STATE OF WEST VIRGINIA

FULL PERFORMANCE EVALUATION OF THE

Department of Tax and Revenue

Tax Division

Random Selection of Audits is Least Effective in Recovery of Revenue and Needlessly Hassles Honest Taxpayers

Issue Area 1: The Audit Selection Process Can be Improved to Maximize Recovery of Revenue, Enhance Voluntary Compliance, and Reduce the Audit Burden on Honest Taxpayers Through Improved Identification of Noncompliant Taxpayers.

The Tax Department's Auditing Division has as its mission statement the responsibility to:

...conduct systematic field audits of taxpayers' returns and records to determine that taxes are being remitted correctly, all in a manner which encourages voluntary compliance and maximizes revenue. [emphasis added]

The Division also states that identifying noncompliant taxpayers is another goal of its audit plans. However, the outcome of the Division's audits suggest that revenue recovery is not maximized, encouraging voluntary compliance and identifying noncompliant taxpayers can be significantly improved. Many of the Division's audits result in no assessments, meaning no unpaid taxes were identified because taxpayers were found to be in compliance. Of the 8,345 audits conducted on 27 different taxes in 1998, 79% of the audits had no assessments (See Table 1). The direct cost of audits without assessments was $415,777. The top six taxes that are audited consist of 92% of the audits. Examining some of these on an individual basis, the percentage of audits without assessments is 90% or higher for some taxes. The results for audits conducted in 1996 and 1997 were similar, nearly 77% of audits had no assessments in both years. Auditing many taxpayers that are in compliance is not indicative of identifying noncompliant taxpayers which hinders achieving the dual goals of encouraging voluntary compliance and maximizing revenue recovery.

Table 1

Percent & Direct Cost of 1998 Audits without Assessments

Table 2 provides data on the results of the 1998 audits for major tax audits and the other audits. The results show that some tax audits generate significant revenue and have COLLECTION/COST ratios that are higher than other tax audits. These ratios include only the direct costs of the audits. Indirect costs would lower these ratios.

Table 2

Results for 1998 Audits

Table 3 provides a summary of the results of the audits conducted in each of the last three fiscal years. This table also calculates the COLLECTION/COSTS ratios using the combined total of direct and indirect costs. The direct cost of conducting all audits in 1998 was $1,015,470, and the indirect cost was $1,447,995, for a total cost of $2,463,466. This results in an ASSESSMENT/ TOTAL COST ratio of 9.5 to 1, and a COLLECTION/ TOTAL COST ratio of 2.9 to 1, which indicates that for every dollar spent on auditing, $9.50 was assessed and $2.90 was collected.

Table 3

Cost/Benefit Ratios

Given the goal of maximizing revenue recovery, audits that result in no assessment 79% of the time should be of concern. In contrast, the Internal Revenue Service (IRS) conducted over 5.6 million audits between 1994 and 1996. The average percent of audits that had no assessments was 33%. Furthermore, enhancing voluntary compliance is less effective when most of the taxpayers audited are in compliance. The Division's dual goal of ensuring voluntary compliance and maximizing revenue recovery are not mutually exclusive. Lowering the number of audits with no assessments will improve both voluntary compliance and maximize revenue recovery. The best way to lower the number of audits with no assessments is to develop a system that identifies a larger percent of noncompliant taxpayers.

Causes of the Ineffectiveness.

There are two causes for the Division's ineffectiveness in achieving its dual goal of optimizing revenue recovery and encouraging voluntary compliance. The two causes are as follows:

• The major audit selection source used by the Division is random selection which does not rely on any indication of a taxpayer being out of compliance. Therefore, extensive random selection is least effective in identifying noncompliant taxpayers. This contributes to many audits not having assessments.
  
• The Division's audit selection sources are limited to two primary sources (random and referral) in contrast to the more numerous sources that are available.

Random selection gives each taxpayer an equal chance of being audited. In this case the initiation of the audit is not based on any evidence that the taxpayer is out of compliance. Other selection processes such as referrals have a basis of possible noncompliance based on internal or external information. Some of the more commonly used selection methods are shown below.

