Does representation matter in IRS office audits?
This research provides empirical evidence regarding one of the perceived benefits of representation--a reduction in the final tax assessment by the IRS--and is a first step in determining when an individual should hire a professional representative for an IRS office audit. The results indicate that the final tax assessment is significantly less for taxpayers with representation during an IRS office audit, both in dollars and as a percentage of the potential deficiency.
Tax practitioners often represent clients before the Internal Revenue Service (IRS) and it is assumed that these clients can expect to see significant benefits during the audit as a result. (1) Articles suggest that a potential benefit from representation is a reduction in the final IRS assessment of tax and penalties (Evanson 1995; Sommers 1995; Hicks 1991).
To the best of our knowledge, however, there is no research that directly analyzes the impact of taxpayer representation on IRS audit outcomes because there is no data on whether the taxpayer was or was not represented. The IRS makes available information regarding the Individual Taxpayer Compliance Measurement Program (TCMP) audits, but the file does not include information regarding whether the taxpayer was represented during the audit. (2)
In a TCMP audit, the IRS audits every line item on the tax return in order to determine the level of compliance for each separate item. In contrast, office audits involve an "in-person" interview and records review with the auditor at the IRS office. (3) Historically, individuals were more likely to be subject to an office audit than a TCMP audit. (4) An office audit is often used for middle-income individual taxpayers including those with small sole proprietorships. The office audit notice usually identifies specific items or areas of interest and requests the taxpayer to support those items, thereby significantly reducing the audit's scope.
This research uses data on 45 taxpayers subject to office audits by the IRS from the Federal Student Tax Clinic operated by the University of North Texas (UNT). This unique dataset provides the opportunity to obtain empirical evidence regarding the impact of tax-payer representation on IRS audit outcomes and to analyze specific audit adjustments for a sample of office audits. This research provides information regarding a much more typical IRS audit than a TCMP audit. Although the study involves student representatives, the students were well prepared for the audit issues identified in the IRS audit notice and had a significantly greater level of knowledge than the taxpayers. (5)
Research regarding the benefits of taxpayer representation during IRS office audits is important for several reasons. First, it provides empirical evidence regarding one of the perceived benefits of representation, a reduction in the final tax assessment by the IRS. This research is a first step in addressing whether an individual should hire a professional representative for an IRS office audit. Since hiring a professional representative is costly, a rational individual would perform a cost benefit analysis and make the decision to hire a professional representative only if the expected tax savings from representation is greater than the cost of representation. This paper provides empirical evidence that representation results in a significant reduction in the final tax assessment from an office audit, both in dollars and as a percentage of the potential deficiency. This information should assist individuals in estimating their expected tax savings from representation. Given an estimate of the expected tax savings from representation, the individual can then compare it to the expected cost of representation to decide whether to hire a professional representative.
This research also provides descriptive statistics of IRS office audit information from a sample of audits. Hite and Hasseldine (2003) examine a large sample of IRS office audits to determine whether there are more audit adjustments on returns completed by paid-preparers than on self-prepared returns. Their study separated audit adjustments into two categories, deduction adjustments and income adjustments. Our results are consistent with Hite and Hasseldine's (2003) findings that most adjustments come from deduction errors rather than income errors. Our study extends Hite and Hasseldine (2003) by separating the audit adjustments into specific types of deduction and income adjustments, providing additional descriptive information.
Our results indicate that representation during an office audit benefits the taxpayer, providing empirical evidence to augment the popular perception. Specifically, we find that taxpayers who are represented see a final tax assessment that is approximately $1,300 lower than those who are not represented (or 40 percent of the potential deficiency determined using the items identified for audit in the IRS notice). This study indicates that well-prepared students significantly reduce the final tax deficiency, adding to Hite and Hasseldine's (2003) evidence "on the positive role that qualified accountants play in terms of decreasing deduction adjustments."
The remainder of the paper is organized as follows. The second section reviews prior research regarding IRS audits and taxpayer compliance and develops the research hypothesis. The third section develops the empirical models, and the fourth section describes the sample. The fifth section reports the empirical results, and the final section summarizes and provides conclusions.
The public press offers numerous articles regarding how to handle an IRS audit that attempt to answer the question of whether you need a representative. The articles argue that the answer depends on (1) the nature of the audit, (2) the issues involved, and (3) the experience and personality of the taxpayer. Generally, such articles in the public press indicate that the taxpayer should retain a tax expert for an IRS audit if that taxpayer cannot substantiate an income or expense item, or if a legal issue is involved (Sommers 1995). Another concern is whether the taxpayer may inadvertently create an issue by saying too much or saying the wrong thing (Weiner 2001; Anderson 1999; Szabo 1994).
Roberts (1995) investigates whether the presence of a CPA representative during an audit influences an IRS agent in his decision to assess negligence penalties on a taxpayer. He found that representation by a CPA had no significant impact on the agent's decision to assess negligence penalties. Roberts' (1995) study is the only academic research that has addressed the impact of representation during an IRS audit and is limited to the specific issue of negligence penalties.
In order to identify other factors that may influence the dollar amount of IRS audit adjustments, it is necessary to review the tax practitioner research and research using the TCMP files. Research on tax practitioners has focused on two areas: (1) the demand for tax return preparation services and (2) whether tax return preparers impact the level of compliance. On the issue of paid preparers affecting the level of compliance, the research suggests that returns prepared by CPAs and lawyers have a higher level of noncompliance than do those prepared by other paid professionals or by the taxpayers themselves (Dubin et al. 1992; Erard 1993). Long and Caudill (1987) find that the reported income tax liability is relatively lower on returns completed by paid-preparers compared to those prepared by taxpayers. Klepper et al. (1991) conclude that tax practitioners as a single group promote compliant reporting on unambiguously defined line items and noncompliant reporting on ambiguously defined items. Contrary to research utilizing TCMP data, Hite and Hasseldine (2003) examine a sample of IRS office audits and find that CPA prepared returns had fewer audit adjustments than self-prepared returns. This research implies that the use of a paid preparer may impact the dollar amount of IRS audit adjustments. Therefore, a control variable for paid preparer is included in this study.
Other research utilizing the TCMP data files provides insight regarding factors other than representation during audit that may influence the dollar outcome of the office audit. Feinstein (1991) found that Schedule C filers are significantly more likely to evade than the average taxpayer. Madeo et al.'s (1987) study on taxpayer compliance behavior indicates that the source of income (wages versus self-employment income) is critical in determining taxpayer compliance. These findings suggest that potential audit adjustments for Schedule C fliers may be greater than for other taxpayer groups. Therefore, a control variable for Schedule C is included in this study.
HYPOTHESIS AND EMPIRICAL MODELS
The purpose of this research is to provide evidence regarding one of the potential benefits of taxpayer representation during an IRS audit, namely the reduction in the final tax liability determined during office audits. (6) In order to determine if representation results in actual dollar benefits, we test the following hypothesis:
H1: Final tax deficiencies are relatively lower for taxpayers with representation during IRS audits than for taxpayers without representation.
