate data base it will be necessary to arrange for the transfer of this information.

Typical third party data includes: Data from other government agencies, quasi-government organizations, and business associations, business organizations that maintain files about industry or professions, and directories that list individuals or business organizations.

Using third party data to compare to data within the data base of the SRC may reveal taxpayers that have not filed but appear to be liable for taxes. Also, it may reveal taxpayers that claim a level of income that is not compatible with the information from the third party.

Using third party data can take the form of computer files or paper files. Obviously, computer files usually will be more efficient. However, the computer files may not be usable in their current format. An I T expert will have to examine the character of these external files and determine if they can be converted to a format that will be usable. Also, the external files must contain some elements that will make it possible to relate to the SRC data and find the right taxpayer. The easiest way would be to use the TIN, but the external file may not have this number. Other relatable data base elements must be found to identify the correct taxpayer. For example, there are sophisticated software prgrams that can compare names and addresses to assure that the same person or business is identified from the two separate files.

It is also possible to use paper files when this is how the records are kept. This will be more labor intensive and care must be taken to not expend many staff hours unless it can be shown that there will be productive results from the effort.

Typical examples of using third party data include: Customs data to compare import and export amounts against tax return data; Real estate records to compare transactions from sales against tax return data; Motor vehicle records to spot cash purchases of expensive cars; Business records that show transactions with customers and data from tax officials from foreign countries.

Informant’s Advice:

Persons who send information to the SRC about tax violations are referred to as “informants”. These people may be privy to private information that would not become available to the SRC during normal operations, even during an audit. Informants may work for a business currently or in the past and know the inner schemes used by the owner to avoid taxes. Or, they may know a person who has a lifestyle that is not compatible with the tax reporting.

Informants may send this information to the SRC for various personal reasons, including: wanting to right an injustice and protect the revenue to the country, or wanting some sort of revenge against the business or acquaintance. The SRC should not be so concerned about the motivation, since the informant does not have to reveal his intent. But, the SRC does need to understand if the information provided is credible and usable and will lead to additional tax and penalties.

Many countries have a policy of giving a monetary reward to the informant if it proves out that the information provided does indeed lead to additional taxes that are collected. The amount is usually a percentage of the tax; and can fall into a range, depending on how much specificity is provided.

Informants usually give information about one single person or business. However, sometimes an informant may be aware of a scheme used by an entire business segment.

Related returns:

During an audit, the auditor may come across transactions that involve another taxpayer. The information about the “related” taxpayer may be useful to see if this taxpayer has properly reported the transaction on his return. Examples include: sales of fixed assets for a large amount, cash transactions, and transactions with a sister entity that may be distorted for tax purposes.

Some countries have set up a formal procedure to relay this information to the audit personnel that determine if an examination may be productive. A written document called an “information notice” is used to describe the transaction and the possible tax consequence. This notice is associated to the returns of the related party and a review is made to determine if an audit would be productive.

Another use of the term “related return” is the expansion of the audit to additional years based on the result of the assigned return. It is recommended that an audit start with the latest return that was filed. If this audit reveals tax errors, the auditor can expand the audit to prior years that are still open; that is, not previously examined. Each year that is examined will count as an additional audit for purposes of accomplishment of the annual audit plan.

Discretionary work:

Local offices sometimes want some flexibility to audit returns or taxpayers that they believe will be productive. These audits are not part of any scoring protocol or national special project.

Headquarters may decide to allow some small percentage of returns to be at the discretion of local audit management. This can be productive if local management has some local knowledge that leads to productive audits. It also opens the door to potential corruption when the local office can pick who they want to audit without controls.

Headquarters would have to review the use of discretionary work to be sure it is not abused.

Mandatory returns:

Many countries have certain categories of returns that must be audited. This is usually done by legislative mandate. This may be justified based on past history or the deep concern for assuring that certain situations are reviewed 100 per cent.

The SRC should consider the categories of mandated work and determine if the requirement is still logical and productive. Sometimes the old reasons for such audits may no longer be rational. Or, it may be that the returns can be reviewed in a more sophisticated manner to determine which should be audited and which can be bypassed.

Chapter 6: Use of Selection Programs at the Tax Inspectorate:

We have described a number of selection programs that can be used to put new returns into the audit stream. So it may be natural to wonder how all these programs work together to provide the workload for the local tax office.

The general scoring system that scores all the returns that are filed should be looked at as the basic selection program. That is, a default method when no other special project returns or other ad hoc returns are taking priority.

