The Mysteries of Erroneous Refunds

I love my yearly tax refund. I know I’m not supposed to, but I do. Theorists argue that my refund represents an interest-free loan to the government and I am supposed to resent that for reasons grounded in the obsessive individualism of our culture. But even accepting the argument’s premise, the value to me of that yearly manifestation of my forced savings outweighs the value of the interest I would theoretically earn in my checking account. I don’t worry; I be happy.

I am not alone. In fiscal 2005 the IRS made lots of folks happy, refunding about $227.6 billion to individual taxpayers. [1] By almost any measure, the IRS does a great job of getting the right refund to the right taxpayer in a timely fashion. But errors happen. Worse, given the huge amounts involved, even small error rates add up. For example, even if the IRS had an error rate of just 1 percent, that would mean it issued more than $2.2 billion in erroneous refunds in 2005. That represents a lot of bridges to nowhere

No one knows the actual error rate. That’s part of the critique in a recent report from the Treasury Inspector General for Tax Administration: The IRS could do better in tracking and preventing erroneous refunds.[2] While TIGTA’s suggestions might decrease the slippage in the gears of IRS campus operations, the report gives scant attention to how the IRS can recover erroneous refunds once they are issued. After all, once we accept that erroneous refunds inevitably happen, it makes sense to consider how to recover them. But TIGTA’s sole recommendation on how to improve collection of erroneous refunds, which I discuss below, is simply laughable in light of how courts have interpreted the relevant statutes in the past 10 years.

Usually, the amount of erroneous refunds escapes attention. Recently, however, USA Today reported in horror that the IRS let slip some $200 million erroneous refunds from the failure of one computer program during 2005.[3] Worse, IRS Commissioner Mark Everson acknowledged that there is “little chance IRS will collect the bulk of the erroneously issued checks.”[4] While the amount issued in error represents less than 0.1 percent of all refunds, it was enough for then-Senate Finance Committee Chair Chuck Grassley, R-Iowa, to grumble that “it is not just that they’re getting off scot-free – it’s that the honest taxpayers become the suckers.”

Honest taxpayers are suckers because current law renders the IRS impotent to collect erroneous refunds. As a result, tens of thousands of taxpayers receive the government’s unintended largesse and wallow in their windfalls. That is bad tax administration that no amount of TIGTAoversight can fix. It did not used to be that way. The IRS used to be able to collect most of the erroneous refunds efficiently. Now it must all too often write them off. This article tells the sad story of how the IRS lost its mojo and what must be done to get it back.

At one level, erroneous refunds are low-hanging fruits. In light of all the huffing and puffing about the tax gap – that mythical beast smokily mirrored more by assumptions than hard data – it is amazing that no one has suggested an admittedly modest, yet simple and solid, reform that would enable a quick and easy recovery of the occasional unavoidable erroneous refund. At another level, as the following details will demonstrate, this is an intricate and arcane area, a part of the tax administration forest that few visit without mishap.

Honest taxpayers are suckers because current law renders the IRS impotent to collect erroneous refunds. Thousands of taxpayers receive the government’s unintended largesse and wallow in their windfalls.

Part I explains how erroneous refunds occur and how they relate to the law of tax administration. Part II reviews the legal doctrines governing their recovery and demonstrates how, as result of a contentious legal battle in the 1990s, current law leaves the IRS basically shooting blanks in trying to recover taxpayer windfalls. The TIGTA report focuses on ex ante solutions precisely because the IRS ability to collect them efficiently ex post is so compromised. Part III explains why the TIGTA collection recommendation makes little sense and, as is my habit, offers a simple legislative provision to help the IRS get its groove back. What is needed is a little statutory power pill to allow the IRS to interact firmly with taxpayers who try to keep money they should not have.

I. What Are Erroneous Refunds, Anyway?

Erroneous refunds come in two flavors: rebate and nonrebate. How they arise involves the interplay of several tax administration powers: the powers to assess, abate, and refund overpayments. I’ll discuss each in turn and explain the difference between rebate and nonrebate erroneous refunds. That difference has important consequences for the recovery of the erroneous refund.

A. Assessments

Section 6201 gives the IRS the power to assess all taxes owed. Section 6203 says that an assessment is made by “recording the liability of the taxpayer” on the IRS books of account. Three points about assessments are important for understanding rebate and nonrebate erroneous refunds: the distinction between the dual functions of an assessment; the distinction between an administrative bookkeeping act and a legal assessment; and the distinction between the summary and the deficiency assessment procedures.

1. Liability versus collection. Tax administration boils down to two main functions: determination of liability and collection. Assessment plays a role in both. It has a dual function, pointing backwards to the determination function and forwards to the collection function. As to the first, assessment is more than just a simple bookkeeping entry. It is the result of an administrative process that represents the IRS’s institutional judgment of what taxes are owed.[5] That the IRS usually accepts the liability shown on the taxpayer’s return as the basis for the assessment does not in any way diminish the fact that it is the IRS’s decision. That is why, as I have repeatedly emphasized, it is erroneous to say that taxpayers “selfassess.” It is also why one must always be alert to distinguish between assessment of the tax and liability for the tax.[6]

Assessment also plays an important role in the second function of tax administration: It is the culmination of the administrative liability determination process, and it also triggers the start of the collection process. That collectionpropelling function is seen in two ways: A proper assessment enables the IRS to invoke its administrative collection powers of lien and levy, and a proper assessment triggers the separate 10-year period of limitations in which the IRS can collect the tax liability.[7]

2. Legal versus administrative. Legally, the act of assessment culminates in the “recording” authorized by section 6203. Reg. section 301.6203-1 expands on the statute by providing that “the assessment shall be made by an assessment officer signing the summary record of assessment. The summary record, through supporting records, shall provide identification of the taxpayer, the character of the liability assessed, the taxable period, if applicable, and the amount of the assessment.” Since the early 1960s, the IRS has kept its books electronically in a master file database.[8]

A legal assessment is thus made on a “summary record of assessment” signed by an authorized IRS employee. Typically, all liabilities to be assessed over a given time period (usually one or two weeks) are aggregated onto a Form 23-C, “Summary of Assessments,” which is printed and signed once a week.[9] Each Form 23-C completes a cycle that begins with the receipt of the return or other supporting list or record identifying a particular tax liability and the creation of an account in the computerized IRS master file system. Each account, called a tax module, tracks a specific tax liability for a specific taxpayer for a specific tax period.[10] As its name suggests, a Form 23-C represents the sum of all assessments made during the cycle; it does not reflect any particular taxpayer account. That really frustrates taxpayers who request “the” record of assessment only to receive a simple one-page form that has a very large number on it and says nothing about their liabilities.

The supporting records required by the regulations come from computer entries in the tax module. When requested, the IRS will transpose the relevant data in the tax module onto a Form 4340, “Certificate of Assessments and Payments.” That is what links a given tax module to a given summary record. A Form 4340 creates a rebuttable presumption of a proper assessment, but really it is just a translation of the computer account and, as such, may be flawed.[11]

Administratively, the IRS uses a three-digit computer transaction code (TC) to record each event in a tax module on the master file.[12] Importantly, no transaction recorded in the IRS computers is ever erased. Instead, the IRS enters an offsetting TC. A TC itself is not the assessment. For example, tax modules are generally opened with a debit TC 150, most often resulting from the liability reported on a taxpayer’s return. The amount associated with the TC 150 is what gets rolled up into the weekly Form 23-C, but it’s not until the Form 23-C is signed that the legal “assessment” has occurred. The act of inputting the computer code is not the legal act of assessment. Similarly, if a payment is erroneously credited to the wrong account, the erroneous credit TC cannot be erased. Instead, an offsetting TC is entered to reverse the mistake by inserting a debit amount equal to the erroneous credit. The offsetting TC is not an assessment and the question whether the amount reflected in the offsetting TC has to be rolled up into any summary document is a legal question, one that goes to the heart of erroneous refunds.[13]

The mere entry of computer transaction codes cannot erase an assessment in the sense that the administrative process by which the assessment was recorded is somehow undone.

In sum, the legal assessment document, the weekly Form 23-C, does not reflect all the TCs in any individual tax module. Moreover, the amount reflected in the Form 23-C does not necessarily include all TCs that adjust accounts. The administrative accounting for the taxpayer’s balance due for any particular tax module is not conterminous with the legal concept of assessment. The court in Simon v. United States got it right when it said:

Thus the positivistic equation of assessment with entry of a debt on the books is misguided. A debt entry is only the culmination of an assessment, which begins when the IRS determines that a taxpayer is liable.[14]

One important consequence of the distinction between the legal act of assessment and the administrative act of bookkeeping is that the latter can never undo the former. That is, the mere entry of computer transaction codes cannot erase an assessment in the sense that the administrative process by which the assessment was recorded is somehow undone. For example, in United States v. Reid the court held that the insertion of credits on a taxpayer’s account to reduce the balance due to zero did not “eliminate” the assessments: “Crediting the taxpayer’s account – no matter what the amount of the credit – does not undo this act of recording described in section 6203.”[15]

3. Summary versus deficiency procedures. The IRS uses a variety of procedures to assess tax liabilities. The two most relevant to the subject of erroneous refunds are the summary procedure and the deficiency procedure.

