The views expressed in this testimony are those of the author and do not necessarily reflect the views of the trustees and employees of the Urban Institute.
Leonard E. Burman is a Senior Fellow at the Urban Institute; Co-director of the Urban-Brookings Tax Policy Center; and Research Professor, Georgetown Public Policy Institute
Mr. Chairman, Mr. Spratt, and distinguished Members of the Committee:
Thank you for inviting me to share my views on waste, fraud, and abuse in the tax system. The views I express are mine alone and should not be attributed to any of the organizations with which I am affiliated.
I applaud the Committee's efforts to rein in waste, fraud, and abuse, and its recognition that fraud is not only a problem on the spending side of the ledger, but also appears on the tax side. Indeed, there is overwhelming evidence that tax fraud is epidemic, and the IRS has already identified tax underpayments that dwarf any amounts the distinguished Inspectors General who testified on the first panel are likely to unearth in their examination of cash transfer programs. The main issue is whether the IRS can deploy its resources to effectively collect a larger share of taxpayers' legal obligations.
Mr. Chairman, in an earlier hearing on the same subject, you noted that "wasteful Washington spending is not a Republican problem or a Democrat problem." I will argue that the same may be said for tax evasion. Whether you see a larger role for government or favor smaller government and lower taxes, tax evasion undermines your objectives.
I'd like to start with some startling statistics on tax evasion. I will then turn to the argument for trying to stem it, and discuss why IRS efforts so far have been disappointing. I will then focus on specific issues related to the earned income tax credit (EITC), since that program has been the center of a disproportionate amount of IRS compliance activity and current compliance plans raise many issues.
I. The Scope of the Tax Evasion Problem
Former IRS Commissioner Charles Rossotti (2002) estimated that in a given year, the IRS assesses almost $30 billion of taxes that it will never collect. This is not theoretical tax evasion. The $30 billion represents underpayments of tax that the IRS has identified, but cannot collect because its staff is spread so thin. Rossotti estimated that it would cost about $2.2 billion to collect that money. If you accept that estimate, the IRS could net almost $28 billion from tax fraud and errors that are identified and ripe for collection.
Assuming that the amount grows with GDP, collecting on assessments would, over the next decade, cover the entire cost of the new prescription drug benefit under Medicare (although not the superfluous new savings accounts in the House version of the bill). It is more than the entire cost of the Jobs and Growth Tax Relief Reconciliation Act of 2003 as scored by the JCT (although not enough to finance the extension of the myriad expiring provisions). It is serious money.
But it is tiny compared with the entire "tax gap"the IRS's estimate of total taxes due but not collected. The IRS estimated that $232 billion in taxes were due in 1998, but never collected. (See Figure 1.) These estimates are highly uncertain because the IRS stopped systematically measuring tax compliance for all but working poor people after 1988, but it suggests that tax compliance is a huge problem, and it has been growing. According to Commissioner Rossotti, "Despite significant improvements in the management of the IRS, the health of the federal tax administration system is on a serious long-term downtrend. This is systematically undermining one of the most important foundations of the American economy."
Why is the gap growing? To begin with, the number of tax returns has been growing much faster than the IRS staff. This has occurred for several reasons. There are more head of household and single returns and fewer married filing joint returns because couples are marrying later, if at all, and the divorce rate is rising. Also, many more children are filing tax returns. (Plumley and Steuerle, forthcoming)
Moreover, after steady growth in compliance resources through the 1980s, IRS staff dedicated to compliance and enforcement plummeted in the 1990s. Between FY 1992 and 2001, the IRS workload increased by 16 percent while its staff declined by 16 percent. Field compliance personnel fell by 28 percentmore than 8,000 FTEsbetween FY 1992 and 2002.
The effect on examinations is even more striking. According to the Internal Revenue Service (2001), the number of field examiners fell by almost two-third between 1997 and 2000. The number of collection cases closed fell by nearly half over the same interval. The number of criminal tax cases not related to income from illegal activities fell by more than two-thirds, from 1,498 in 1997 to 409 in 2000.