Selection bias results whenever there is non-random sampling. The term self-selection bias is used when the non-randomness arises from individual choices. In this study, individuals chose whether to use a representative based on their perception of the relative costs and benefits of representation, resulting in a potential self-selection bias. OLS procedures that ignore the non-randomness of the sample may be biased. The most widely used methodology for addressing self-selection bias is the two-stage process employing a correction term that was originally developed by Heckman (1976, 1979). (7)
The first stage involves the representation decision. The individuals in our study faced a choice regarding representation during an IRS audit. Define a latent variable R* for individual i that reflects his/her preference for representation and let it be denoted by [R.sub.i]*. The individual selects representation if [R.sub.i]* > 0 and remains unrepresented if [R.sub.i]* < 0. Our model hypothesizes the [R.sub.i]* is a linear function of variables describing the individual's characteristics, including income level and audit issues, along with a random error term:
(1) [R.sub.i]* = [theta][Z.sub.i] - [[epsilon].sub.i]
where [[epsilon.sub.i] is normally distributed with zero mean and unit variance, [Z.sub.i] is a vector of explanatory variables and [theta] is an unknown vector of coefficient parameters. The latent variable is not observed. Instead we observe:
[R.sub.i] = 1 if [R.sub.i]* > 0, i.e., [theta][Z.suib.i] > [[epsilon.sub.i]
[R.sub.i] = 0 if [R.sub.i]* [greater than or equal to] 0, i.e., [theta][Z.sub.i] [greater than or equal to] [[epsilon.sub.i].
The second stage addresses the item of interest, the potential benefit of representation during IRS audits. For an individual, [y.sub.i] denotes the outcome of the IRS audit, which is a function of explanatory variables including the representation variable and a random error term:
(2) [y.sub.i] = [beta][X.sub.i] + S[delta][R.sub.i] + [u.sub.i]
where [u.sub.i] is normally distributed with zero mean, X denotes a vector of variables that impact the final audit outcome, R is the representation dummy variable described above, and [beta] is a vector of unknown parameters.
The parameters in Equation (2) cannot be estimated directly by OLS because the presence of self-selection bias causes [u.sub.i] to be correlated with [[epsilon].sub.i]. The covariance matrix of [[u.sub.i], [epsilon].sub.i]] is assumed to be normally distributed, and the conditional mean of the error term [u.sub.i] can be shown as follows (Heckman 1976; Lee 1982; Maddala 1983):
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.]
where [[sigma].sub.[epsilon]] = Cov[[u.sub.i], [[epsilon].sub.i]], [PHI] is the cumulative distribution of the standard normal function,
and [theta] is its density function. The term [- [phi][[theta]Zi]/[PHI][theta]Zi]], known as the Mills ratio and designated as [lambda], represents the selectivity correction term for the sample.
Using Heckman's two-stage process, the representation choice equation is estimated using Probit analysis. The estimated value of [??][Z.sub.i] is then used to generate the Mills ratio ([lambda]) for each sample observation. In the second stage, the Mills ratio ([lambda]) is added to Equation (2), which is then estimated using OLS. Equation (2) is rewritten as follows:
(3) [y.sub.i] = [beta][X.sub.i] + [delta][R.sub.i] + [[sigma].sub.[epsilon]] [??] + [w.sub.i]
where [w.sub.i] is a random error term with zero mean and constant variance. A natural test for self-selection bias arises, since the coefficient of [??] estimates the degree of covariance between the error terms in Equations (1) and (2). A statistically significant selection coefficient [??] implies that unobserved factors leading some individuals to choose representation are correlated with unmeasured factors that concurrently affect the audit outcome.
Model Specification Representation Choice Model
The representation choice model used to develop the Mills ratio to control for self-selection bias is as follows:
(4) [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.]
where REP is a dummy variable coded 1 if a student represented the taxpayer during the office audit, or 0 otherwise. PDEFIC equals the potential tax deficiency calculated based on the specific items identified in the IRS audit notice, reduced by the tax effect of errors on the return as originally filed. Higher potential tax deficiencies may influence a taxpayer's decision to seek representation. PAID is a dummy variable coded 1 if a paid preparer signed the original return filed by the taxpayer, or 0 otherwise. ORGSCHC is the income or loss originally reported on Schedule C. ERR is a dummy variable coded 1 if the tax return as originally filed contained an error, or 0 otherwise. If the tax return as originally filed includes an error, then taxpayers may be less likely to select representation because they may be concerned that the representative will disclose the error to the IRS. The sample section includes additional discussion of tax return errors.
Potential Benefit of Representation Models
The model for testing the potential benefit of representation during IRS audits on the dollar amount of the final tax assessment is as follows:
(5) [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.]
where DEFIC is the final tax assessment (refund) resulting from the audit reduced by the tax effect of gross income adjustments and errors on the return as originally filed. PDEFIC equals the potential tax deficiency calculated based on the specific items identified in the IRS audit notice and reduced by the tax effect of errors on the return as originally filed. REP is a dummy variable coded 1 if a student represented the taxpayer during the office audit, or 0 otherwise. PAID is a dummy variable coded 1 if a paid preparer signed the original return filed by the taxpayer, or 0 otherwise. SCHC is a dummy variable coded 1 if the return included a Schedule C, or 0 otherwise. The selection term, [lambda], determined from the representation equation, is included to correct for self-selection bias.
Our hypothesis regarding the benefit of representation on the final tax deficiency predicts a negative sign on REP. PDEFIC provides a control for differences in the magnitude of the audits based on the original items identified by the IRS for audit. We expect taxpayers with greater dollar amounts of items identified for audit to pay a larger final deficiency suggesting a positive sign for PDEFIC. PAID controls for the potential impact of tax return preparers on the level of compliance (Erard 1993; Dubin et al. 1992; Long and Caudill 1987). The research regarding the impact of paid preparers on reporting is mixed, so we do not predict a sign for PAID (Hire and Hasseldine 2003; Dubin et al. 1992; Klepper et al. 1991). If taxpayers filing a Schedule C are more likely to evade taxes, then we predict the sign on SCHC to be positive (Feinstein 1991; Madeo et al. 1987).
We adjusted the amount of potential tax deficiency (PDEFIC) and/or final tax deficiency (DEFIC) for the tax effect of gross income adjustments and return errors where representation during the audit could not impact the audit outcome. For example, the final tax deficiency includes adjustments to gross income not considered in the calculation of the potential deficiency. The gross income adjustments resulted when the taxpayer failed to include on the return amounts reported to the IRS on Forms 1099 as originally filed (see complete discussion in the "Results" section). In addition, the final tax deficiency and potential tax deficiency amounts include adjustments for errors made on the return as originally filed. A review of the taxpayer files found that one taxpayer included an office-in-home deduction even though the Schedule C already showed a net loss, two cash-basis taxpayers deducted bad debt expenses on Schedule C, and three taxpayers reporting net income on Schedule C failed to pay self-employment tax. Since representation during the audit would not impact these adjustments, we adjusted the amount of DEFIC and/or PDEFIC in Equation (5).
From 1993 through 1997, UNT operated a Federal Student Tax Clinic (Tax Clinic) in conjunction with the Dallas (Texas) District Internal Revenue Service. (8) The IRS included a notice of the Clinic's availability in the letter they sent notifying taxpayers of an impending office audit. Taxpayers interested in using the service contacted the Tax Clinic and were assigned to a student representative by the clinic director. UNT provided the service free of charge. (9) All of the returns referred to the Tax Clinic were office audit cases.