The summary procedure is the general rule, authorized by section 6201. All other assessment procedures are either statutory or administrative exceptions to the summary assessment process. It is called the summary procedure because the IRS simply and summarily records the taxpayer’s liability, payments, and credits, based on the information before it, and then notifies the taxpayer if there is a balance due. The most typical example is when the IRS makes an assessment based on the taxpayer’s return, such as the Form 1040 for income taxes or the Form 941 for employment taxes. The form might also be generated by an IRS employee. Before 1924 the IRS also used the summary process to record the results of income tax audits. It still uses that procedure to record the results of excise tax audits.

The key to section 6211 is the phrase “tax imposed.’

The main statutory exception to summary assessment applies to the assessment of income, estate, and gift tax deficiencies. Whenever the IRS seeks to assess a deficiency, it must follow the special procedures set out in section 6213. Those require the IRS to send the taxpayer a notice of deficiency to the taxpayer’s last known address. The notice of deficiency opens a window of opportunity for the taxpayer to petition for Tax Court review. While that window is open, the IRS cannot assess, absent a determination of jeopardy. The window stays open for at least 90 days, and if the taxpayer timely petitions the Tax Court, it stays open until the Tax Court issues its decision and the decision has become final. Only when that window closes can the IRS assess the tax liability and start to collect any balance due.

Understanding when the IRS must use the special deficiency procedures turns on what constitutes a deficiency. Section 6211 defines a deficiency. Good luck understanding it! It has to be one of the most difficult and densest statutes in the tax code. For this column, however, the important point of section 6211 is how it reinforces the distinction between a tax liability and an assessment. Section 6211 is all about finding the true and correct tax liability.

The key to section 6211 is the phrase “tax imposed.” That phrase means the true tax liability – the one that arose at the end of the applicable tax period – and is distinct from what is reflected on the IRS books. To see if what the IRS seeks to assess is a deficiency, the statute instructs one to first find what has been previously recorded in the IRS books. That is, section 6211(a)(1) says to first add the tax reported by the taxpayer to any tax previously assessed for that tax period. Call that the recorded tax. From that amount, one subtracts any rebates as defined in subsection (b)(2).Arebate results from the IRS’s substantive redetermination of tax, a discovery that the true tax was less than what was previously recorded on the books. Let’s call what remains on the IRS books after a rebate the recorded tax as substantivelyn adjusted.[16] If what the IRS claims to be the tax imposed is greater than the recorded tax as substantively adjusted, the IRS seeks to assess a deficiency and must follow the special procedures.[17]

B. Abatements

Section 6404 gives the IRS the power to abate assessments. Just as assessments do not create tax liabilities, abatements do not extinguish tax liabilities. Abatements simply adjust the record on the IRS books. An abatement may or may not have anything to do with the taxpayer’s tax liability. Moreover, just as not every upward adjustment to a tax module is an assessment, neither is every downward adjustment an abatement. I’ll now expand on both points.

First, some abatements result from (and reflect) a substantive redetermination of tax and some do not. When the IRS discovers that the true tax liability (the tax imposed) is less than what is reflected in the assessment on the IRS books (the recorded tax as substantively adjusted), or when the assessment was made illegally, section 6404(a) authorizes abatement to properly reflect the liability or eliminate the illegal assessment. A common cause of a section 6404(a) abatement is an audit reconsideration after a section 6020(b) substitute for return.[18]

Just as assessments do not create tax liabilities, abatements do not extinguish tax liabilities. Abatements simply adjust the record on the IRS books.

If a section 6404(a) abatement later proves to have been erroneous on the merits, the IRS must follow the applicable assessment procedure to reassess the proper liability.[19] For income, gift, and estate taxes, that means the IRS must follow the deficiency procedures to reassess the tax and will be bound by the section 6501 limitations period.[20] That’s a sensible result because in that situation there is a potential disagreement over the true tax liability; requiring the IRS to follow the deficiency procedure here gives taxpayers not just prepayment but preassessment access to judicial review.

Other abatements do not result from (or reflect) a substantive redetermination of tax. For example, when the IRS decides that the assessed amount is not worth collecting because the costs of collection outweigh the amount to be recovered, section 6404(c) authorizes ann abatement of the assessment. In contrast to section 6404(a) abatements, if the IRS later decides that an abatement made per section 6404(c) was erroneous (perhaps having found taxpayer assets that can be collected), it may reinstate the prior assessment without going through the deficiency procedures because no deficiency is at issue.[21] That is, there is no concern that the assessment may not properly reflect the true tax any more than when it was first made. So the taxpayer has no need for any preassessment judicial review.

A good example of the distinction between section 6404(a) and section 6404(c) abatements is United States v. Buckner, 264 B.R. 908, Doc 2001-10251, 2001 TNT 69-13 (N.D. Ind. 2001). There, the IRS levied on the taxpayer’s retirement plan. The taxpayer then filed bankruptcy and was personally discharged from liability for three years of income taxes. But the retirement account, not being part of the bankruptcy estate, remained liable in rem.[22] After receiving the discharge order, the IRS abated the assessments for those years but did not remove the levy from the retirement accounts. The taxpayer asked the bankruptcy court to order the IRS to release the levy, arguing that the abatement extinguished the assessment and the IRS could not revive it without creating a new assessment for the discharged years, which the bankruptcy discharge injunction forbade. The court refused to issue the order and instead held that the IRS could accept the levy proceeds and reverse the abatement without having to make a new assessment because the abatement did not result from a substantive redetermination of the taxpayer’s liability. “Stated simply, a ß 6404(c) abatement reflects a determination by the I.R.S. of a taxpayer’s collectability, and since the I.R.S. may account for collectable assets by simply entering a debit to reverse a prior credit, no formal reassessment procedure must be followed.” [23]

As to the second point, not every downward adjustment to a tax module is an abatement within the meaning of section 6404. For example, because erroneous increases cannot be erased, they must be offset by downward adjustments. It is not always clear, however, which administrative actions constitute a legal abatement and which do not. Unlike the careful description in the code of what constitutes an assessment, neither section 6404 nor its regulations provide guidance as to what it takes to effect a legal abatement. In practice, IRS employees may use different forms to make section 6404(a) abatements.[24] Further, one of the forms used to make a section 6404(a) abatement is also used to make a section 6404(c) abatement for taxes discharged in bankruptcy.[25]

It’s not always clear which administrative actions constitute a legal abatement and which do not.

In deciding which accounting adjustments represent legal abatements (and, if so, what kind of an abatement), courts have traditionally looked at the substance of the transaction over the IRS form used to initiate it or the label of the transaction code used to enter it. The most important substantive issue is whether the administrative action represents, as does a proper assessment, the culmination of a proper administrative process, or whether it is just the result of a glitch in the system, a miscommunication, or a processing error. If the latter is the case, courts hold that no abatement has occurred and the IRS will not be bound by the administrative action of its employees, unless equity demands.[26] And, as anyone experienced in government litigation can tell you, getting equity against the government is a very demanding task.[27]

An excellent example of the focus on substance over form is In re Bugge.[28] There, in the course of preparing a taxpayer’s account for collection, a revenue officer thought he discovered that the computer account was double-counting one of the taxpayer’s liabilities. He therefore filled out the proper form and sent it to the service center, requesting that it abate the liability, ““since this is a duplicate assessment that [has] already been done using the correct [master file tax] and tax period.”'[29] The Fifth Circuit analyzed the situation as follows:

The collections manager in this case never intended or approved an abatement of Bugge’s entire tax liability. * * * In requesting the abatement, the collections officer intended, and received approval from his manager for, a reduction of Bugge’s tax liability from two assessments of $327,379.82 each . . . to a single assessment of $327,379.82. This request was in accord with the IRS’s discretionary authority under section 6404(a)(1) to abate an assessment that is “excessive in amount.” * * * However, when the regional service center processed the request and inadvertently eliminated Bugge’s entire tax liability, it failed to act within the scope of the request that had been approved by the collections manager. In addition, by abating Bugge’s actual and correct tax liability, it failed to act within the IRS’s statutory authority to abate an excessive amount. * * * Because of a purely accidental and unintended processing error, the regional service center executed an unintended abatement lacking any authorization.[30]

Note that the Bugge opinion refers neither to the particular form used to request the abatement nor to the label of the particular transaction code used to reflect the adjustment to the taxpayer’s account. Both were irrelevant to the court’s analysis. Instead, the proper legal interpretation of the IRS action required consideration of institutional intent: whether the administrative action at issue resulted from a process truly reflective of an institutional decision to perform a section 6404(a) abatement or whether it resulted from serendipity, a comedy of errors.[31]

In sum, despite the lack of applicable code provisions, cases like Buckner establish that not all abatements involve substantive redeterminations of a taxpayer’s liability, and cases like Bugge establish that not all downward adjustments of the taxpayer’s liability have the legal effect of abatements.

C. Rebate and Nonrebate Refunds

Section 6402 authorizes the IRS to refund overpayments to taxpayers. Practitioners and academics alike often overlook the distinction between an overpayment and a refund. They are two separate concepts. An overpayment occurs when a tax module shows a credit balance.[32] A refund is what gets paid to the taxpayer. Section 6402(a) does not require the IRS to refund an overpayment but instead permits the IRS to set off the overpayment against other outstanding kinds of taxes or other tax periods owed by the taxpayer. Section 6402(c), (d), and (e) also requires the IRS to set off any remaining overpayments against some state and federal nontax liabilities – such as federal contractor debt or child support payments – if properly requested by the benefiting entity. Thus, a refund is whatever the taxpayer gets after the IRS either exercises its discretion or obeys the statutory commands regarding disposition of the overpayment.[33]

A rebate refund occurs when the taxpayer gets money back because the amount paid or credited is greater than the true liability properly reflected in the IRS books.