A large part of the problem, according to the Commissioner, is budgets with "unrealistic assumptions about such items as pay raises, inflation and other mandates, including specific mailing and notification requirements." In the late 1990s, a key factor was the taxpayer bill of rights, which required the IRS to answer its telephones and focus its efforts on "customer service." The better service, while surely welcome, came at the expense of audit activity. This decade, Congress has twice mandated that the IRS interrupt its ordinary operations to mail out springtime checks to most taxpayersadvance payments on the low-end tax rate cut in 2001 and on the child credit increase in 2003. Without a supplemental appropriation to pay for additional staff, the staff managing these huge mailings must come out of existing employees, typically compliance staff.
Finally, the opportunities for evasion have been growing. While the overall number of returns grew by 16 percent, the number of tax returns reporting more than $100,000 of income grew by 342 percent. These people who face the highest marginal tax rates have the most to gain from tax evasion, and the most opportunities to engage in it. Commissioner Rossotti reported that "enormous amounts of money ... flow through 'pass-through entities'such as partnerships, trusts, and S-corporations," which are ideally suited to hiding income. In tax year 2000, pass-throughs accounted for 4.8 million tax returns with over $660 billion of income.
Commissioner Everson has taken up where Mr. Rossotti left off calling for a renewed focus on enforcement: "...(T)he IRS is committed to ensuring everyone pays his or her fair share, including those who have the resources to move money offshore or engage in abusive schemes or shelters. We must focus our efforts on achieving greater corporate accountability and ensure that high-end taxpayers fulfill their responsibilities. Honest taxpayers should not bear the burden of others who skirt their responsibility." (May 20, 2003)
II. Why Tax Evasion Matters
Tax evasion undermines the tax system in numerous ways. It is unfair. It costs revenues that could be used to make the tax system better, pay down the debt, or provide additional government services. It wastes resourcesi.e., hampers economic growth. And it feeds on itself, reducing respect for the integrity of the tax system and leading to more cheating.
Tax evasion is fundamentally unfair: unless they are caught, cheaters pay less tax than their law-abiding neighbors. But Figure 1 shows that getting caught is highly unlikely. Of the $282 billion of taxes not paid on time in 1998, only about $50 billion was eventually collected, and about half of that was voluntarily remitted by tardy taxpayers. Thus, the IRS only collected about 10 percent of underpaid tax through enforcement activity. The individual audit rate has fallen from 2.15 percent in 1978 to 0.58 percent in 2001. Almost 4 percent of individuals with business income were audited in 1995, because they were known to be comparatively noncompliant. That rate fell in halfto 2 percentin 2001. In 1993, more than 3 percent of corporate income tax returns were audited. By 2001, despite a well publicized epidemic of questionable and illegal corporate tax shelters in the late 1990s, less than 1 percent of corporations were audited. (Indeed, that statistic makes one suspect that the corporate tax shelter boom was fed by the IRS's apparent indifference.)
Tax evasion undermines both Republicans' and Democrats' notion of a good government. The lost tax revenue inevitably means higher taxes on law-abiding citizens, less government services, or both. If we could close half of the tax gap, the IRS could raise close to $150 billion on tax year 2003 returns (assuming that the tax gap grows at the same rate as GDP). Over the decade, collections would increase by something like $1.7 trillionthe entire cost of the 2001 tax cut as scored by the JCT. With that money, we could (1) eliminate more than two-thirds of the public debt according to CBO projections, or (2) cut income tax rates across the board by more than 10 percent, or (3) provide health care for the uninsured and a generous prescription drug benefit under Medicare, or (4) fully fund the transition to individual accounts under Social Security. I don't mean to endorse any of these policy proposals (my four kids, however, think that paying down the debt is a very good idea), but they illustrate that this huge hole in our income tax is keeping us from getting the government any of us wants.
Second, some argue that tax evasion might be okay because it lowers tax burdens. That argument is obviously false in the aggregatetax evasion simply reallocates tax burdens from noncompliant to compliant taxpayers. But, it also is a uniquely inefficient way to cut taxes. Companies alter their business practices to hide income from the IRS, as Bob McIntyre explained in his testimony in the earlier hearing. A good tax system interferes as little as possible in businesses' and individuals' decisions, but abusive tax shelters virtually always involve substantial distortions. Some companies now view their tax departments as profit centersthat is, they make money by hiding it from the IRS rather than by producing more and better products. Individuals make investment decisions not based on where they will earn the highest pre-tax rate of return, but where they can make the most money after subtracting taxes, promoters' fees, and legal fees. Thus, money is not going to where it can produce the most return, but to where it can produce the most tax savings. Moreover, the fees paid to tax shelter promoters, unethical lawyers, financial wizards, etc. are a pure waste of resources. Most of these intermediaries could be doing productive work if inadequate enforcement did not make tax evasion so lucrative.