Graduate tax students pursuing the Master of Science degree in accounting with a specialty in tax operated the Clinic while a member of the graduate accounting faculty and a CPA supervised its operation. The students contacted each taxpayer who had requested assistance, reviewed the taxpayer's returns, researched potential issues, and met with them to discuss the upcoming audit. The director worked with the student representatives after they met with the taxpayers to ensure that all the potential problems and opportunities on the returns were identified prior to their audit appointments. If a taxpayer chose to have student representation at the audit, then the director reviewed the proposed audit changes before the taxpayer agreed to the adjustments. The Tax Clinic followed the procedures outlined in Price (1995). The Appendix contains these procedures.
The Tax Clinic worked 32 cases during the 1995-96 academic year and 49 cases during the 1996-97 academic year for a total of 81 cases. (10) Of the 81 files, only 45 contained a complete set of information necessary for this study. Of the 36 other cases, 24 were individuals who set up an initial appointment, but either did not show up at all or decided not to use the Tax Clinic after the initial appointment. In 11 of the 36 cases, information regarding the final agreed-upon audit adjustments was not included in the file. In the final case, the file included information regarding the final audit adjustments, but did not include the tax return as originally filed.
Each of the 45 files that contained the information necessary for the study included: (1) a projection of the anticipated refund or deficiency based on the initial letter notifying the taxpayer of the audit, (2) the actual refund or deficiency agreed to by the taxpayer, (11) (3) the individual items originally questioned by the IRS, (4) the individual items resulting in actual audit adjustments, (5) information about whether a paid tax return preparer originally prepared the return, (6) information about whether the audit was extended into other tax years or issues and (7) information about whether the taxpayer was represented by a student during the audit. Taxpayers utilizing the services of the Tax Clinic signed a consent form allowing use of their information for educational and research purposes. Of the 45 taxpayers included in the study, 31 taxpayers requested student representation and 14 tax-payers declined student representation during the office audit. The taxpayers declining representation attended the audit by themselves. None of them hired another representative.
Taxpayers generally declined representation for one of three reasons. First, the taxpayer was uncomfortable with the fact that the student knew all of the potential problems with the return and could potentially disclose that information to the IRS auditor. Although the students explained that they could not disclose problems without the taxpayer's permission, the taxpayer still did not want the student to attend the meeting. Second, after working with the student to understand the audit process and potential issues, the taxpayer was comfortable attending the audit alone. Third, in a few cases, the tax return contained obvious errors that the Clinic director concluded must be disclosed to the IRS. (12) When the taxpayers refused to allow disclosure of the errors, the Clinic director did not allow the students to attend the IRS meetings.
Descriptive Statistics and Univariate Tests
Table 1 shows whether the taxpayer used a paid preparer and summarizes the filing status of the returns included in the final sample. Twenty-eight returns (62 percent) indicated married filing jointly, ten returns (22 percent) selected head of household, five returns showed single filing, and two returns chose married filing separately. Of the ten head of household returns, the IRS questioned the dependency exemption on seven returns (70 percent). Paid preparers signed eight (18 percent) of the returns, while the individual tax-payer prepared the remaining 37 (82 percent). (13)
Table 2 provides descriptive statistics for specific line items from the sample returns. The average taxable income for the sample was $18,792, an indication that the sample generally consists of lower- and middle-income taxpayers. Taxable income as originally reported ranged from -$9,002 to $82,582, with five returns reporting negative taxable income, 26 returns reporting taxable income between $0 and $25,000, nine returns with taxable income between $25,000 and $50,000, and five returns reporting over $50,000 of taxable income. A Schedule C was included in 37 of the 45 returns (82 percent). The Schedule C income (loss) reported ranged from -$17,915 to $47,461, with 24 returns reporting a loss, 10 returns reporting net income between $0 and $25,000, and three returns reporting net income between $25,000 and $50,000. Of the 13 returns reporting net income on Schedule C, nine paid self-employment tax on the return as originally filed, one reported net income of less than $400 (the threshold amount for paying self-employment tax), and three reported net income over $400 but failed to calculate self-employment tax.
The t-tests for differences between the returns with and without representation indicate no significant differences between the groups for adjusted gross income, taxable income, and total tax. Total tax equals the tax computed on taxable income less tax credits plus self-employment tax. The only significant difference between the two groups of returns is the net income (loss) reported on Schedule C (t = -1.98, p = .0540). The group without representation reported an average of -$6,933, and the group with representation reported an average of $1,767. (14)
Table 3 summarizes the specific items identified in the IRS audit notice, the number of IRS notices including each item, and the average dollar amount claimed on the return as originally filed. The most frequently questioned items were deductions taken on Schedule C, with 36 (80 percent) of the 45 audit notices identifying at least one Schedule C deduction. Of the 37 returns that contained a Schedule C, deductions were questioned on 36 of them and self-employment tax was questioned on the other return. Car and truck expenses were questioned most frequently (16 returns), followed both by depreciation expense and other expenses (13 returns each). The IRS questioned travel, meal, and entertainment expenses on 12 returns, cost of goods sold on nine returns, and phone/utility expenses on eight returns. The combined average dollar amount identified for audit on a Schedule C was $13,560. (15) The largest individual item was wage/labor expense ($10,013).
Twenty-two taxpayers reported itemized deductions on their returns as originally filed (Table 2). The IRS questioned the itemized deduction amount on nine returns. The most frequently identified item for audit was employee business expenses (identified on five returns). (16) The combined average dollar amount of itemized deductions identified for audit was $7,683, and the largest individual item was employee business expenses ($7,751).
The IRS questioned the number/qualification of dependents on ten returns. Eight tax-payers claimed the child care credit (Table 2), and it was identified as an audit issue on five returns. The earned income credit was originally claimed by three taxpayers (Table 2) and was identified as an audit area on all three returns. The IRS specifically identified self-employment tax as an audit issue on one return. Although not specifically identified in the IRS notice, some of the previously discussed Schedule C adjustments may impact both the amount of self-employment tax and the deduction for one-half of self-employment tax.
The t-tests of differences in potential audit adjustments between taxpayers with representation and those without representation indicate that none of the differences between the groups are significant. Even though individual items identified for audit differed in the two groups, the overall potential difference in total tax was not significantly different between the groups with and without representation. This statistic suggests that any difference in final audit adjustments between the two groups is not explained by a difference in the tax impact of potential adjustments identified in the IRS audit notices.
Table 4 summarizes the actual agreed-upon audit adjustments for the sample returns. Adjustments to gross income were made on seven returns. Gross income increased on five returns, three due to unreported interest income, one due to an unreported state income tax refund and one due to underreported gross receipts on Schedule C. Gross income decreased on two returns. One taxpayer originally double reported Schedule C gross receipts as wage income and a second taxpayer did not originally report a loss on sale.