A rebate refund occurs when the taxpayer gets money back because the amount paid or credited is greater than the true liability properly reflected in the IRS books.[34] My yearly refund that I love so much is a rebate refund because my amount paid is greater than my true tax. Erroneous earned income tax credit refunds are rebate refunds because the refundable credit is greater than the true tax liability. Likewise, the IRS sometimes determines the previous assessment was excessive because a taxpayer files an amended return that is accepted by the IRS, or because an audit results in a lower tax liability. In those cases, the assessment is abated under section 6404(a) and any resulting refund becomes a rebate within the meaning of section 6213(b). If the IRS later determines that it committed a substantive error in crediting my payments, in crediting the refundable credit, or in making the abatement, the refund becomes a rebate erroneous refund. [35]

In contrast, a nonrebate refund results from an action other than a redetermination of tax liability. For example, a refund arising from a section 6404(c) abatement would be a nonrebate refund. If that abatement was later determined to be erroneous, the refund would be a nonrebate erroneous refund. Likewise, a refund arising from a clerical error-either in adjusting the assessed tax downwards or adjusting the payments or credits upwards is a nonrebate erroneous refund. In those cases, because the substance of the transaction does not involve a redetermination of tax liability, neither does the resulting refund. For the same reason, the erroneous crediting of another’s payment, or a math error involving the taxpayer’s own payments or refundable credits, are acts that result in nonrebate erroneous refunds.[36]

II. The Problem of Collecting Erroneous Refunds

A. Collecting Rebate Erroneous Refunds

On one hand, the law regarding the collection of rebate erroneous refunds is settled. The IRS and the courts have long agreed that just as the IRS must make a new assessment to undo an abatement made under section 6404(a), the IRS can collect a rebate erroneous refund – such as one resulting from a section 6404(a) abatement – only by reassessing the abated tax liability. [37] That’s because the amount of the error (the difference between the true tax and the tax reflected in the IRS books after the abatement) is an error of judgment about the amount of true tax liability and so constitutes an alleged deficiency of tax. If the IRS discovers the error within the applicable limitations period, the taxpayer deserves the protections of the deficiency procedure (access to preassessment judicial review) before the liability can be assessed and collected.

B. Collecting Nonrebate Erroneous Refunds On the other hand, the law regarding the collection of nonrebate erroneous refunds has long been confused and haphazard. Recall that nonrebate erroneous refunds do not represent any redetermination of the taxpayer’s tax liability. They result from either an erroneous collection decision (abatement under section 6404(c)) or an erroneous processing action (for example, a clerical error). In theory, the IRS can collect nonrebate erroneous refunds in one of three ways: moral suasion; authorizing the Department of Justice to file a lawsuit under section 7405 (within the limitation period set out in section 6532(b)) for recovery of the erroneous refund; or administrative collection action. It is the last one in which confusion prevails. I’ll now discuss each possibility in turn.

The law regarding the collection of nonrebate erroneous refunds has long been confused and haphazard.

The first two methods are not controversial. The IRS engages in moral suasion through pattern “beg letters” (forms 510C and 4728, “Notice of Erroneous Refund”). The letters ask for the money back, explaining that the taxpayer has no right to the money. As to legal action, the government has the same common-law cause of action as does any other litigant for money wrongfully or erroneously paid. “No statute is necessary to authorize the United States to sue in such a case. The right to sue is independent of statute.”[38] That common-law action for “money had and received” is declared in section 7405. Congress has self-imposed a limitation period in section 6532(b), which allows two years from the “making of such refund” to file suit under section 7405, unless the refund was induced by the taxpayer’s fraud or misrepresentation of a material fact, in which case the IRS has five years to act. The courts have interpreted the phrase “making of such refund” to mean the date the check was cashed and “cleared the Federal Reserve.”[39]

Although not controversial, neither of those two collection methods is satisfactory. The problem with moral suasion is, I hope, self-evident. I am not aware of any studies on how effective the IRS letters are. I assume many taxpayers do the right thing and return the payments. I also assume many taxpayers may not read the letters, may be confused about their account, may not believe the refund is erroneous, or else believe that nine-tenths of the law allows them to keep it anyway. Practitioners can help persuade clients to voluntarily return refunds by reminding them that taxpayers have been criminally prosecuted for cashing erroneous refund checks.[40]

Recovering erroneous refunds by suit under section 7405 is unsatisfactory for multiple reasons. First, it is a retail solution for a wholesale problem. It is working the case at the individual level and not the bulk processing level. And each case is worked, not by just one field agent, but by teams of government employees. First, the IRS team must discover the error and authorize the Department of Justice to sue. Then the DOJ team must file suit, and the court “team” must adjudicate, which by itself can take well over a year, assuming a summary judgment resolution.[41] So it forces collection of what is usually an obvious error into an inefficient adversarial process. Second, a section 7405 liability is harder to collect than is a tax liability. Courts are unlikely to treat a section 7405 judgment as a tax debt, which means no tax lien arises to protect the claim, no priority is given the claim in an ensuing bankruptcy, and the IRS cannot perform administrative levies.[42] Finally, the DOJ may not even agree to file suit. Only erroneous refunds larger than a specified base amount will be worth the cost to the government to prosecute. While I have no idea what that amount might be, of the few section 7405 cases reported, most are for hundreds of thousands, or millions, of dollars. I cannot find a reported case of the government suing for less than $10,000.[43]

C. The Erroneous Refund Revival Theory

The IRS would much prefer to fix administratively what is, almost by definition, an administrative error. And from the mid-1970s until early 1998 it did just that. The idea was first raised in 1970. The IRS Office of Chief Counsel studied the issue, and in two important general counsel memoranda, one issued in 1975 and the other in 1976, it developed a theory whereby the IRS could administratively recover so much of a nonrebate erroneous refund as was equal to or less than the original assessment.[44] The gist of the reasoning was this:

If a non-rebate erroneous refund is made with respect to a year for which the taxpayer’s tax has been assessed, which it would be if tax were shown by the taxpayer on a return, and to the extent that the amount of erroneous refund is not greater than the amount of tax that has been assessed and paid, then the erroneous refund is in effect a giving back to the taxpayer of his payment of assessed tax. In this case the tax has already been assessed, and in our opinion there is no requirement that the tax be reassessed before the normal collection procedures can be utilized.[45]

Based on that reasoning, the GCMs concluded that, as to nonrebate erroneous refunds, “the amount of the refund may be recovered by the usual tax collection procedures, including offset under Code ß 6402(a), without use of the assessment or deficiency procedures.”[46] Further, the IRS was bound by the 10-year limitation period for collection in section 6502 and not the 3-year period for assessment in section 6501. That theory became known as the erroneous refund revival theory and resulted in a significant change in IRS practice.[47] To see how the IRS theory worked – and why it ultimately was rejected by the courts – I will use three simplified examples (I omit interest and penalties). The examples also illustrate how refunds may be smaller or larger than the assessed tax and how refunds may arise because of either erroneous crediting of payments or erroneous abatements of assessments.

Example 1 (Alex): Alex reports a tax liability of $4,000 on the return, claims $2,000 in withholding credits, and sends a check to cover the $2,000 balance due. During processing, a clerk erroneously inputs the check twice. Assuming no setoffs, the system generates a refund to Alex of $2,000. When the error is discovered, a clerk enters a new debit TC to reverse the error and the tax module shows a balance due of $2,000. The new debit TC is not rolled into the summary record of assessments.

Example 2 (Blair): Blair reports a tax liability of $4,000 on the return, claims $2,000 in withholding credits, and sends no payment to cover the balance due. The return is properly processed and shows a balance due of $2,000. A year later, Blair sends in the remaining payment. However, at that time a clerk also erroneously credits Blair’s tax module with a payment of $7,000 received from another taxpayer for another tax liability. Assuming no setoffs, the system generates a refund to Blair of $7,000. When the error is discovered, the clerk enters new TC codes to reverse the error, leaving the account with a balance due of $7,000. Again, no new assessment is made.

Example 3 (Cody): Cody reports a tax liability of $4,000 on the return, claims $2,000 in withholding credits, and does not pay the balance. The return is properly processed and shows a balance due of $2,000. However, four years later, when a field employee later submits a Form 3870, “Request for Abatement,” requesting abatement of a different taxpayer’s account in the amount of $4,000, the IRS campus processing clerk erroneously processes the request on Cody’s account. The TC used by the clerk (TC 291) is translated onto the Form 4340 as “abatement prior to tax assessment.” The reduced liability results in the tax module showing an overpayment of $2,000 that is then refunded. When the error is discovered, the IRS clerk enters a new debit TC to reverse the credit, thus creating am balance due of $4,000 in Cody’s account. Again, the entry of the reversing credit transaction is not a new assessment.

The IRS erroneous refund revival theory would allow the IRS to administratively collect all $2,000 of the erroneous refund to Alex, $4,000 of the $7,000 erroneous refund to Blair, and $4,000 from Cody (the $2,000 erroneous refund and the $2,000 unpaid on the original assessment). All of those amounts represent proper assessments that were never legally abated and remain unpaid. Even though Alex and Blair paid all their liability, the IRS theory says that the erroneous refunds mistakenly returned their tax payments to them (and then some, in Blair’s case). In that way, the IRS theory focuses on the liability determination function of an assessment, reasoning that if the assessment properly reflects the true tax liability and if the account shows a balance due, what remains unpaid is a tax liability.[48]

The IRS would much prefer to fix administratively what is, almost by definition, an administrative error.