In contrast, if the IRS stemmed tax evasion and used the money to pay for debt reduction or tax rate cuts, the economy would surely grow faster. First, there would be fewer distortions from the tax shelter arrangements. Second, debt reduction would reduce government crowding-out of private investment: that is, it would lower interest rates, making capital less costly for businesses. Or tax rate reductions would reduce the incentive to avoid tax by working less, saving less, or engaging in legal or illegal tax shelters.
Finally, tax evasion can create a vicious cycle of growing disrespect for the tax system, which undermines voluntary compliance. The IRS has some evidence that this is happening now from Roper surveys they commissioned in 1999 and 2001. In 1999, 87 percent of respondents said that cheating on taxes was unacceptable; in 2001, only 76 percent. In 1999, 96 percent of respondents agreed that it is everyone's duty to pay their fair share of taxes; in 2001, 91 percent. And, in 2001, respondents were skeptical that cheaters would be caught. A plurality of respondents (37 percent) said that cheaters were less likely to be audited in 2001 than in the past. Only one in three thought the odds of detection had increased.
What can be done about the epidemic of tax evasion? Two things can deter those who are inclined to cheat: a high probability of detection and a high penalty if caught. In this regard, the first order of business ought to be to make sure that, barring extenuating circumstances, everyone who is caught underpaying their tax is made to pay what they owe. Correcting the alarming statistics reported by Commissioner Rossotti should be the first order of business: (1) 60 percent of identified tax debts are not collected; (2) 75 percent of identified nonfilers are not pursued; (3) 79 percent of taxpayers who use known abusive devices to avoid tax are not pursued; (4) 78 percent of taxpayers identified through document matching programs are not pursued; and (5) 56 percent of noncompliant taxpayers with incomes over $100,000 get off scot-free.
Given the large amounts of money involved, it should be possible to solve this problem without costing anything. One option would be to raise the penalties and/or interest for taxpayers once they are identified as noncompliant. The clock on these excess penalties could stop for nonfrivolous legal challenges, but taxpayers who decided to try a rope-a-dope strategy with the IRS would find it unprofitable. A second option would be to allow the IRS to divert a fraction of the revenues it collects from enforcement action into a trust fund that could be tapped to pay for other enforcement activities. (Since money is fungible, this strategy only works if the Congress does not cut the rest of the IRS's budget at the same time that they are tapping the trust fund to finance enhanced enforcement.)
The IRS is taking steps to raise the probability of detection, which is good, both by expanding its document-matching program and increasing the number of examiners (although the latter might be derailed by the rebate program and other competing demands for scarce resources). It is well known that compliance is much higher when the IRS has an independent source of verification.
There is, of course, a risk that compliance activity could go too far. Arguably, that is why the Congress terminated the taxpayer compliance measurement program (TCMP), which involved highly intrusive random audits. Arguably, the taxpayer bill of rights was aimed at redressing a system that favored the tax collector too much at the expense of law-abiding citizens. Unfortunately, the resources to protect taxpayer rights came out of the resources used for enforcement, so the balance may have shifted too far in the other direction.
Given scarce resources, it is important that the IRS targets them where the payoff is greatest. The TCMP was designed to allow that, but was terminated because it was too intrusive on lawful taxpayers. The IRS is now engaging in a new audit strategy called the National Research Program, which will adjust audit rates based on the yield from less intrusive auditsmany of which will not involve any taxpayer contact unless a problem is discovered. This is clearly a promising approach to balancing taxpayer rights with the imperative to improve collections.
IV. The EITC Compliance Program
Amid all this enlightened activity by the IRS, one example stands out as a misallocation of resources and a failure to balance the rights of taxpayers against the need for enforcementthe EITC compliance initiative. EITC noncompliance appears to be a problem. The IRS estimates that somewhere between 27 and 31 percent of earned income tax credits were issued erroneously in 1999, either because of taxpayer confusion or fraud. They estimate the EITC compliance gap at $7.8 billion in 1998 (See Figure 1), about 0.5 percent of revenues and about 2.8 percent of the total tax gap. But EITC enforcement accounts for 3.8 percent of total enforcement budget in 2003. Indeed, the IRS has requested a 68.5 percent increase in its EITC enforcement budget, while increasing other enforcement by only 3.3 percent. In fact, the increase in EITC enforcement would account for 45 percent of all new compliance dollars. (Internal Revenue Service 2003) On its face, this seems like an inefficient way to spend scarce compliance resources.