Thirty-one taxpayers agreed to Schedule C expense adjustments, with an average adjustment of $4,758. The other five taxpayers ended up with no Schedule C adjustments. Car and truck expenses were adjusted on 15 returns, with an average adjustment of $2,307. (17) All nine taxpayers agreed to itemized deduction adjustments (Table 3), with an average adjustment of $4,348. (18) Seven taxpayers agreed to adjust the number of dependents claimed including five taxpayers originally identified by the IRS. The other two taxpayers provided support for dependents not claimed on the original return. The child care credit was adjusted on only one return. No adjustments were made to the earned income credit. On the one return with an identified self-employment tax issue, the taxpayer agreed to a $5,568 adjustment. Ten additional taxpayers agreed to self-employment tax adjustments because of adjustments to Schedule C. (19) For all 10 returns, disallowed Schedule C deductions increased net earnings from self-employment, which increased self-employment tax.
A t-test of the overall difference in actual tax deficiency between taxpayers with representation and those without representation indicates that taxpayers without representation paid a significantly higher tax deficiency (t = 2.98, p = 0.07) than taxpayers with representation. This test supports HI indicating that the final assessments of taxpayers represented during an IRS audit are lower than those of taxpayers not represented. The t-tests of differences in the specific audit adjustments between the two groups are significant for the adjustment to Schedule C expenses (t = 3.38, p = 0.05). Reviewing the total dollar amount of potential audit adjustments (averages reported on Table 3) and the actual dollar amount disallowed as a result of the audit (averages reported on Table 4), only 55 percent of Schedule C deductions were allowed for taxpayers without representation, while 78 percent of Schedule C deductions were allowed for taxpayers with representation. The differences between itemized deduction and dependency exemption adjustments are not significant.
Table 5 reports the descriptive statistics for the regression variables. The Probit estimate of the representation choice equation (Equation (4)) is presented in Table 6, Panel A and the OLS estimate of the impact of representation equation (Equation (5)) is reported in Panel B. The results in Panel A indicate that the representation model is significant. The only significant explanatory variable is the income or loss on the Schedule C as originally filed. This is consistent with the univariate tests indicating that the only significant difference between the "with representation" and "without representation" groups is the average Schedule C net income (loss).
Table 6, Panel B reports the results for Equation (5) analyzing the impact of representation on final tax deficiencies. The results are consistent with the prediction that representation reduces the final tax deficiency. Specifically, the significant, negative coefficient on REP indicates that taxpayers with representation during the office audit paid final tax deficiencies that were lower by an average of $1,329. As expected, a significant positive relation exists between the control variable, PDEFIC, and the actual deficiency. This is consistent with taxpayers who have a larger potential deficiency based on the items identified for audit paying a larger final assessment. The remaining control variables, PAID and SCHC, are not significant, so the final audit deficiency is not related to the use of a paid preparer or the presence of Schedule C. (20) The selection variable, X, is not significant. This indicates that the sample does not show evidence of covariation between unobserved factors in the decision of whether to choose representation.
The impact of representation on the final audit adjustment in Equation (5) may be influenced by the relative size of the potential versus final tax deficiency. In order to evaluate the impact of representation without this potential confounding effect, we utilize a change model where the dependent variable (CHANGE) is the difference between the potential audit deficiency (PDEFIC) and the final tax deficiency (DEFIC) as a percentage of the potential audit deficiency (PDEFIC). If representation during audit results in a relatively lower final tax deficiency, then the difference between the potential audit deficiency and the final tax deficiency should be greater for taxpayers with representation and, therefore, we predict a positive sign on REP. The resulting model is as follows:
(6) CHANGE = [[alpha].sub.0] + [[alpha].sub.1]REP + [[alpha].sub.2]PAID + [[alpha].sub.3]SCHC + [[alpha].sub.4] [[lambda].sub.i] + [w.sub.i].
REP, PAID, SCHC, and [lambda] are the same as Equation (5). Since the research on paid preparers is mixed, we do not predict a sign for PAID. We predict a negative sign on SCHC. The selection variable (h) is again included to correct for self-selection bias.
Table 7 reports the results for regression Equation (6) analyzing the impact of representation on the change between the potential and final tax deficiency. The coefficient on REP is positive and significant. The results indicate that the difference between the potential audit deficiency and the final tax deficiency is approximately 40 percent greater for taxpayers with representation. The coefficients on PAID and SCHC are not significant. (21) The change model provides additional support for the hypothesis that final tax deficiencies are significantly lower for taxpayers with representation during IRS audits.
The final tax deficiency refund) and the potential tax deficiency in Equation (5) include self-employment tax. When adjusted self-employment income is greater than $400, adjustments that increase Schedule C net income automatically increase self-employment tax. Arguably, representation matters only with respect to the potential Schedule C adjustments and not for the automatic adjustments to self-employment tax. To test the sensitivity of our results to self-employment tax adjustments, we removed self-employment tax adjustments from the final tax deficiency and the potential tax deficiency variables in Equation (5). The results are reported in Table 8.
The coefficient on REP remains significant and negative. Therefore, HI is still supported after reducing the final deficiency and/or potential audit deficiency for the automatic adjustments to self-employment tax. Consistent with Equation (5), the significant positive relation for the potential deficiency (APDEF NO SE) continues, and the coefficients on PAID, SCHC, and X remain insignificant.
The final tax deficiency (refund) and the potential tax deficiency in Equation (5) depend not only on adjustments to taxable income, but also on the taxpayer's current year marginal tax rate. Representation can only impact the adjustment amount. To test the sensitivity of our results to the impact of marginal tax rates, we use the final adjustment to taxable income and the potential adjustment to taxable income as the dependent and independent variables in Equation (5). (22) The results are reported in Table 9. The coefficient on REP remains significant and negative. Therefore, H1 is still supported based on adjustments to taxable income rather than total tax.
CONCLUSIONS AND LIMITATIONS
The study examines the potential benefit of taxpayer representation during an IRS office audit on the final tax assessment. Because data on representation during IRS audits is not readily available, previous IRS audit research focuses on other factors influencing the final audit outcome, including income sources and the use of a paid preparer. Our access to specific audit information on a limited number of IRS office audits allows us to analyze the potential benefits of taxpayer representation.
Overall, the results indicate that representation during an office audit results in a final tax assessment that is approximately $1,300 lower for our sample of taxpayers. The change model suggests that the final deficiency is 40 percent lower for taxpayers with representation. The descriptive analysis extends the work of Hire and Hasseldine (2003) by detailing the specific deductions identified for audit by the IRS and the resulting adjustment for each type of deduction.
The representatives in this study were not tax professionals, but students without prior experience representing taxpayers before the IRS. Although they lacked experience, the students did extensive research on the issues involved in the audit and were coached by the Tax Clinic director (a CPA specializing in individual and corporate taxation with an accounting doctorate), who had been a Revenue Agent for six years. Therefore, the students had a significantly greater level of tax knowledge than the taxpayers they represented. We find that well-prepared students significantly reduce the final tax deficiency, adding to Hite and Hasseldine's (2003) evidence "on the positive role that qualified accountants play in terms of decreasing deduction adjustments."
The study involves a small sample of taxpayers subject to office audits in the Dallas IRS District, thereby limiting the generalizability of the results. (23) However, given the limited availability of data regarding representation, the analysis adds to our knowledge regarding IRS audits and the benefits of representation. We believe our results provide a useful foundation for addressing the issue of when an individual should hire a professional representative for an IRS office audit. Since hiring a professional representative is costly, a rational individual would do a cost benefit analysis and make the decision to hire a professional representative only if the expected tax savings from representation is greater than the cost of representation. This paper provides empirical evidence that representation results in a significant reduction in the final tax assessment from an office audit both in dollars and as a percentage of the potential deficiency.