Thus, the IRS could administratively collect the entire $2,000 refund from Alex because Alex could not gainsay that the properly made assessment reflects the true tax liability. Likewise, the IRS can administratively collect $4,000 from Blair because that is the amount of the true tax liability that – when all is said and done – remains unpaid. However, because the erroneous refund to Blair exceeded the true tax liability by $3,000, the IRS would have bifurcated remedies: It could collect the $4,000 tax liability administratively but would have to file suit under section 7504 to collect the excess $3,000. Similarly, the IRS can collect $4,000 from Cody because the clerical error that caused the refund was not a legal abatement.[49]

D. The Extinguishment Theory

For the first 20 years, the IRS often won challenges to its erroneous refund revival theory in the lower courts.[50] However, the GCMs that developed the theory acknowledged one weakness: the doctrine of extinguishment. The courts first articulated that idea in 1929, in Kelley v. United States: “Once paid, a tax is gone, and a refund of the money does not restore it.”[51] Thirty years later, the district court in In re Marshall relied on Kelley to reject the idea that a bankruptcy claim based on an erroneous refund was a claim for taxes entitled to special treatment. [52] In turn, the district court in Rodriguez v. United States relied on Marshall to conclude that “a refund is not, properly speaking, a tax amount,” and therefore “the act of sending a refund cannot of itself revive or continue a preexisting tax liability.”[53] The Fifth Circuit picked up on that argument in United States v. Wilkes in 1991 and, agreeing with both Rodriguez and Marshall, rejected the IRS erroneous refund revival theory. After that, it was all downhill, as the government lost in four other courts of appeals over the next six years.[54] Ultimately, the IRS conceded the issue in an action on decision dated May 4, 1998.[55] The government then abandoned the position on two appeals from lower court decisions that had upheld the theory, and the IRS changed its collection procedures in the Internal Revenue Manual.[56]

In contrast to the erroneous refund revival theory, the extinguishment theory focuses on the second function of an assessment as a precursor to administrative collection. The idea is that once a taxpayer pays any portion of the liability, the assessment can no longer serve as a proper precursor to administrative collection. Thus, as applied to the above three examples, the IRS would have no power to administratively collect any amount from Alex or Blair because they had fully paid the assessed liability. The extinguishment theory would permit collection of $2,000 from Cody because that was the portion of the original assessment that remained unpaid. But Cody’s partial payment (the $2,000 in withholding credits) could not be recovered, even by courts that improperly apply the extinguishment theory and say the IRS can perform a supplemental assessment of a nonrebate erroneous refund if done within three years. In short, properly applied, the extinguishment theory holds that the IRS can recover nonrebate erroneous refunds of “paid taxes” only through suit under section 7405.

In sum, under both the revival and extinguishment theories, the IRS will sometimes be forced into bifurcated remedies for administrative errors. The chief difference is that the IRS theory uses the original assessed liability to measure what can be recovered administratively because the assessment’s liability determination function was unimpaired by the error. In contrast, the extinguishment theory focuses only on the unpaid assessed liability as the proper measure of what can be recovered administratively because the assessment’s “collection-propelling” function is impaired, or cut down, by the taxpayer’s payment.

E. Problems With Extinguishment Theory

The extinguishment theory does not jibe with good tax administration. It characterizes nonrebate erroneous refunds as a nontax liability yet insists the IRS can collect it using administrative collection tools if the IRS just timely assesses it. But the IRS can assess only tax liabilities. So by saying that nonrebate erroneous refunds are not tax liabilities, courts in fact disable the IRS from pretty much any administrative collection action.[57] Make no mistake about it: A nonrebate erroneous refund, by definition, results in a windfall to the lucky taxpayer. Whether the taxpayer is fully aware of the windfall or not does not change its character. The taxpayer is still getting the free use of the government’s money. We’re talking handouts here.[58] In that way, the extinguishment theory hurts those taxpayers who comply with their obligations and rewards taxpayers who avoid returning (whether through ignoranceorobstinance)theirundeserved(evenifunsolicited) gain.

There is a central contradiction to the extinguishment theory. On one hand, courts hold that nonrebate erroneous refunds are nontax liabilities that neither raise nor revive any tax liability. They are instead simply commonlaw liabilities that arise from the wrongful holding of the government’s money by the taxpayer. On the other hand, the same courts say that the IRS can reassess the refund if it acts within the assessment limitations period, using either its authority under section 6201 to assess or under section 6204 to make a “supplemental assessment whenever ever it is ascertained that any assessment is imperfect or incomplete in any material respect.” Here is how the O’Bryant court explained it:

The money the O’Bryants have now is not the money that the IRS’ original assessment contemplated, since that amount was already paid. Rather, it is a payment the IRS accidentally sent them. They owe it to the government because they have been unjustly enriched by it, not because they have not paid their taxes. Because it is a refund, the money the O’Bryants received is not part of the taxpaying transaction as the IRS asserts and therefore cannot be recovered through the ß 6502 post-assessment collection procedures. It would not make sense to allow the IRS to use those procedures (which are premised on the taxpayer’s not having paid his tax debt) to recover money it accidentally sent to the taxpayer. Rather, the agency is confined to the erroneous refund collection procedures available to it under the Tax Code-ß 7405 and the deficiency/ assessment procedures.[59]

Some courts – the ones that do not understand the difference between rebate and nonrebate refunds-insist the IRS must use the deficiency procedures to perform that theoretical reassessment. Look at the language from Singleton:

Sections 6204(b) and 6213(a) [prescribing deficiency procedure] prohibit the IRS from issuing a supplemental assessment without first issuing a notice of deficiency and giving the taxpayer an opportunity to contest the assessment in Tax Court. The Court finds no applicable exception to these procedural requirements. Section 6213(a) applies to all assessments; it does not distinguish between assessments intended to reclaim rebate refunds versus those intended to reclaim non-rebate refunds.[60]

The contradiction here is that the IRS can assess only “taxes . . . imposed by this title.”[61] If the liability to return nonrebate refunds is not “imposed by this title” but is instead a common-law liability for money had and received, the IRS cannot assess them. It has no authority to do so. And yet courts insist – in dicta mostly – that the IRS can.[62]

The reason courts keep saying that the IRS can assess refunds is because the legislative history to section 7504 instructs that the section was not meant to limit the IRS to recovery of erroneous refunds by lawsuit. When Congress first enacted the language now in section 7504, back in 1928, the committee report was clear that it did not intend the legislation to cut off the IRS’s administrative remedies: “Obviously, if the limitations period on the making of the assessments has not expired, the erroneous refund may be recovered by assessment in the ordinary manner.”[63]

The extinguishment theory hurts taxpayers who comply with their obligations and rewards taxpayers who avoid returning their undeserved (even if unsolicited) gain.

When the 1928 committee report was written, it was well established that erroneous refunds created tax liabilities. Recall that the distinction between rebate and nonrebate refunds did not arise until 1944: In 1928 all refunds were rebate refunds.[64] Recall further that the courts and the IRS have always agreed that to recover erroneous rebate refunds, the IRS must reassess.[65] That’s based on the view that rebate refunds result from redetermination of the tax liability.[66] So, in 1928 it was true that every erroneous refund resulted in a deficiency of tax. The above quote from Singleton would have been perfectly true . . . in 1928.

So the question becomes whether Congress in 1944, when it changed the definition of rebate in section 6211, meant to undo what had been true in 1928. Did Congress mean to no longer allow the IRS the ability to administratively recover that class of refunds now removed from the definition of deficiency? Did Congress intend, by a definitional change, to suddenly restrict the IRS to suits under section 7504 to recover refunds that no longer qualified as a deficiency? The two GCMs issued in 1975 and 1976 thought not. The 1975 GCM expressed its conclusion this way:

A rational interpretation of the statutory collection scheme as a whole requires the conclusion that tax collection procedures are applicable here [in the case of a nonrebate refund] as well as in the case of a rebate. If the procedures were not available here, the only means for recovery of the refund would be civil suit under Code ß 7405. Yet there seems no reason why Congress [in 1944] would have intended a more restrictive rule for recovering refunds which are not rebates than for recovering rebates. Rather, the logical interpretation would be that rebates and non-rebate refunds may be recovered in the same manner except that the deficiency procedure must be used prior to the assessment of a rebate so the taxpayer may have access to the Tax Court where there is a question of liability for tax, whereas a non-rebate refund, not made on the basis of a determination of tax liability, may be recovered by simple assessment.[67]

The 1976 GCM modified that analysis by eliminating the requirement for reassessment. Focusing on the liability determination function of the assessment, it pointed out that a nonrebate refund did not change the true liability, which was already accurately reflected in the existing assessment. That is, just as a clerical error does not result in an abatement of tax under section 6404(a), neither does a nonrebate refund result in a new tax liability because it is just the giving back of a tax payment. Accordingly, not only was no reassessment necessary but it was not even permitted because there was no new tax liability to assess.

F. Problems With the Revival Theory

Although more rational than the extinguishment theory, the revival theory is far from perfect. It creates the opposite problem of the extinguishment theory: It gives the IRS more power to collect erroneous nonrebate refunds than rebate refunds. That is, while one cannot say that Congress, in changing the definition of rebate in 1944, intended to make the IRS less able to collect nonrebate erroneous refunds, neither can one say that Congress intended to suddenly enhance the IRS’s ability to collect erroneous nonrebate refunds over rebate refunds. Yet that is what the revival theory does.