The apparently high rates of noncompliance are troubling, but it is necessary to put them in context. Indeed, it is likely that much EITC noncompliance reflects compliance problems that are endemic to the entire income tax. If that is true, then targeting compliance activity at EITC participants alone may not be the most effective use of IRS resources.
The IRS's current compliance initiative, which will for the first time since 1988 collect information about other than low-income taxpayers, may help resolve some of these issues.
A. EITC Noncompliance in Perspective
Two Treasury economists (Holtzblatt and McCubbin, forthcoming) used data from the IRS's 1999 EITC compliance study to draw out some comparisons between EITC compliance and compliance with other tax provisions that require some definition of an "eligible child." Of children claimed for both the EITC and the dependent exemption (97 percent of "qualifying children" claimed for EITC were also claimed as dependents), more tax filers failed the test for dependency status (for the exemption) than the test for qualifying child (for the EITC). It is striking that one-third of children were claimed in error for the dependent exemption, the EITC, or both. However, while six percent qualified as a dependent but not as an EITC-qualifying child, 11 percent (almost twice as many) were eligible for qualifying child status but not for a dependent exemption. That is, there were more children claimed in error as a dependent for purposes of the exemption than as an EITC-qualifying child. An additional 17 percent of children were ineligible for both.
While this level of noncompliance with both provisions is troubling, the statistics only apply to low-income tax filers who were audited as part of the EITC compliance program. These statistics raise the question of whether higher income people have the same propensity to claim dependent exemptions for children who do not qualify. There is some historical evidence (from 1986) that people are prone to cheat with dependent exemptions when they think they can get away with it. In that year, seven million children disappeared when the IRS started requiring reporting of Social Security numbers to verify dependent exemptions. (Graetz 1997)
The ineluctable conclusion is that there are likely to be many dependents claimed incorrectly at all income levelsnot just among the poor. Thus, the relevant policy response would be to study compliance in the entire taxpaying population, not just among low-income people.
Another fascinating set of statistics drawn from the EITC compliance data relates to homemade marriage penalty relief. In 1999, 0.5 million people filed as head of household when they were actually married and living together, possibly to avoid EITC marriage penalties. Another 0.4 million filed as single when they should have claimed another unspecified status. Three-quarters of a million filed as head of household when they lived apart from their spouse for at least part of the year, but were still married and should have filed as married filing joint or married filing separate. The obvious question is the extent to which this type of roll-your-own marriage penalty relief occurs among higher-income taxpayers (who often have a far greater incentive to misstate their filing status).
Some EITC recipients with income in or beyond the phase out range of the credit underreported their income and thus increased their tax refund. Half of the unreported income was from self-employment, consistent with ancient evidence from the TCMP that self-employment income is an area of rampant evasion. Again, while the noncompliance among EITC recipients is troubling, there is no reason to think that it is any worse than exists among the taxpaying public generally.
B. How Much Noncompliance is Intentional?
A key question is how much of EITC noncompliance is intentional, and how much inadvertent. If intentional tax evasion is rampant, then the solution is to ramp up enforcement. However, if a major source of noncompliance comes from taxpayer confusion, then education, assistance in preparing tax returns, and simplification of the tax law would be better-targeted policy responses.
Janet McCubbin (2000) reported that at least 28 percent of qualifying child errors are systematic, and thus intentional attempts to overclaim the EITC. Some of the remaining 72 percent may be influenced by other elements of code, such as the dependent exemption. How many of the 72 percent are simply confused tax filers?
There's certainly evidence of confusion. As Holtzblatt and McCubbin report, the IRS mailed notices to 194,000 taxpayers who appeared to be eligible for the EITC based on income and the presence of dependent children reported on their 1998 return. About one-third responded requesting the credit. The IRS also sent 680,000 notices to low-wage single filers notifying them that they appeared to be eligible. About 45 percent of them responded requesting the credit. The people who only requested the credit after being notified by the IRS almost surely underclaimed the credit unintentionally. Some of those who overclaimed are surely similarly uninformed.