Procedures for the UNT Federal Student Tax Clinic
The University of North Texas (UNT) Federal Student Tax Clinic was operated in conjunction with the Dallas District (Dallas, Texas) Internal Revenue Service. Most of the returns assigned to the tax clinic were office audit cases requiring the taxpayer and representative to travel to a local IRS field or district office for examination. The IRS included notice of the clinic's availability in all letters notifying a taxpayer of an impending office audit.
The clinic director had the primary responsibility for all operational aspects of the tax clinic program. The UNT clinic director is a member of the graduate accounting faculty and is also a CPA. Graduate tax students pursuing the Master of Science degree in accounting with a tax specialty staffed the clinic. Students participated in the clinic when enrolled in the graduate corporate tax course, which was taught by the clinic director. Participation in the clinic was mandatory and usually constituted 10 percent of the students' grade in the course. Enrollment in the corporate tax class was limited to 25 students but averaged 20 students per semester.
Variability of student caseload was experienced from semester to semester. However, all students worked at least one case but never more than two. There were always enough cases to assign each student at least one case. The percentage of the student's grade for participation in the clinic was adjusted depending on the number of cases completed. The clinic director would not accept additional cases from the IRS after each student was assigned a second case. Managing the caseload was coordinated with the local IRS.
A summary of the procedures utilized for each case follows:
Step Procedure 1 Each student is given a clinic file that outlines clinic objectives and procedures, responsibilities of student participants, a copy of Circular 230 governing practice before the IRS, and several forms that must be completed at various stages in working assigned cases. 2 Taxpayers call the clinic secretary after receiving notice of an IRS audit and request assistance. The clinic secretary completes a "Taxpayer Information" form that includes background information on the taxpayer and the tax return including the items to be substantiated during the audit. The clinic secretary delivers the case file to the clinic director who assigns the case to the student. 3 The student: a. Immediately contacts the taxpayer to set up the pre-audit conference. Conferences are held at any location that is mutually agreeable to the student and the taxpayer. Taxpayers are reminded to bring to the conference a copy of tax returns, the IRS appointment letter, and substantiation for the items being questioned. b. Reviews the basic rules of the tax law related to the areas marked on the checklist in preparation for the consultation with the taxpayer. 4 The student conducts the pre-audit conference with the taxpayer in an informal, yet professional manner. The option of having the student accompany the taxpayer to the audit is discussed during this conference. Students are required to keep the discussion pointed at the objectives as follows: a. Go over the entire return, not just the items marked on the checklist. b. Learn as much as possible about the taxpayer's business affairs. Think about income and deductions on the return and use them as a basis for questions. c. Try to think of tax benefits missed by the taxpayer. d. Look at the taxpayer's substantiation and think of other sources of evidence if they are needed. Stress the need for well-organized supporting documents. e. Make notes as necessary during the conference, especially about points of law that require additional research. 5 Following the conference(s) with the taxpayer, the student prepares a "Pre-Audit Consultation Report" showing questionable items uncovered during the pre-audit conference. For each item listed on the form, students are required to discuss: a. Relevant facts uncovered. b. The precise issued posed by the facts. c. Conclusions and authorities to support them. d. Recommended advice to the taxpayer. 6 Students arrange a conference with the clinic director prior to the taxpayer's audit appointment. The director reviews the file, including the return and student report. The purpose is to uncover additional problems and opportunities on the return. If the taxpayer requests representation, then the student and the director decide whether the student should accompany the taxpayer to the audit. If the student attends the audit, then the taxpayer must sign a Power of Attorney. Regardless of whether the taxpayer requests representation, after the pre-audit conference, the student relays the results of the review with the clinic director to the taxpayer before the audit. 7 Taxpayers are instructed that they are not to agree to audit changes until the clinic director has had a chance to review the proposed adjustments. 8 The office audit usually closes the case. Following the audit, the student completes the "Post-Examination Report" that shows the outcome of the audit and contains an evaluation of the clinic services by the taxpayer. A comparison of the actual adjustments and clinic director's and student's expectations are also summarized in this report. 9 The final requirement for the student is preparation of a memo to the file. This memo summarizes the facts, issue(s), discussion of authorities, conclusions, and recommendations given to the taxpayer as well as a summary of the actual outcome of the case. Additionally, the student makes written comments on the tax clinic experience. All completed cases are passed to the clinic director who closes and files the case. Adapted from Price (1995, 420-421). TABLE 1 Filing Status and Use of Paid Preparer for Taxpayers with and without Representation during the IRS Office Audit Without With Representation Representation Total Number % Number % Number % Filing Status Single 1 7 4 13 5 11 Head of Household 3 21 7 23 10 22 Married Filing Joint 8 58 20 64 28 62 Married Filing 2 14 0 0 2 5 Separately Total 14 100 31 100 45 100 Use of Paid Preparer Paid Preparer 2 14 6 19 8 18 Prepared by Taxpayer 12 86 25 81 37 82 Total 14 100 31 100 45 100 TABLE 2 Dscriptive Statistics Returns as Originally Filed for Taxpayers with and without Representation during the IRS Office Audit Without