The first way in which the revival theory prefers nonrebate refunds concerns how liens arise. The law under section 6321 is well settled that the tax lien arises when there is a proper assessment, proper notice and demand of the unpaid amount, and a failure to pay.[68]

Alex illustrates the problem. Recall that Alex fully paid the assessed tax and so, under the established law regarding section 6321, no tax lien arises. Further, the IRS could not administratively collect a rebate erroneous refund made to Alex unless it could reassess within the section 6501 assessment limitations period, which is generally three years. But the revival theory allows the IRS to administratively collect a nonrebate erroneous refund from Alex at anytime within the section 6502 collection limitations period, which is generally 10 years. That’s because the erroneous revival theory views an assessment, once established, as inviolate, so once the nonrebate refund is issued, all that would be needed to trigger the tax lien would be notice and demand and a failure to pay. The revival theory would thus give more tax lien protection to nonrebate refunds than rebate refunds.

The revival theory transforms section 6321 into the zombie lien statute.

The second way in which the revival theory prefers nonrebate refunds over rebate refunds concerns how liens are satisfied. Section 6322 provides that the tax lien “continues until the liability for the amount so assessed . . . is satisfied or becomes unenforceable by reason of lapse of time.” The revival theory transforms section 6321 into the zombie lien statute.

Blair illustrates the problem. Recall that Blair did not fully pay the tax initially. Accordingly, after notice and demand was made, and he continued his failure to pay, the tax lien arose. But then Blair sent in the remaining payment. At that point, when the “liability for the amount so assessed . . . [was] satisfied,” the tax lien was extinguished; it died.[69] But according to the revival theory, the processing error creating the nonrebate refund brings the tax lien back from the dead.

Cody illustrates how the tax lien is supposed to work. There, the tax lien arises because Cody has an unpaid balance due when the assessment is made. As long as any part of the assessment remains unpaid, the tax lien remains outstanding. It may secure a greater or lesser claim, depending on the payments and credits, but it can be extinguished only by full payment or running of the limitations period on collection. Once extinguished, however, it’s dead and should stay dead. That is the teaching of section 6325, which provides that once the IRS determines a taxpayer has satisfied the “liability,” the IRS must issue a certificate releasing the lien. And the effect of the certificate of release is that the lien is extinguished and cannot be reinstated, in explicit contrast to the statute’s allowance for the revival of liens on property from which they had been removed (through certificates of discharge or nonattachment).

III. The Needed Legislative Solution

A. The Limits of Administrative Solutions

In its September 29, 2006, report, TIGTA criticized the IRS for not doing more to collect erroneous refunds. After harrumphing about how much better the IRS could be in tracking and analyzing erroneous refunds, it came up with this recommendation for how the IRS might better collect them:

The Wage and Investment Division, with input from the IRS Office of Chief Counsel, should consider revising its erroneous refund procedures to include a financial analysis conducted electronically by Submission Processing Accounting function employees working erroneous refund cases to determine collectibility on cases above a specific dollar tolerance, and to refer those cases with collection potential directly to the local IRS Office of Chief Counsel. We believe adopting this change would be a more efficient use of resources.

That is a laughably lame suggestion. TIGTA wants low-level IRS employees to figure out whether taxpayers who have received erroneous refunds have enough assets to make the effort of obtaining a judgment worthwhile. The IRS management politely objected to the recommendation, citing lack of training. But there are much worse problems with TIGTA’s idea.

First and most fundamentally, TIGTA’s proposed solution puts the collection cart before the judgment horse. TIGTA wants the IRS to first determine whether a taxpayer has assets and only then refer the case to the DOJ. In almost no other context does the IRS make a collectibility decision before a liability decision. You do not see revenue agents inquiring whether taxpayers have the wherewithal to pay a deficiency![70] And for good reason: What is here today can be gone tomorrow, and vice versa. There is no correlation between a taxpayer’s assets today and the ability of the government to collect from the taxpayer some two or three years down the road whenever the DOJ gets the section 7504 judgment.

TIGTA’s suggestion is laughably lame:

Low-level IRS employees should figure out whether taxpayers who have received erroneous refunds have enough assets to make the effort of obtaining a judgment worthwhile.

Second, TIGTA’s suggestion would badly misapply IRS resources by draining significant amounts of fulltime equivalents (FTEs) – the measurement of worker hours – from the primary mission of submissions processing. That’s because a collectibility determination requires financial information about the taxpayers. The only source of financial information readily available to submission processing employees is the data used by the automated collection system (ACS): the data in the IRS data systems gathered from various returns by the taxpayer and third parties. That data is not reliable. People change bank accounts. People change employers. That data is also incomplete: Just because a bank sent in a 1099-INT for a taxpayer doesn’t mean you know anything about the balance in that account.

At the collection stage, that bad and incomplete data does not matter because ACS just sends out a levy and if it hits, it hits. But what works well to collect an assessed tax would not at all work well to perform TIGTA’s suggested financial analysis. You cannot use a levy to collect information unless you have an assessed tax. TIGTA’s suggestion would require submissions-return employees to issue and track summonses. They would have to track down taxpayers to secure a valid Form 433A, “Collection Information Statement.” That is the work of field agents. It takes huge hunks of FTEs. Even if you could train submission processing employees to do all that, you would throw them into a hugely inefficient operation.

The basic problem facing the IRS is that current law does not provide a satisfactory framework for the collection of nonrebate erroneous refunds, no matter which theory you subscribe to. The IRS’s erroneous revival theory and the alternative extinguishment theory create problems in the law. Meanwhile, the inability of the IRS to administratively collect nonrebate erroneous refunds results in huge windfalls for many taxpayers. What is needed is a legislative fix.

B. The Legislative Fix

In many ways, the current situation parallels that facing Congress when, in 1954, it enacted section 6201(a)(3). That section allows the IRS to assess the entire amount of erroneously allowed withholding credits, even when the resulting refund is larger than the assessed tax liability. The Senate committee report explained that Congress believed the IRS faced a problem with bifurcated recovery of erroneous refunds based on overstated withholding credits and accordingly wrote the statute to allow the IRS to make a single assessment to recover the entire erroneously refunded amount.[71] For example, if the erroneous refund to Cody had resulted from a mistake in recording the withholding credit (erroneously posting it as $12,000 instead of $2,000), resulting in a $10,000 refund, section 6201(a)(3) would permit the IRS to assess the entire $10,000 refund as a tax liability (thus allowing the IRS to administratively collect $12,000, the $2,000 erroneously applied to satisfy Cody’s liability and the $10,000 refund). Unfortunately, while the purpose of section 6201(a)(3) was to prevent the necessity of bifurcated recovery procedures for a single erroneous refund, it covers only refunds made on the basis of one particular type of error.

Congress should act to prevent the necessity of bifurcated recovery procedures for any erroneous nonrebate refund. Congress should create two new provisions, one in Chapter 68, “Additions to Tax, Additional Amounts, and Assessable Penalties,” and one in Chapter 66, “Limitations.” The first would be a new section 6658 (current section 6658 would be renumbered as section 6659), titled “Failure to Return Erroneous Refund.” It would read: “In the case of any person failing to return or repay any nonrebate erroneous refund after the Treasury secretary has requested its return or repayment by written notice (either in hard copy or electronic form) delivered in the manner prescribed in section 6303, the secretary may assess and collect a penalty in the amount of the erro- [70] While there are provisions for evaluating offers in compro- neous refund.” The second would be a new subsection (n) in 6501 (the assessment limitations period; current sections 6501(n) and 6501(o) would be renumbered). It would be titled “Special Rule for Certain Erroneous Refunds” and would read: “Assessment of the amount authorized by section 6658 must be made within five years from the date that the erroneous refund has cleared the Federal Reserve or otherwise been released to the person assessed.”

The reason that I suggest that particular arrangement, instead of adding another subsection to the section 6201 assessment authority, is that the policy behind the threeyear period for assessments of liabilities is inapplicable to the kind of errors that create nonrebate erroneous refunds. The three-year limitations period for assessment represents congressional policy that the substance of the taxpayer’s liability be settled-that is, properly reflected in an assessed liability-within three years from the due date of the return. A substantive error resulting in a rebate refund does not affect that policy because the policy goes toward the first function of an assessment – to properly reflect the true tax liability of a taxpayer. But a nonrebate refund does not implicate that policy because it has nothing to do with the first function of an assessment: the proper reflection of a tax liability. That is the insight from the 1975 and 1976 GCMs and is what formed the basis for the revival theory. Further, because a nonrebate erroneous refund is not connected with a substantive redetermination of tax, it can happen at any time. It does not make sense that the IRS’s ability to administratively collect a nonrebate erroneous refund should turn on the happenstance of when it occurred.[72] For those reasons, Congress should give

For those reasons, Congress should give the IRS a reasonable time in which to discover and collect back money erroneously refunded to taxpayers for reasons unconnected with substantive redeterminations of tax. The $200 million in erroneous refunds identified by USA Today should not be given up without giving the IRS a reasonable shot at collecting it back.

IV. Conclusion

I view law in general, and tax law in particular, as a slow-moving conversation between the various rulemaking authorities: the courts, Congress, and the government agencies charged with administering the law (here, the IRS). The conversation is ongoing and in the great tradition of fragmented democracy, each participant acts to check and balance the others. Sometimes that works to make the law better, but sometimes it makes the law worse. In the area of erroneous refunds, there has been a long awkward pause in the conversation and it is Congress’s turn to speak. I hope Congress will do so here to advance the cause of good tax administration.

[1]2005 Data Book, Table 1.