At the local office, there will be a responsible person to see that the auditors are kept busy with work and that there are always new returns to be assigned to the auditors as they need additional assignments. The scoring program returns are always available. If no other returns from supplemental systems are prescribed for the office, there is no problem to reach for the next higher scores from inventory and start these cases. The manager of assigning workload needs to pay attention to the annual audit plan for his office and select work that will fulfill the goals of the plan.

Certain cases will be mandated and expected to be expedited. These may be cases mandated by legislation, or by the headquarters demanding that certain special projects be carried out by time deadlines. These type cases will be selected on a priority basis. When completed, these cases will be tallied toward meeting the goals of the annual audit plan.

If the local office is given discretion about using a supplemental system or returns identified by score, the local manager must decide on the returns that appear to have the most potential for audit results. He may also want to keep an eye on the plan goals and that may also influence his decision.

Usually, when the headquarters is mandating that some returns for a special project be worked in the local offices, they will dictate the number of returns and may also provide the actual cases to be worked.

The local office will have more discretion over local selection techniques, such as informant’s advice, or related returns. Also, the local office may select a limited number of returns as discretionary work. If the local office wants to create and work on their own special project, they must request permission from headquarters to do this. This would normally require a detailed description of intentions and criteria to be used and the number of returns to pilot test the idea. Headquarters would weigh the proposal in light of knowledge of other efforts that may have been undertaken and determine if the local project has merit.

Many good special projects start by the local office coming up with the original idea. But these need to be controlled by headquarters so that the workload plan is managed. Local projects that prove to be successful may be adopted for national use.

All cases that enter the audit stream will come from a specific source. The source must be captured by a code in the computer data base for audit results. So every type of selection system, and every unique special project, will have a unique “source code” input with the start of the audit.

In a later chapter, we describe how this data can be used to evaluate audit results.

Chapter 7: Audit Results and the Capture of Data:

The Audit Department will want to have a Management Information System (MIS) to track the assignment of work and results of audits. The MIS will serve many purposes including: a) improving the planning programs, b) refining the selection systems to make them more powerful, c) tracking inventory and work in process to assist in meeting goals in the annual plan, d) keeping results on time to complete audits for each class, e) making comparative analysis of offices, sources of work, etc. and f) reporting on year-end audit results.

The Audit MIS computer system may be set up as a sub file of the SRC main computer data base. It will serve to track inventory by the status of cases. When cases are completed, the MIS will capture the results of the audit, return by return. It will also capture the time spent on cases, broken down for each separate return. In this way we can make analysis of the revenue (yield) and the time it took to generate the yield. The yield per hour is an important factor in analyzing the comparative productivity of different sources of work and variations among classes.

In order to track the cases and to capture all the elements of closed cases, systems will need to be in place. These may be a combination of paper records and computer inputs.

When the auditor has completed a case, there are a number of elements to capture, including:

-Date of assignment

-Date of closing

-Amount of additional tax and penalties

-Agreed or not agreed

-The number of hours spent each return

-The name and grade of the auditor

Periodically, Audit management will want to print out certain reports to evaluate the various programs and make decisions on whether to make any adjustments to audit planning.

Chapter 8: Analysis of Results:

A robust Audit MIS can lend itself to many forms and styles of analysis for management. Periodic reports of critical elements should assist management to be knowledgeable of audit programs and their level of success. Special, ad hoc reports can be gleaned from the data as well, when there is a need to know a particular outcome.

Some of the ways to analyze the data is as follows:

-Yield per hour, in total and for different classes, different sources of work, for the local offices, etc.

-Time per return, in total and for different classes, different sources of work, for each local office, etc.

-Yield per return, for various breakouts.

-Lapse time to complete cases, for various breakouts.

-The per cent of cases agreed or not agreed, by breakouts.

-The per cent of returns that have no adjustments, by breakouts.

-Frequency distributions for any of the above parameters.

It is also possible to use Audit MIS data and compare results to the data in the SRC main computer data base for further analysis. This is particularly important for the analysis of criteria for selection programs.

At some future date when more robust data is available, it will be possible to make correlations between criteria and the audit result outcomes. This will enhance the capability to determine the most powerful criteria to use in the scoring formulas. It can also be used to improve the criteria for special projects.

Technical Assistance to Support Tax Administration Reforms (TASTAR)

PAGE

Technical Assistance to Support Tax Administration Reforms (TASTAR)