Representation n Mean Median Min. Max. Std. Dev. Wages 12 35,834 35,772 2,272 61,397 19,846 Interest 5 981 571 422 1,762 640 Dividends 3 843 709 154 1,667 765 Capital Gains 5 11,175 2,614 -3,000 27,906 14,955 Retirement 1 59,070 59,070 NA NA NA Income Other 3 4,277 7,283 -2,410 7,957 5,800 Income (a) Schedule C 14 -6,933 -7,221 -17,915 8,554 8,789 Adjustments 4 495 552 229 647 187 for AGI AGI 14 33,374 32,211 5,317 69,908 16,619 Itemized 6 11,701 12,030 3,950 18,429 5,142 Deductions Taxable Income 14 20,031 21,686 -5,583 49,129 15,655 Self 2 928 928 647 1,209 397 Employment Tax Child Care 1 480 480 NA NA NA Credit Earned Income 1 508 508 NA NA NA Credit Total Tax (b) 14 3,602 3,520 0 11,493 3,096 With Representation Std. n Mean Median Min. Max. Dev. Wages 30 28,880 25,174 765 90,674 12,952 Interest 10 568 134 20 3,210 975 Dividends 9 925 217 2 2,938 1,108 Capital Gains 4 10,296 4,507 41 32,128 14,733 Retirement 3 19,049 22,190 3,648 31,308 14,095 Income Other 10 4,445 1,270 -7,986 38,596 12,986 Income (a) Schedule C 23 1,767 -6,467 -15,273 47,461 16,728 Adjustments 10 1,666 1,013 249 6,308 1,840 for AGI AGI 31 33,807 26,054 3,125 101,348 24,147 Itemized 16 12,070 11,043 8 088 20,901 3,439 Deductions Taxable Income 31 17,514 7,978 -9,002 82,582 22,907 Self 7 1,561 1,018 498 4,814 1,493 Employment Tax Child Care 7 710 730 480 1,008 230 Credit Earned Income 2 860 860 552 1,168 435 Credit Total Tax (b) 31 3,381 1,564 -1,168 (c) 18,324 4,767 TOTAL n Mean Wages 42 30,867 Interest 15 706 Dividends 12 905 Capital Gains 9 10,784 Retirement 4 29,054 Income Other 13 4,445 Income (a) Schedule C 37 -1,525 Adjustments 14 1,331 for AGI AGI 45 33,673 Itemized 22 11,969 Deductions Taxable Income 45 18,792 Self 9 816 Employment Tax Child Care 8 681 Credit Earned Income 3 976 Credit Total Tax (b) 45 3,322 (a) Other income includes rental income (loss), partnership income nonemployee compensation, and ordinary loss reported on Form 4797. (b) Total tax equals tax computed on taxable income less tax credits plus tax credits plus self-employment tax. (c) Refund of Earned Income Credit. TABLE 3 Potential Audit Adjustments Identified in IRS Audit Notice for Taxpayers with and without Representation during the IRS Office Audit Without With Representation Representation # of Average # of Average Returns Amount Returns Amount Schedule C 13 13,614 23 13,529 Car and Truck 5 4,433 11 6,141 Depreciation 4 4,174 9 3,394 Other Expenses 7 6,063 6 5,897 Travel, Meals, and 5 4,149 7 4,378 Entertainment Cost of Goods Sold 4 5,842 5 4,231 Phone/Utilities 1 1,456 7 1,470 Legal and Prof. Fees 2 1,628 5 1,353 Office Exp/Supplies 3 4,079 4 4,266 Repairs 3 5,544 4 1,751 Office in Home 1 3,761 4 3,167 Advertising 2 1,001 2 5,941 Interest 1 2,846 3 2,610 Rent 1 2,200 3 3,337 Wages/Labor 0 0 4 10,013 Bad debts 1 1,200 1 348 Employee Benefits 1 4,346 0 0 Adjustments for AGI 0 0 1 1,155 Itemized Deductions 1 8,709 8 7,555 Employee Business 1 3,032 4 8,931 Exp Charitable 0 0 3 5,160 Contributions Medical 0 0 2 4,811 Interest 1 5,677 0 0 Dependents 3 3,233 7 8,214 Child Care Credit 0 0 5 831 Earned Income Credit 1 508 2 1,210 SE Tax 0 0 1 6,706 Total tax 14 3,273 31 3,730 adjustment (b) Total # of Average Returns Amount t-statistic (a) p-value Schedule C 36 13,560 0.79 0.432 Car and Truck 16 5,607 Depreciation 13 3,634 Other Expenses 13 5,986 Travel, Meals, and 12 4,282 Entertainment Cost of Goods Sold 9 4,947 Phone/Utilities 8 1,468 Legal and Prof. Fees 7 1,471 Office Exp/Supplies 7 4,186 Repairs 7 3,376 Office in Home 5 3,285 Advertising 4 3,471 Interest 4 2,669 Rent 4 3,109 Wages/Labor 4 10,013 Bad debts 2 774 Employee Benefits 1 4,346 Adjustments for AGI 1 1,155 Itemized Deductions 9 7,683 -1.03 0.311 Employee Business 5 7,751 Exp Charitable 3 5,160 Contributions Medical 2 4,811 Interest 1 5,677 Dependents 10 6,720 -1.08 0.287 Child Care Credit 5 831 Earned Income Credit 3 976 SE Tax 1 6,706 Total tax 45 3,588 -0.52 0.602 adjustment (b) (a) t-statistic based on all returns in each group. (b) Total tax adjustment equals difference between "total tax computed on taxable income after potential adjustments" and "total tax on return as originally filed plus adjusted tax credits and self-employment tax." TABLE 4 Actual Audit Adjustments Agreed to with IRS for Taxpayers with and without Representation during the IRS Office Audit Without With Representation Representation # of Average # of Average Returns Amount Returns Amount Gross Income 1 5,453 6 -1,602 Schedule C 12 6,581 19 3,607 Car and Truck 3 1,697 12 2,460 Depreciation 2 1,602 6 275 Other Expenses 5 5,916 7 1,161 Travel, Meals, and 4 2,860 5 410 Entertainment COGS 2 2,930 3 2,299 Phone/Utilities 0 0 1 387 Legal and Prof. Fees 2 1,121 0 0 Office Exp/Supplies 2 1,960 5 1,074 Repairs 2 3,126 4 1,582 Office in Home 1 3,761 2 1,771 Advertising 1 2,000 0 0 Interest 1 34 3 -62 Rent 1 40 0 0 Wages/Labor 1 4,346 1 4,515 Bad Debts 1 1,200 1 348 Employee Benefits 0 0 0 0 Adjustments for AGI 0 0 2 44 Itemized Deductions 1 5,677 8 4,181 Employee Business 0 0 2 13,745 Exp Charitable 0 0 3 1,288 Contributions Medical 0 0 3 699 Interest 1 5,677 0 0 Dependents 1 4,900 6 3,525 Child Care Credit 0 0 1 1,008 Earned Income Credit 0 0 0 0 SE Tax (b) 0 0 1 5,568 Total Tax 14 1,543 31 780 Deficiency (c) Total # of Average Returns Amount t-statistic (a) p-value Gross Income 7 -594 Schedule C 31 4,758 3.38 0.049 Car and Truck 15 2,307 Depreciation 8 606 Other Expenses 12 3,143 Travel, Meals, and 9 1,499 Entertainment COGS 5 2,551 Phone/Utilities 1 387 Legal and Prof. Fees 2 1,121 Office Exp/Supplies 7 1,327 Repairs 6 2,096 Office in Home 3 2,434 Advertising 1 2,000 Interest 4 -38 Rent 1 40 Wages/Labor 2 4,431 Bad Debts 2 774 Employee Benefits 0 0 Adjustments for AGI 2 44 Itemized Deductions 9 4,348 1.24 0.268 Employee Business 2 13,745 Exp Charitable 3 1,288 Contributions Medical 3 699 Interest 1 5,677 Dependents 7 3,721 0.88 0.331 Child Care Credit 1 1,008 Earned Income Credit 0 0 SE Tax (b) 1 5,568 Total Tax 45 1,031 2.98 0.071 Deficiency (c) (a) t-statistic based on all returns in each group. (b) An additional 10 returns required self-employment tax adjustments due to changes in Schedule C net income. (c) Total tax deficiency equals tax due on agreed to IRS audit report. TABLE 5 Descriptive Statistics for Regression Variables (n = 45) Variable (a) Mean Median Std. Dev. Min. Max. DEFIC 1,031 360 1,565 -1,204 6,358 PDEFIC 