[2]TIGTA Report 2006-40-137, “Improvements Are Needed to Better Identify and Prevent Erroneous Refunds.” TIGTA notes that while the IRS campuses are supposed to create quarterly reports about erroneous refunds, the reports are not consistently produced by all campuses nor retained nor reviewed at a national level. “No effort has been made to use these reports to track the increase or decrease in the numbers and types of errors being made that cause erroneous refunds.” Id. at 5. Note that for the 2007 filing season, the IRS plans to offer split refunds for the first time, allowing taxpayers to split their refund among up to three direct deposit accounts. See Dustin Stamper, “Taxpayer Assistance Blueprint Will Be “Mind-Shaking,’ DuMars Says,” Tax Notes, Nov. 13, 2006, p. 612, Doc 2006-22837, 2006 TNT 217-4. That added complexity potentially increases the error rate.

[3]See “How the IRS Failed to Stop $200M in Bogus Refunds,” USA Today, Dec. 4, 2006, available at money/perfi/taxes/2006-12-04-irs-bogus-refunds_x.htm (last visited Dec. 6, 2006).


[5]Rambo v. United States, 492 F.2d 1060, 1061 n.1 (6th Cir. 1974) (“Assessment is an administrative determination that a certain amount is currently due and owing as a tax. It makes the taxpayer a debtor in much the same way as would a judgment”); Cohen v. Mayer, 199 F. Supp. 331, 332 (D. N.J. 1961), affirmed sub nom., Cohen v. Gross, 316 F.2d 521 (3d Cir. 1963) (“assessment is a prescribed procedure for officially recording the fact and the amount of a taxpayer’s administratively determined tax liability, with consequences somewhat similar to the reduction of a claim of judgment”); Simon v. United States, 261 F. Supp.2d 567, 573, Doc 2003-20449, 2003 TNT 181-21 (M.D. La. 2003) (assessment “is the culmination of a process whereby liability is determined”); Pipola v. Chicco, 169 F. Supp. 229, 231 (S.D.N.Y. 1959) (“The assessment is an administrative determination that one is indebted to the Government for taxes – in effect, it is a judgment for taxes found due”).

[6]Taxpayers are liable for taxes independent of the assessment. See Ewing v. United States, 914 F.2d 499, 502-503 (4th Cir. 1990), cert. denied, 111 S. Ct. 1683 (1991) (rejecting taxpayer’s argument that, before assessment, there can be no tax liability and therefore no “payment” of taxes). That is why section 6501(a) allows the IRS to either assess or bring proceedings in court without assessment within three years after the return is filed. Think “form” for assessments and “substance” for liabilities.

[7]Hibbs v. Winn, 542 U.S. 88, 102, Doc 2004-12400, 2004 TNT115-11 (2004).

[8]Discussion of the woeful deficiencies of the master file database is beyond the scope of this article. I was once a small part of a large task force to develop a replacement, called the customer account data engine (CADE). CADE is a relational database that can be accessed and updated in real time and will hopefully eliminate many of the problems with the creaky master file system. See Dustin Stamper, “Support Growing for IRS E-Filing Portal, Everson Says,” Tax Notes, Nov. 6, 2006, p. 608, Doc 2006-22589, or 2006 TNT 214-1 (“The IRS is tentatively planning to process 30 million returns in CADE in 2007”). My particular task force worked on developing definitions for various data elements in the database. CADE has great potential to significantly improve tax administration.

[9]The newer RACS Report-006 also serves as the summary record of assessment and, as I understand it, is signed electronically.

[10]There are two master file systems, one for individual taxpayers and one for business taxpayers. The IRS also keeps a separate system of non-master-file accounts whenever it needs to account for situations not covered by the software written for the master file. Changes to the master file software can take years to implement, and the barrage of tax code changes made by Congress means that the master file system is almost always out of date in some respect. The term “tax module” is imprecise, and while it is most commonly used the way I describe it in the text, it has other meanings as well. See my discussion in “Failure of Collection Due Process, Pt. 1: The Collection Context,” Tax Notes, Aug. 30, 2004, p. 969, n.12, Doc 2004-16770, 2004 TNT 169-32.

[11]For an example of an erroneous Form 4340, see Freije v. Commissioner, 125 T.C. 14, Doc 2005-15064, 2005 TNT 135-11 (2005) at note 5. The presumption of correctness is, however, very strong. The “Pursifull saga” contains a most illuminating discussion. Compare Pursifull v. United States, 92-2 U.S.T.C. para. 50,346 (S.D. Ohio 1992) (IRS motion for summary judgment denied and taxpayer entitled to further discovery on showing genuine dispute of the validity of the Form 4340 presented), with Pursifull v. United States, 1993 U.S. Dist. LEXIS 11738 (S.D. Ohio 1993) (taxpayer convinces magistrate judge to reject IRS renewed summary judgment motion after discovery), and with Pursifull v. United States, 849 F. Supp. 597, Doc 93-9370, 93 TNT 184-12 (S.D. Ohio 1993) (district judge reverses magistrate’s recommendation and grants IRS summary judgment on strength of Form 4340).

[12]Transaction codes and their explanations are collected into a yearly bound publication, IRS Document 6209, ADP Handbook. The one I work from in this article is from 2003 and can be found at Transaction codes are complex and difficult to decipher, even for most IRS employees. Each primary three-digit code can have secondary and tertiary codes associated with it. The ADP Handbook covers many other database coding systems as well and is useful for understanding transcripts.

[13]For example, TC 610 records a credit from a payment submitted along with the return and TC 612 records a debit equal to the amount of an erroneous TC 610 credit. The amount recorded by TC 612 is not reflected in a summary record of assessments

[14]261 F. Supp.2d 567, 573, Doc 2003-20449, 2003 TNT 181-21 (M.D. La. 2003).

[15]2000-1 U.S. Tax Cases (CCH) para. 50,340 (S.D. Ga. 2000). I will return to this point again when discussing how and why the IRS should be allowed to recover administratively erroneous refunds that result from bookkeeping errors.

[16]A rebate is any adjustment on the IRS books “as was made on the ground that tax imposed . . . was less than the excess of the amount specified in subsection (a)(1) [that is, the recorded tax] over the rebates previously made.” The iterative and self-referential nature of the definition makes it difficult to understand. The easiest way to approach it is to read the definition as if there were no prior rebates. Then one quickly sees that a rebate refers to some substantive redetermination of the proper tax owed. The “rebate” language was added in 1944 to deal with the new “pay as you go” system of tax collection. See S. Rep. No. 78-885, 78th Cong. 2d Sess., 1944 C.B. 858.

[17]For another good explanation, see Judge Niemeyer’s excellent dissent in Singleton v. United States, 128 F.3d 833, Doc 97-28978, 97 TNT 203-11 (4th Cir. 1997), gently pointing out the majority’s complete failure to understand the concept. It is the iterative nature of the statute that makes it difficult to read. One sees that not only in the definition of rebate but also in the definition of deficiency, which is defined with reference to prior deficiencies. The self-referential structure of the statute makes it read like a chicken-and-egg puzzle. I think section 6211 could be successfully simplified by making more explicit the distinction between what the IRS now has determined to be the true and correct tax and what is recorded as reflecting the true tax. But taxwriters are loath to change administrative provisions. In 1952, as a young man of 25, Sheldon Cohen read through the entire IRC as part of a team making comprehensive revisions to the regulations. He noted in an e-mail to the author that, even today, most of the administrative provisions have not changed. He’s right. In fact, a significant amount of language dating to the various revenue acts of the 1860s is still in the code – like statutory DNA – chopped up and recombined through the various codifications, but still there. For one example of that, see Bryan T. Camp, “Tax Administration as Inquisitorial Process and the Partial Paradigm Shift in the IRS Restructuring and Reform Act of 1998,” 56 Fla. L. Rev. 1, 36-77 (2004) (reviewing history of the summons powers). The historical reluctance to revise the administrative provisions means that the administrative provisions of the code no longer reflect a unified vision of tax administration. They are a jumble and in serious need of an overhaul.

[18]Similarly, section 6404(d) authorizes abatement when the IRS determines that interest assessed on a tax liability resulted from an unreasonable error or delay on the part of an IRS employee. That is again a redetermination of the liability decision represented by the assessment.

[19]Carlin v. United States, 100 F. Supp. 454, 455 (Ct. Cl. 1951) (referring to a time when the abatement power in section 6404(a) was lodged with the commissioner only and the abatement power in section 6404(c) was lodged with the collectors only, the court noted, “If the Commissioner abates the assessment, it ceases to exist or to have any effect thereafter. The Commissioner cannot subsequently rescind his actions or restore the assessment, but must rather make a new assessment unless, of course, the statute of limitations has previously expired”) (citations omitted).

[20]Id. The running of the assessment limitations period, however, does not require the IRS to return money it has properly collected within the limitation period if the true and correct tax is more than the amount collected, even if the assessment is incorrect. See Lewis v. Reynolds, 284 U.S. 281 (1932) (expiration of assessment limitations period without assessment being recorded does not bar the IRS from retaining payments already received if they do not exceed the amount that could have been – but was not – properly assessed within the limitations period).

[21]See, e.g., Crompton-Richmond v. United States, 311 F. Supp. 1184, 1186 (S.D. N.Y. 1970) (“An assessment abated under (a)(1) is thereby canceled and cannot be resurrected if the IRS later decides that its decision was incorrect. On the other hand, the IRS can revive an assessment abated under (c), because the abatement of an uncollectible tax . . . in effect . . . excuses its collector’s obligation (in this case the Brooklyn District Director) to account for the tax liability, but does not excuse the taxpayer’s liability”).