Common Audit Selection Methods

• Random:Each taxpayer has an equal chance of being selected for an audit. Evidence of possible noncompliance is not the basis for these audits.
  
• Referrals:Taxpayers for audits can be referred by internal or external sources. For example, during a current audit, a field auditor may identify transactions between another company that justifies an audit of the other company. An external source is the public who may notify the Division of potential noncompliant taxpayers. These audits are initiated based on information that suggest possible noncompliance.
  
• By-Cell:Taxpayers can be selected for an audit by reviewing taxpayers who are within certain "cells," such as top remitters of a tax, top utilizers of tax credits, or taxpayers with the highest gross sales.
  
• Prior Audit:The results of a prior audit may dictate a current audit.
  
• The IRS has about 40 audit sources to select taxpayers for audits. The IRS' audit sources do not rely on random selection. Instead, taxpayers are selected for audits based on characteristics indicative of potential noncompliance. One such source is the Discriminant Function (DIF), which is an automated system for scoring individual tax returns according to their potential for recommended tax adjustments (noncompliance). Each individual tax return goes through this system to arrive at a score for each return. The higher the score the greater the probability of noncompliance. Each score is compared to a national cutoff. Those returns that exceed the cutoff are audit candidates. The overall intent of the IRS' selection process is to identify taxpayers that are likely to be out of compliance. IRS officials have stated that they have little incentive to randomly select audits because its other sources generally find more noncompliance at lower costs.

In contrast, the Auditing Division has eight sources to select audits. However, for the last three years (FY 1996 through 1998) the Division has used primarily only two sources to select taxpayers for audits: 1) random; and 2) referrals. Table 4 shows that random and referral selections account for about 97% of the selections. Nearly two-thirds of the audits were randomly selected in 1998.

Table 4

Audit Selection Methods Used From FY 1996 - 1998

The reason the Division has a high percentage of audits that yield no assessments is that it relies substantially on random selection. Random selection will generally yield lower COLLECTION/COST ratios than other selection methods because it is not intended to target taxpayers that are at risk of being out of compliance. Whereas, other selection methods can be designed to audit taxpayers that have a higher probability of noncompliance based on certain characteristics. When a stated goal is to identify Noncompliant taxpayers, the extensive use of random selection is not likely to achieve the desired results to a great extent.

The Auditing Division explains its extensive use of random selection in its Audit Selection Procedures, as follows:

Many times the administrators of Auditing Division are queried on why certain taxpayers are consistently being audited or a taxpayer may feel 'picked on'. Legislative members are particularly sensitive to ensuring that no particular taxpayers are being targeted for needless reasons. The random selection process utilized will ensure that no bias consideration exists.

The Legislative Auditor understands the Division's concerns that Legislative members have for their constituents. However, other selection methods that are more cost-efficient can be used that do not target taxpayers for "needless reasons". For example, a greater use of cell selection is advisable. The Division has used it sparingly in the past, and it can be more cost-efficient than random selection. Furthermore, improving the Division's identification of Noncompliant taxpayers will alleviate bias considerations. Little objections can be made when taxpayers are selected for audits based on indications of noncompliance.

Table 5 illustrates the COLLECTION/DIRECT COST ratio between random, referral and cell selection. ⁽¹⁾ In many cases, random selection yields the lowest COLLECTION/DIRECT COST ratio for all audits conducted from FY 1996 through 1998. The difference is more revealing when the comparison is made on sales tax audits, which were the second largest number of audits conducted. For each year, referral and cell selections were by far more productive in revenue recovery.

Table 5

COLLECTION/DIRECT COST Ratios from FY 1996 - 1998

Moreover, the Division's audit selection procedures indicate the intention to minimize audits that historically result in little recovery of revenue. This is indicated in the following statements:

Other than the occasional review to maintain voluntary compliance there would be little need to concentrate on certain segments of the taxpayer base which have historically proven that there was a low ratio of revenue recovery verses audit costs....Through the review of historical data we may find that the audit of certain accounts results in little revenue recovery. Therefore, we may not concentrate so highly in that area and perhaps review that segment on a random basis, only, to maintain compliance.