3,588 2,710 2,677 262 12,339 REP 0.69 1.00 0.47 0.00 1.00 PAID 0.18 0.00 0.39 0.00 1.00 SCHC 0.82 1.00 0.39 0.00 1.00 ADJ DEFIC 762 315 1,733 -3,996 6,358 ADJ PDEFIC 3,321 2,638 2,679 0.00 12,338 CHANGE 0.70 0.85 0.44 -0.47 1.70 ADJ CHANGE 0.76 0.89 0.51 -0.54 2.00 ADEF NO SE 658 315 1,088 -2,252 3,495 APDEF NO SE 2,473 2,030 1,787 0.00 8,166 ADJUST 4,368 2,079 6,857 -8,042 24,710 PADJUST 13,686 11,419 9,413 0 47,224 (a) See Tables 6, 7, 8, and 9 for definition of variables. TABLE 6 Impact of Taxpayer Representation during Audit on Agreed Upon Audit Deficiency Panel A: Representation Choice Equation Model: [REP.sub.i] = [[alpha].sub.0] + [[alpha].sub.1][PDEFIC.sub.i] + [[alpha].sub.2][PAID.sub.i] + [[alpha].sub.3][ORGSCHC.sub.i] + [[alpha].sub.4][ERR.sub.i] n = 45; Chi-Square Statistic = 8.13 Intercept PDEFIC PAID ORGSCHC ERR 0.946 0.001 0.783 0.001 -2.263 Chi-square 1.93 0.27 0.53 4.78 2.31 p-value 0.165 0.600 0.465 0.028 0.128 Panel B: Impact of Taxpayer Representation Equation Model: DEFIC = [[alpha].sub.0] + [[alpha].sub.1]PDEFIC + [[alpha].sub.2]REP + [[alpha].sub.3]PAID + [[alpha].sub.4]SCHC + [[alpha].sub.5][lambda] n = 45; F-Statistic = 3.70 PDEFIC REP PAID SCHC Intercept (+) (-) (?) (+) 1598.62 0.319 -1328.58 -395.74 -1035.22 t-statistic (1.99) (3.61) (-4.84) (-0.66) (-1.61) p-value 0.053 0.001 0.001 0.511 0.114 [lambda] Adj. [R.sup.2] -31.25 .235 t-statistic (-0.58) p-value 0.566 REP = 1 if tax payer represented during office audit, 0 otherwise; PDEFIC = amount of potential tax deficiency calculated based on audit items identified in initial IRS audit notice; PAID = 1 if original return signed by paid preparer, 0 otherwise; ORGSCHC = amount of income or loss originally reported on Schedule C; ERR = 1 if error on return as originally filed, 0 otherwise; DEFIC = amount of final tax deficiency (refund) reduced by tax effect of gross income adjustments and errors on the return as originally filed; and PDEFIC = amount of potential tax deficiency calculated based on audit items identified in initial IRS audit notice reduced by tax effect of errors on the return as originally filed. TABLE 7 Impact of Taxpayer Representation during Audit on the Change between the Potential and Final Tax Deficiency Model: CHANGE = [[alpha].sub.0] + [[alpha].sub.1]REP + [[alpha].sub.2]PAID + [[alpha].sub.3]SCHC + [[alpha].sub.4] [lambda] n = 45; F-Statistic = 3.58 REP PAID SCHC Intercept (+) (?) (-) [lambda] 0.15 0.403 0.147 0.341 0.0150 t-statistic (0.58) (4.48) (0.75) (1.67) (0.86) p-value 0.567 0.001 0.455 0.100 0.392 Adj. [R.sup.2] .087 t-statistic p-value CHANGE = (PDEFIC -DEFIC)/PDEFIC See Table 6 for other Nariahle definitions. TABLE 8 Impact of Taxpayer Representation during Audit on Agreed Upon Audit Deficiency with Adjustments for Self-Employment Tax Model: ADEF NO SE = [[alpha].sub.0] + [[alpha].sub.1]APDEF NO SE + [[alpha].sub.2]REP + [[alpha].sub.3]PAID + [[alpha].sub.4]SCHC + [[alpha].sub.5][lambda] n = 45; F-Statistic = 3.07 APDEF NO SE REP PAID SCHC Intercept (+) (-) (+) (+) 1352.79 0.18 -814.58 -338.59 -587.82 t-statistic (2.49) (2.04) (-4.46) (0.84) (-0.40) p-value 0.017 0.048 0.001 0.407 0.169 [lambda] adj. [R.sup.2] -16.56 0.106 t-statistic (0.55) p-value 0.583 ADEF NO SE = amount of final tax deficiency (refund) without self-employment tax reduced by tax effect of gross income adjustments and errors on the return as originally filed; and APDEF NO SE = amount of potential tax deficiency calculated based on audit items identified in initial IRS audit notice without self-employment tax reduced by tax effect of errors on the return as originally filed. See Table 6 for other variable definitions. TABLE 9 Impact of Taxpayer Representation during Audit on the Final Adjustment to Taxable Income Model: ADJUST = [[alpha].sub.0] + [[alpha].sub.1]PADJUST + [[alpha].sub.2]REP + [[alpha].sub.3]PAID + [[alpha].sub.4]SCHC + [[alpha].sub.5][lambda] n = 45; F-Statistic = 5.94 PADJUST REP PAID SCHC Intercept (+) (-) (+) (+) 6472.53 0.426 -4792.08 -1816.5 -4752.14 t-statistic (2.13) (4.66) (-4.84) (-0.84) (-2.06) p-value 0.039 0.001 0.001 0.405 0.045 [lambda] Adj. [R.sup.2] -211.63 .359 t-statistic (-1.04) p-value 0.304 ADJUST = amount of final adjustments to taxable income reduced by gross income adjustments and errors on the return as originally filed; and PADJUST = amount of potential adjustments to taxable income based on audit items identified in initial IRS audit notice reduced by gross income adjustments and errors on the return as originally filed. See Table 6 for other variable definitions.
The authors gratefully acknowledge the helpful comments of the editor, two anonymous reviewers, and workshop participants at the 2002 AAA Annual Meeting in San Antonio. We are grateful for the financial support from James Madison University College of Business.
(1) Weiner (2001) suggests that a face-to-face meeting with an IRS auditor "can turn into a chess game of strategy and psychology, which is why many taxpayers may want to seek professional help and skip attending the audit." Other articles suggest potential benefits include the limitation of the audit to the items originally identified (Anderson 1999; Chaffin and Busby 1998) and a reduced anxiety level on the part of the taxpayer (Sommers 1995; Szabo 1994).
(2) The IRS discontinued the TCMP program in 1998. The last TCMP audits of individual income tax returns were in 1988 (Smith 2001).
(3) Audit methods include examinations by mail (i.e., "mail audits") or through an in-person interview and review of records (i.e., "office audits" or "field audits"). Interviews held at the IRS office are referred to as "office audits." Interviews held at the taxpayer's home, place of business, or accountant's office are referred to as "field audits" (Tax Analysts 1997).
(4) The TCMP survey in 1988 audited approximately 54,000 individual returns (Dolan 1993). In 1995, the IRS audited 1,941,546 individual returns. The three audit methods used included mail audits (1,179,696 returns), office audits (509,420 returns) and field audits (252,430 returns) (IRS 1996).
(5) See the "Sample" section and the Appendix for further discussion of the Federal Student Tax Clinic.
(6) A second potential benefit of representation is to limit the extension of the audit into other areas. Since only 12 audits included in the sample were extended, the analysis of this potential benefit is not included.