[22]See Johnson v. Home State Bank, 501 U.S. 78, 84 (1991) (“A bankruptcy discharge distinguishes only one mode of enforcing a claim – namely, an action against the debtor in personam”); see also In re Conston, 181 B.R. 769, 773 (D. Del. 1995) (collectingcases)

[23]264 B.R. at 912.

[24]Form 5344, “Examination Closing Record,” is generally used for cases in Exam, see IRM, while Form 3870, “Request for Adjustment,” is used by other IRS functions; see also IRM (01-31-2002), “Abatement of Interest Claim Cases,” for use of Form 3870 to make interest and penalty abatements per section 6404(e).

[25]IRM, “Adjustment Methods for Discharged Liabilities.”

[26]See, e.g., Kroyer v. United States, 55 F.2d 495, 499 (Ct. Cl. 1932) (the government will not be “bound by the bookkeeping errors of its agents, when such errors in no way affect the real equities of the case or result to the prejudice of” the taxpayer).

[27]See In re Becker, 407 F.3d 89, Doc 2005-9433, 2005 TNT 86-9 (2d Cir. 2005) (discussing the requirements of proving estoppel against the government for misapplication of payments).

[28]99 F.3d 740, Doc 96-30396, 96 TNT 226-14 (5th Cir. 1996).

[29]Id. at 744 (quoting the request).

[3]Id. at 745 (footnote omitted).

[31]See also Crompton-Richmond v. United States, 331 F. Supp. 1184, 1187 (S.D. N.Y. 1970) (“whenever an abatement is issued because of a mistake of fact or bookkeeping error, the assessment can be reinstated, at least so long as this does not prejudice the taxpayer” because the government is not bound by clerical errors of its agents); In re Range v. United States, 245 B.R. 266, 274-275 (S.D. Tex. 1999) (upholding as not clearly erroneous the Bankruptcy Court’s decision that administrative action was not an abatement). The Second Circuit in In re Becker, 407 F.3d 89, 100, Doc 2005-9433, 2005 TNT 86-9 (2d Cir. 2005), thought Bugge inconsistent with Crompton-Richmond, but despite some difference in language, both cases hold that not all administrative acts taken by IRS employees rise to the level of legal acts that bind the IRS as a matter of law. Both cases also leave open the possibility that equity (through the doctrine of estoppel) will undo any undue harm resulting from the legal rule.

[32]Section 6401 also provides that payments assessed or collected after the expiration of the relevant limitation period are to be treated as overpayments. Those are statutory overpayments.

[33]See, e.g., Pettibone v. United States, 34 F.3d 536, 538, Doc 94-8454, 94 TNT 182-15 (7th Cir. 1994) (“The Internal Revenue Code leaves to the Commissioner’s discretion whether to apply overpayments to delinquencies or to refund them to the taxpayer. Until the Commissioner exercises this discretion, the taxpayer has no right to payment”) (internal citations omitted); In re Luongo, 259 F.3d 323, Doc 2001-23380, 2001 TNT 175-46 (5th Cir. 2001) (mere overpayment could not constitute “property of the estate” within meaning of Bankruptcy Code section 541 because debtor had no interest in the overpayment until after the IRS had exercised its discretion to offset overpayment against other tax liabilities).

[34]Refunds are deemed made as of the date that an authorized IRS employee signs a schedule of overassessments, anothersummary document similar to the summary record of assessments. Reg. section 301.6407-1. That document used to be Form 1166 but can now be various other forms, some computergenerated. See Rev. Rul. 2001-40, 2001-2 C.B. 276, Doc 2001- 23872, 2001 TNT 180-10 (modifying Rev. Rul. 78-127, 1978-1 C.B. 436). The signing date is chiefly important for calculating interest, if any, due on refunds.

[35]There is an open question whether the IRS’s erroneous determination that a prior assessment was illegal – and thus abated under section 6404(a)(3) – gives rise to a rebate or nonrebate refund. Discussion of that point is beyond the scope of this article.

[36]See United States v. Frontone, 383 F.3d 656, Doc 2004-18042, 2004 TNT 176-13 (7th Cir. 2004) (discussing distinction).

[37]The earliest case I can find is Carney Coal Co. v. Commissioner, 10 B.T.A. 1397 (1928), and the latest case to reiterate the rule is In re Becker, 407 F.3d 89, 97 (2d Cir. 2005) (“If an assessment is properly abated pursuant to subsection (a)(1), (2), or (2) of I.R.C. ß 6404, quoted above – all of which pertain to assessments made in error – the abatement entirely extinguishes the assessment. In order to undo that abatement, the IRS would be required to impose a new assessment; and, to be effective, that new assessment would need to be imposed within the limitations period”); see also GCM 36263, “Legality of Overpayment Offsets to Collect Unassessable Erroneous Refunds” (1975) (reviewing case law).

[38]United States v. Wurts, 303 U.S. 414, 415 (1938).

[39]United States v. Greene-Thapedi, 398 F.3d 635, Doc 2005-3339, 2005 TNT 33-17 (7th Cir. 2005); United States v. Commonwealth Energy, 235 F.3d 11, Doc 2001-145, 2000 TNT 248-72 (1st Cir. 2000). That is consistent with the rule that interest owed on erroneous refunds starts on the date the taxpayer cashes or negotiates the check. United States v. Mallah, 882 F. Supp. 779, Doc 95-1847, 95 TNT 26-34 (S.D. Ind. 1995); La Follette v. United States, 173 F. Supp. 388 (S.D. Cal. 1959). Accordingly, taxpayers who receive an erroneous refund should neither cash the check nor send it back, but should instead hold onto it, inform the IRS of the error, and wait for instructions. They should send it back only to someone who knows it is coming and knows how to process it. Sending it back blind just invites additional error.

[40]United States v. McRee, 7 F.3d 976 (11th Cir. 1996) (en banc) (upholding conviction under 18 U.S.C. section 641 of a taxpayer who, even though doing nothing to induce the erroneous refund, took elaborate actions to disperse the $350,000 refund check in offshore accounts).

[41]See, e.g., United States v. Daum, 968 F. Supp. 1037, Doc 98-7010, 98 TNT 35-36 (W.D. Pa. 1997) (error discovered in Feb. 1995, moral suasion letter issued that Mar., followed by suit in Apr., and motion for summary judgment in Oct.; summary judgment granted 18 months later, on Apr. 30, 1997).

[42]See, e.g., In re Able Roofing & Sheet Metal Co., 425 F.2d 699, 701 (5th Cir. 1970) (refusing to give a section 7405 claim the same priority in bankruptcy as taxes because “a claim for refund erroneously made does not create a liability for taxes”). The status of a section 7405 judgment is not settled but treating it as a nontax debt is the most consistent result from the string of cases that I discuss below denying the IRS the ability to administratively recover nonrebate erroneous refunds.

[43]The smallest amount I could find, in an admittedly quick search, was at issue in United States v. Korda, 2005-2 U.S.T.C. para. 50,541 (M.D. Fla. 2005) ($10,209.81).

[44]GCM 36263 (May 9, 1975), as modified by GCM 36624 (Mar. 11, 1976) (“We regret that our reply to your inquiry of June 17, 1970 has been so long delayed”). I would sure love to know the story behind that unusual delay. The 1975 GCM thought that the IRS had to reassess, but the 1976 GCM concluded that to require reassessment made no sense for nonrebate refunds because the error causing the refund could not undo the prior assessment and the IRS could make only one assessment of a tax liability.

[45]GCM 36624 (Mar. 11, 1976).

[46]Recall that, as discussed above in note 16, the distinction between rebate and nonrebate refunds did not arise until after 1944. The GCMs pointed out that before 1944 the IRS had established the right to reassess erroneous refunds through the deficiency procedures. They reasoned that when Congress put the distinction between rebate and nonrebate refunds into section 6211’s definition of deficiency, it could not have meant to suddenly deny what had previously been allowed: the ability to rectify the error administratively. If the IRS no longer had to reassess the nonrebate liability, that meant the original assessment’s power to unleash the administrative collection tools was revived.

[47]See GCM 36624 (“We believe it would be particularly desirable to publish a decision to recover non-rebate erroneous refunds through the usual tax collection procedures, since this represents a change in past practice”). The IRS eventually put the procedure to recover nonrebate erroneous refunds in IRM I can no longer find that section of the IRM online.

[48]This reasoning is best seen in Groetzinger v. Commissioner, 69 T.C. 309 (1977).

[49]Bugge, supra note 28.

[50]Davenport v. United States, 136 B.R. 125 (W.D. Ky. 1991); Sanfellipo v. United States, 1990 U.S. Dist. LEXIS 18654 (N.D. Cal. 1990); and Groetzinger v. Commissioner, 69 T.C. 309 (1977). One district judge bravely rejected the opinions of five courts of appeals, only to have the government abandon the argument on appeal. See Mildred Cotler Trust v. United States, 2 F. Supp.2d 264, 270, Doc 98-11750, 98 TNT 67-10 (E.D.N.Y. 1998), rev’d, 184 F.3d 168, 171, Doc 1999-23590, 1999 TNT 132-10 (2d Cir. 1999) (“I recognize that a number of courts have reached a different conclusion in similar or related situations. Relying on a doctrine known as “extinguishment,’ these courts have concluded that a payment of taxes “extinguishes’ an underlying assessment, thus barring collection of taxes on the basis of that assessment. I find these opinions to be unpersuasive”) (citations omitted).