It is also worth mentioning that not all of the EITC tax gap would be collected if EITC enforcement were perfect. In many cases where one person wrongly claims the EITC as the eligible custodial adult, another person might be eligible for an EITC, albeit possibly a smaller one. We have no evidence on whether someone else is eligible for the EITC when a person is found to be disqualified, although this is clearly an important measure of the costs of noncompliance to the Treasury. In addition, because of flaws in the design of the compliance studies, it is possible that actual noncompliance is much less than the IRS estimates. (Greenstein 2003b)
C. Addressing EITC Noncompliance
As in other areas of the tax law, there is a trade-off between administration and compliance costs on the one hand and targeting, compliance, and participation on the other. The question for policy makers is how to strike the right balance. The IRS could audit every return, which would minimize noncompliance, but would maximize enforcement and compliance costs. At the other extreme, the IRS could make all low-earning families eligible for EITC, without regard to children, which would also reduce noncompliance, but at great cost in terms of tax revenues. In that context, one might argue that the current system does not do a bad job of balancing competing objectives.
The compliance problems with EITC may be viewed as comprising two parts, each of which has a specific policy implication: systemic problems and those specific to the EITC. There are errors and fraud that are endemic to the income tax, such as children claimed incorrectly, understated income, and incorrect filing status. The solution to that problem is system-wide enforcement, not a specific EITC compliance program. Indeed, targeting scarce enforcement resources on low-wage returns to catch systemic noncompliance would be a highly inefficient audit strategy, since so much more money is at stake on the high-income returns.
Certain errors are specific to the EITC. For example, a major factor in the 1999 data involves parents who violated the confusing AGI tie-breaker rule or were disqualified because of too much non-cash earned income (such as pensions, parsonage benefits, and the like). In these cases, Congress ultimately decided that the targeting rule was not worth the cost and the rules were simplified to reduce chances of inadvertent errors.
A similar example is the inconsistent definition of a child for different purposes. The Treasury has proposed rules to make the definitions more consistent and intuitive, and the Senate included them in the Relief for Working Families Tax Act Of 2003, but they have not yet been enacted. (Treasury 2002). Further simplifications would be possible, such as automatically allowing a dependent to be a qualifying child for EITC purposes so long as the other parent does not claim the child for the EITC. These simplifications all involve some cost in terms of tax revenues, but they would significantly reduce confusion for low-income working families who do not tend to think like tax lawyers.
Another promising approach is to enlist the help of those who prepare tax returns for lowincome people. Almost two-thirds of EITC returns are prepared by paid preparers. IRS statistics show that more competent preparersaccountants, lawyers, enrolled agents, major tax preparation firmsproduce returns with fewer errors than less competent preparers. Volunteer tax preparers have the lowest error rate, although the sample is too small to draw firm inference. It is at least possible that spending more time on tax returns reduces the likelihood of errors. It is also possible that differences in performance among preparers reflect self-selectionthat noncompliant taxpayers are more likely to seek the help of disreputable tax preparersbut this conjecture should be tested.
In 1999, the IRS initiated a large-scale outreach program aimed at tax return preparers who had recently prepared at least 100 EITC returns. During those visits, preparers (other than national firms, CPAs, lawyers, and enrolled agents) received one-on-one instruction from Revenue Agents on EITC compliance and preparers' due diligence responsibilities. Because most EITC claimants use paid preparers, such a strategy could prevent both unintentional and intentional errors on tax returns claiming the EITC. The value of this approach could be measured by comparing the accuracy of trained preparers with similar preparers who did not get training. However, no data are available yet and it is not clear that the IRS followed up. If not, they lost an important opportunity to improve compliance without adding extra burdens for low-income taxpayers.
The other tool to improve compliance is to strengthen EITC enforcement. The IRS is about to start a new pre-certification program for the EITC. This probably would improve compliance, but also could significantly reduce participation, and might not save the government much money. The cash assistance programs that you heard about this morning cost about as much to administer as the EITC, including both the administration and compliance costs and the revenues lost due to noncompliance, but EITC participation is much higher than participation in direct transfer programs. (Holtzblatt and McCubbin, forthcoming). So the result of the IRS's EITC compliance offensive may be less payments to low-income families, including many who are eligible but deterred by the new hurdles to participation, but little or no overall budget savings.