(7) For accounting studies using the Heckman procedure see Schaefer and Zimmer (1998) and Shehata (1991). Maddala (1991) analyzes self-selection models in accounting studies.
(8) The clinic closed in 1997 when the Director assumed additional professional responsibilities.
(9) Usually a taxpayer's decision regarding representation during an IRS audit involves a cost benefit analysis since professional representation is generally fee based. Since the Tax Clinic provided free representation, this study provides information regarding the potential tax savings from representation for assisting taxpayers in a cost benefit analysis.
(10) Although the Clinic represented taxpayers for the first time during the 1993-1994 academic year, the files for the two years, 1993-1995, were destroyed before this study began.
(11) All of the taxpayers included in the sample agreed to the audit adjustments determined at the office audit. None of the cases required an Appeals Conference.
(12) Both the "with representation" and "without representation" groups include tax returns with obvious errors. Taxpayers in the "with representation" group allowed the student to disclose the error to the IRS agent.
(13) Of the eight returns signed by paid preparers, H&R Block prepared two, small CPA firms prepared four, and self-employed individuals prepared two returns.
(14) The median Schedule C net income for taxpayers "without representation" is -$7,221 and for taxpayers "with representation" is--$6,467 (Table 2). All three taxpayers reporting Schedule C net income >$25,000 are in the "with representation" group. This explains the much higher average Schedule C net income for such group.
(15) Other Schedule C items identified for audit include legal and professional fees, office expenses or supplies, and repairs (seven returns each), office-in-home deduction (five returns), advertising, interest, and rent (four returns each), bad debts (two returns) and employee benefits (one return).
(16) Charitable contributions on three returns, medical expenses on two returns, and interest expense on one return were also identified as areas of concern.
(17) The largest average adjustment was to wage/labor expense with an average adjustment of $4,431 to two returns. Other frequent Schedule C expense adjustments include other expenses on 12 returns, travel, meals, and entertainment on nine returns, depreciation on eight returns, office expense or supplies on seven returns, repairs on six returns, and cost of goods sold on five returns.
(18) The largest adjustments were made to employee business expenses with an average adjustment of $13,745 on two returns. Charitable contributions and medical expenses were each adjusted on three returns. Interest expense was reduced on one return.
(19) Of the ten returns requiring adjustments to SE tax, seven were represented (average adjustment $1,010) and three were not represented (average adjustment $1,604). Schedule C audit adjustments resulted in six taxpayers going from an original Schedule C loss to Schedule C net income in excess of $400 (four were represented and two were not represented).
(20) Using a DFFITS > 1, we identified two potential outliers for Equation (5). Excluding the outliers, the coefficient on REP is -1,406.09 (t = -5.52).
(21) No potential outliers were identified for Equation (6) (DFFITS > 1).
(22) Since only one final adjustment involved a tax credit, it is appropriate that this sensitivity test focuses solely on the adjustments to taxable income.
(23) All taxpayers in the Dallas IRS District selected for office audit received information regarding the Tax Clinic. Only a small portion of those taxpayers contacted the clinic for assistance.
Anderson, C. 1999. Self-employed, big itemized deductions head list of IRS audit targets. The Associated Press Newswires (April 7).
Chaffin, R. E. and (3. Busby. 1998. Dealing with Internal Revenue Service income tax audits. The National Public Accountant 43 (2): 47-52.
Dolan, M. 1993. IRS' plans to measure tax compliance can be improved. Tax Notes Today 93: TNT 76-24.
Dubin, J., M. Graetz, M. Udell, and L. Wilde. 1992. The demand for tax return preparation services. Review of Economics & Statistics 74(1): 75-82.
Erard, B. 1993. Taxation with representation: An analysis of the role of tax practitioners in tax compliance. Journal of Public Economics 52: 163-197.
Evanson, D. 1995. 10 tips for surviving an audit. Entrepreneur 23 (October): 53-55.
Feinstein, J. 1991. An econometric analysis of income tax evasion and its detection. RAND Journal of Economics 22 (1): 14-35.
Heckman, J. 1976. The common structure of statistical models of truncation, sample selection, and limited dependent variables and a simple estimator for such models. Annals of Economic and Social Measurement 5 (Fall): 475-492.
--. 1979. Sample selection bias as a specification error. Econometrica 47: 153-161.
Hicks, (3. 1991. Tips for handling IRS audits. Practical Accountant 24 (1): 63-65.
Hite, P., and J. Hasseldine. 2003. Tax practitioner credentials and the incidence of IRS audit adjustments. Accounting Horizons 17 (March): 1-14.
Internal Revenue Service (IRS). 1996. Internal Revenue Service Data Book 1996. Washington, D.C.: Government Printing Office.
Klepper, S., M. Mazur, and D. Nagin. 1991. Expert intermediaries and legal compliance: The case of tax preparers. Journal of Law and Economics 34: 205-299.
Lee, L. 1982. Some approaches to the correction of selectivity bias. Econometrica 49: 355-372.
Long, J., and S. Caudill. 1987. The usage and benefits of paid tax return preparation. National Tax Journal 40 (March): 35-46.
Maddala, G. S. 1983. Limited Dependent Variables in Econometrics. New York, NY: Cambridge University Press.
--. 1991. A perspective on the use of limited dependent variable models in accounting. The Accounting Review 66 (October): 788-807.
Madeo, S., A. Schepanski, and W. Uecker. 1987. Modeling judgments of taxpayer compliance. Accounting Review 62 (2): 323-342.
Price, J. 1995. Interactive learning through the use of student federal tax clinics. Journal of Accounting Education 13 (4): 413-423.
Roberts, M. 1995. The influence of contextual factors on IRS agents' assessment of taxpayer negli-gence. The Journal of the American Taxation Association 17 (Supplement): 62-77.
Schaefer, J., and M. Zimmer. 1998. Liability insurance decisions and premium charges among CPA firms: An empirical analysis. Journal of Applied Business Research 14 (1): 69-82.
Shehata, M. 1991. Self-selection bias and the economic consequences of accounting regulation: An application of two-stage switching regression to SFAS No. 2. The Accounting Review 66 (4): 768-787.
Smith, D. 2001. The IRS market segment specialization program. The CPA Journal 71 (2): 22-28.
Sommers, R. 1995. How to deal with IRS demands. San Francisco Examiner (March 28).
Szabo, J. 1994. How to survive an IRS audit. Nation's Business 82 (April): 4243.
Tax Analysts. 1997. Fact sheet explains IRS audit process. Worldwide Tax Daily 97: TNI 41-28.
Weiner, L. 2001. The odds of an IRS audit are still in the taxpayer's favor, but don't bet the house. U.S. News & World Report (April 20).
Nancy B. Nichols is an Associate Professor at James Madison University and John Ellis Price is a Professor and Vice Provost at the University of North Texas Dallas.
Submitted: December 2001
Accepted: September 2003
|Printer friendly Cite/link Email Feedback|
|Author:||Nichols, Nancy B.; Price, John Ellis|
|Publication:||Journal of the American Taxation Association|
|Date:||Mar 22, 2004|
|Previous Article:||The influence of biased tax research memoranda on supervisors' initial judgments in the review process.|
|Next Article:||Local sales tax competition and the effect on county governments' tax rates and tax bases.|