[51]30 F.2d 193 (9th Cir. 1929). There, the IRS had erroneously refunded an estate tax and brought suit, on the equity side of the court, to recover the amount as an unpaid tax. The circuit court dismissed the suit because the government had an adequate remedy at law – being an action under the predecessor to section 7405 for recovery of an erroneous refund. Thus, the court made the critical distinction between liability for tax and liability for return of erroneously paid money, a distinction picked up later by the Seventh Circuit in O’Bryant.

[52]158 F. Supp. 793, 795 (N.D. Tex. 1958) (“When the plaintiffs paid their respective income taxes for the year 1943, such taxes as to the plaintiffs were extinguished and the subsequent refunds to the plaintiffs of a portion or all of the money paid by them as 1943 income taxes did not restore the taxes”).

[53]629 F. Supp. 333, 344 (N.D. Ill. 1986).

[54]The circuits in which the IRS argued for the theory and lost were, in chronological order of the opinions being issued: United States v. Wilkes, 946 F.2d 1143, 1152 (5th Cir. 1991); O’Bryant v. United States, 49 F.3d 340, Doc 95-3184, 95 TNT 57-15 (7th Cir. 1995); Clark v. United States, 63 F.3d 83, Doc 95-8258, 95 TNT 171-10 (1st Cir. 1995); Bilzerian v. United States, 86 F.3d 1067, Doc 96-19205, 96 TNT 130-4 (11th Cir. 1996); Singleton v. United States, 128 F.3d 833 (4th Cir. 1997).

[55]AOD GL-118964-97, 1998-1 C.B. 972 (action only), 1998 AOD LEXIS 8 (full memo). The acquiescence was for Bilzerian.

[56]Mildred Cotler Trust v. United States, 184 F.3d 168, 171 (2d Cir. 1999), reversing 2 F. Supp.2d 264 (E.D.N.Y. 1998) (“On appeal, the government has expressly abandoned its argument below, on which the district court relied”); Stanley v. United States, 140 F.3d 1023, 1027-1028, Doc 98-11458, 98 TNT 65-62 (Fed. Cir. 1998), reversing 35 Fed. Cl. 493, Doc 96-13281, 96 TNT 88-15 (Ct. Fed. Cl. 1996) (“the Government argued before the Court of Federal Claims that the erroneous refund could nevertheless be recovered on the basis of that [previous] assessment. . . . The Government on appeal wisely does not pursue this argument”). The new IRM provisions are at 25.6.7.

[57]The IRS still has a limited ability to collect the erroneous refund through setoff, but courts do not permit setoffs after the applicable section 6532(b) limitations period for filing an erroneous refund recovery suit has expired. PG&E v. United States, 417 F.3d 1375, Doc 2005-17029, 2005 TNT 154-7 (Fed. Cir. 2005).

[58]Some examples are Stanley, supra note 56 ($600,000 + windfall); O’Bryant, supra note 54 ($7,000 + windfall); Wilkes, supra note 54 ($20,000 + windfall); Clark, supra note 54 ($25,000 + windfall); Bilzerian, supra note 54 ($125,000 + windfall); Mildred Cotler Trust, supra note 56 ($175,000 + windfall).

[59]49 F.3d at 347-348 (citations omitted). Accord, Clark, supra note 54, at 87 (“there is a fundamental difference between money taxpayers possess as the result of an erroneous refund and money they originally owed the IRS (their tax liability); taxpayers who received erroneous refunds owe the IRS because they have been unjustly enriched by it, not because they have not paid their taxes”) (quotations omitted); Bilzerian, supra note 54, at 1069 (“Today, we join these circuits and hold that once a tax liability is paid, no erroneous refund – whether rebate or non-rebate – can revive it”). But see United States v. Frontone, supra note 36 (answering yes to the question “whether a claim for taxes based on an erroneous refund is a claim for taxes” and noting that “we acknowledge the tension between O’Bryant’s conception of when assessment is available [for erroneous refunds] and the broader conception suggested by Bilzerian and Clark”) (citations omitted).

[60]Singleton v. United States, supra note 17, at 837. In Singleton the IRS did try to make a supplemental assessment under section 6204. What the majority got wrong was reading section 6204 as always requiring the IRS to use deficiency procedures for all supplemental assessments. As the dissent correctly pointed out: “The Tax Code does not prohibit supplemental assessments without a notice of deficiency. It prohibits supplemental assessments of a deficiency without a notice of deficiency.” Id. at 840. But from the facts as recited in the case, it appears the majority correctly identified the error as substantive. The taxpayer had included a schedule reporting a business tax credit but had correctly not taken the credit against the reported tax liability. The error correction/reject unit in the processing function determined that the taxpayer had made a mistake and so allowed the credit and assessed a much smaller tax liability than reported. That determination was an error. But it was substantive. Accordingly, the IRS was required to follow the deficiency procedures before making the supplemental assessment.

[61]Section 6201(a).

[62]O’Bryant v. United States, 839 F. Supp. 1321, 1325 (C.D. Ill. 1993), aff’d, 49 F.3d 340 (7th Cir. 1995) (“Most courts hold that the government is not limited to a Section 7405 action to recover an erroneous refund, but may collect by assessment in the ordinary manner, because the refund creates an underpayment”) (citations omitted) (collecting cases); see also Bilzerian v. United States, 887 F. Supp. 1509, 1514 (M.D. Fla. 1995), rev’d on other grounds, 86 F.3d 1067 (11th Cir. 1996) (The IRS “is not limited to filing an action under ß 7405 to recover an erroneous refund. Where the IRS has made [a] new assessment [of] the erroneously refunded amount, the IRS may collect this amount through ordinary collection procedures within a ten (10) year period after the assessment of the tax. 26 U.S.C. ß 6502”) (citations omitted); Purcella v. United States, 1992 U.S. Dist. LEXIS 426 (D. Colo. 1992) (“The IRS could have recovered the [erroneous] refund by assessment in the ordinary manner. Yet it never assessed the [erroneous] refund as a tax and the previous assessment had already been satisfied”) (citations omitted).

[63]S. Rep. No. 960, 70th Cong., 1st Sess. 42 (1928).

[64]See note 16 supra.

[65]See note 37 supra. The requirement to reassess was independent of the process required for reassessment. For example, courts permitted the IRS to use the summary process to reassess erroneous refunds of excise taxes. United States v. Tuthill Spring Co., 55 F.2d 415 (N.D. Ill. 1931). Recall that assessment of excise taxes never requires the IRS to follow the deficiency procedures, which apply only to income, estate, and gift taxes.

[66]The 1975 GCM lists the additional cases upholding the recovery of erroneous rebate refunds by reassessment: Burnet v. Porter, 283 U.S. 230 (1931); Clark v. Comm’r, 158 F.2d 851 (6th Cir. 1946); Comm’r v. Newport Industries, 121 F.2d 655, 657 (7th Cir. 1941); Page v. LafayetteWorsted, 66 F.2d 339 (1st Cir. 1933); Richard E. Warner, T.C. Memo. 1974-243; Lucy L. Lawton, 16 T.C. 725 (1951); Etta Craig, 18 B.T.A. 86 (1929).

[67]The O’Bryant court disregarded that argument, snapping, “It is an unjustified leap of logic to say that because nonrebate refunds cannot be recovered by reassessment, they must be collectible by resort to the original assessment. There is no indication in the Code that Congress intended such a result and we refuse to reach it, especially when doing so would require us to mischaracterize an erroneous refund as a tax liability.” 49 F.3d at 347.

[68]United States v. National Bank of Commerce, 472 U.S. 713, 719-720 (1985) (proper assessment); Blackston v. United States, 778 F. Supp. 244 (D. Md. 1991) (notice and demand); United States v. Wintner, 200 F. Supp. 157 (N.D. Ohio 1961), aff’d, 312 F.2d 749 (6th Cir. 1963), rev’d on other issues, 375 U.S. 393 (1964) (failure to pay).

[69]This is another area in which the tax code needs rewriting. Section 6322 provides that the tax lien is extinguished when the taxpayer’s liability is satisfied. Likewise, section 6325(a) requires the IRS to issue a certificate of release when it determines that a tax liability is satisfied. However, section 6432 provides for damages only when the IRS negligently or knowingly failed to release a lien within 30 days of determining that the related assessment is satisfied. Thus, current law leads to the anomalous result that the tax lien may be extinguished but the IRS incurs no penalty for not issuing a certificate of release, leaving in place the notice of federal tax lien declaring to the world that the tax lien still exists. See Henderson v. United States, 91 F. Supp.2d 995, Doc 2000-23495, 2000 TNT 176-67 (E.D. Wis. 2000) (no damages to the taxpayer who claimed to have satisfied his liability but concededly did not satisfy the erroneously assessed amount).

[70]While there are provisions for evaluating offers in compro- mise during examination, IRC (“Offer in Compromise Filed During Audit”) and (same), those provisions are used only in exceptional circumstances. See IRM (stating that considering collection aspects of a deficiency determination should be “highly unusual”).

[71]See H. Rep. No. 1337, 83d Cong., 2d Sess. (1954) at A404; S. Rep. No. 1622, 83d Cong., 2d Sess. (1954) at 572.

[72]The proposal also parallels how the IRS uses the section 6672 trust fund recovery penalty.

Previously published by the University of Wisconsin – Law School, January 2008

George H. Mahon Professor of Law, Texas Tech University School of Law