With 77% to 79% of audits resulting in no assessments over the last three years, this suggests that the Division has not adequately reviewed the data to minimize audits with no recovery of revenue. The COLLECTION/TOTAL COST ratio is low (2.9 to 1), suggesting the need for improvement. The Division has stated in its performance based budget the need to improve its audit selection process. However, there does not appear to have been much progress in this area. Random selection and referrals are consistently the primary sources for audits, with most of these being random.

Also, the Division's goal is to identify Noncompliant taxpayers. Yet, for the past three years nearly 80% of the audits conducted do not identify Noncompliant taxpayers. The Division does have a scoring system for Corporate Net Income Tax returns. Based on thresholds for certain lines of the tax return, a corporate net income tax return will be audited if it has a score of more than 150 points. However, Corporate Net Income Tax audits had nearly 90% of the audits result in no assessments in 1998. Overall, to have 79% of audits result in no assessments should concern the Division when identifying Noncompliant taxpayers is an important goal to achieve.

As stated previously, the Division's audits can become more effective by relying less on random selection and developing sources that rely more on indicators of noncompliance. By decreasing random selections, other selection methods can be used more extensively. To illustrate this point, the Legislative Auditor's Office conducted a survey of other state tax departments that collect a state sales tax. Information was collected on what extent tax departments use various audit selection methods in auditing their sales tax accounts. Appendix A contains the results of the survey. Of the 46 states that have a state sales tax, 31 responded to the survey. Compared to other states, the Division uses random selection to a much greater extent than other states, and West Virginia's use of other selection methods is more limited than other states. Most of the states (18) indicated no use of random selection. The highest use was 48% by Arkansas and the average use of random selection for sales tax audits is 7.9%, compared to 67.5% of sales tax accounts for West Virginia.

It is also interesting to note how other selection methods are used more extensively by other states. Table 6 provides a summary. For example, cell selection was not used by West Virginia in 1998, and in years in which it was used, the usage was less than 1%. Whereas, with few exceptions most states use the cell selection extensively, with the average near 27%. Other (unnamed) selection methods account for nearly 27%, compared to 2.4% for West Virginia. This indicates that most states use a greater variety of selection methods than West Virginia, and random is the least used method.

Table 6

Selection Processes Used by Other States in Auditing Sales Tax Accounts

Effect number 1:Loss of Additional Revenue

The Auditing Division has collected a total of $24,098,135 in the three fiscal years 1996 through 1998. That is an average collection of a little more than $8 million per year. The average percentage of audits that resulted in an assessment over this time period was 22.5%. This means that on average, $8 million has been collected from 22.5% of the audits conducted each year. This indicates that an average of $355,555 is collected for each percentage of audits that results in an assessment, or $3,555,555 for every ten percentage points.

If the Auditing Division can improve its methods of identifying taxpayers who are potentially out of compliance, the additional revenue to the state will be significant. For example, lowering the percentage of audits that have no assessments from 80% to 60% could potentially result in over $7 million in additional tax collections each year.

Effect number 2:Honest Taxpayers are Needlessly Hassled

Another reason that the Division should consider reducing its use of random selection is the possible sentiment that taxpayers object to being audited based on no evidence of error, wrongdoing or noncompliance. There has been an attempt to pass federal legislation to prohibit the IRS from randomly selecting taxpayers for audits. ⁽²⁾ The objection has been that innocent taxpayers should not have to go through the burden of an audit when there is no basis that they are out of compliance.

The costs and burdens of an audit imposed on taxpayers has not been measured, however, a GAO report done on the IRS made the following statement:

According to IRS, any audit imposes some level of costs and burdens on taxpayers. IRS has not measured these costs and burdens for any type of audit. As discussed in several of our testimonies, accurately measuring taxpayer costs and burdens is difficult. Even so, IRS has recognized the importance of these measures and has been making efforts to develop them. ⁽³⁾

3. Tax Administration: IRS' Use of Random Selection in Choosing Tax Returns for Audit, United States General Accounting Office, Washington D.C., (Letter Report, 02/06/98, GAO/GGD-98-40), p. 3.

4. Internal Revenue Service Restructuring and Reform Act of 1998 (Public Law 105-206, July 22, 1998).