The proposed pre-certification program is supposed to be non-intrusive, but it is not clear how the IRS can accomplish that. How can they determine that the residency requirement is met in advance, especially for households that are highly mobile? Arguably, it is unfair to single out the EITC. Eligibility for other tax benefits, such as head of household status and the dependency exemption, also theoretically require extensive record keeping. Resolving filing status errors would require fairly intrusive tests, which again might be hard to certify in advance. The fear among those who care about the EITC is that the pre-certification strategy is tantamount to a 100 percent audit rate (in advance) for certain people who claim the EITC.
There are also real issues in subjecting EITC recipients to a pre-certification process that does not apply to any other tax filers. People do not need to pre-certify before taking a charitable deduction for a used car or clothing, even though there is ample evidence that these deductions are overstated. Sole proprietorships do not need to pre-certify that they are not hiding cash from the tax authority before claiming deductions for inventories, rent, and equipment, even though sole props are notoriously noncompliant. And so on.
In point of fact, the IRS's proposed strategy now is to select about 45,000 single fathers, grandparents, and other adults who claim to care for a qualifying child for the precertification process. Bob Greenstein (2003a) has documented the ways in which the precertification requirements create a Catch-22 for many grandparents and fathers who are lawfully eligible for the credit. For example, a grandparent who leaves her grandchild with a nonlicensed family daycare center cannot rely on an affidavit from the daycare provider or from a relative or neighbor to prove that the child lived with her for the year. Since most low-income people cannot afford expensive licensed daycare facilities, this means that many eligible people will not be able to prove eligibility to the IRS. Add to this the problems of establishing eligibility for people who are transient or have language problems and you have a recipe for excluding many eligible recipients.
At a minimum, we should check that pre-certification meets its objectives before subjecting 2 million or more taxpayers to it.
Noncompliance is a serious issue that undermines the tax system and carries a huge cost in terms of higher taxes on law-abiding citizens, fewer government services, and more government debt. The IRS is taking a number of important steps to improve tax compliance. However, the IRS's preoccupation with EITC recipients seems like a poor use of scarce audit resources, is likely to undermine the EITC program, and is unfair. It would be better to address the endemic problems in the income tax at all income levels. EITC compliance, and compliance in other areas, could also be improved by simplifying the tax law.
Graetz, Michael. 1997. The Decline [and Fall?] of the Income Tax, W. W. Norton and Company.
Greenstein, Robert. 2003a. "The New Procedures for the Earned Income Tax Credit," Center on Budget and Policy Priorities. http://www.cbpp.org/5-20-03eitc2.pdf.
Greenstein, Robert. 2003b. "What is the Magnitude of EITC Overpayments?" Center on Budget and Policy Priorities. http://www.centeronbudget.org/5-20-03eitc3.pdf
Holtzblatt, Janet and Janet McCubbin. Forthcoming. "Complicated Lives: Tax Administrative Issues Affecting Low-Income Filers," in Henry Aaron and Joel Slemrod, The Crisis in Tax Administration, Brookings Institution Press.
Internal Revenue Service. 2001. "IRS Strategic Plan: Fiscal Years 2000-2005," http://www.irs.gov/pub/irs-utl/irs_strategic_plan.pdf.
Internal Revenue Service. 2003. "Budget in Brief, Fiscal Year 2004," http://www.irs.gov/pub/irs-utl/budget-brief.pdf
McCubbin, Janet. 2000. "EITC Noncompliance: The Determinants of the Misreporting of Children," National Tax Journal, 53(4): 1135-1164.
Plumley, Alan H., and C. Eugene Steuerle. Forthcoming. "What Should the Ultimate Objective of the Internal Revenue Service Be? A Fresh Look from an Historical Perspective," in Henry Aaron and Joel Slemrod, The Crisis in Tax Administration, Brookings Institution Press.
Rossotti, Charles O. 2002. "Report to the IRS Oversight Board: Assessment of the IRS and the Tax System," Internal Revenue Service, September.
U.S. Department of the Treasury. 2002. "Proposal for a Uniform Definition of a Qualifying Child," http://www.ustreas.gov/press/releases/docs/child.pdf, April.