concurring in the result.
With respect, I concur only in the result.
On its face, this case is abstract and unapproachable; few people have ever looked at subchapter K of the tax code, let alone familiarized themselves with terms like “notice of deficiency,” “computational adjustment,” or the difference between “partnership items” and “nonpartnership-items.” I write separately to emphasize the importance of the issues at play and ramifications of the majority opinion. Ultimately, the majority and I reach the same result. However, I do so by simply applying the statute; in my view, the statute is clear, in need of no interpretive gloss or embellishment. I would thus adopt the logic and careful reasoning of the U.S. Court of Federal Claims. The majority, by contrast, reaches the right result by misreading the statute, then creating a new harmless error “exception” never raised or advocated by anyone. What the majority decides has implications far beyond the outcome here.1 Particularly in tax law, it is unwise and even perilous to craft a new rule that will likely affect future cases in unpredictable ways. For these reasons, I concur only in the result.
This case is about a taxpayer’s right to know what taxes the government says he owes, but which the taxpayer has not paid and may not know are owed. The majority decides that a taxpayer who enters a settlement agreement with the Internal Revenue Service (“IRS”) nonetheless has an unrestricted right to challenge issues resolved by that agreement again, in Tax Court.
In addition to allowing a taxpayer to selectively relitigate (and potentially invalidate) portions of a settlement he dislikes, the majority creates a harmless error exception.2 This exception is broad and *1367sweeping. It applies whenever the IRS fails to give a taxpayer notice that he owes more taxes. In practice, the majority’s new rule means the IRS no longer has to honor a taxpayer’s statutorily-guaranteed right to notice or follow the tax code’s detailed notice procedures. See I.R.C. § 6212. Instead, the IRS is free to assess and collect the allegedly due taxes. See id. §§ 6201, 6203, 6212-6213. This new rule conflicts with the conclusions of our circuit and every other circuit to have considered the issue. See Olson, 172 F.3d at 1317 (“Under the non-partnership ‘standard’ deficiency procedures of the Code, the IRS may not assess additional tax without first mailing a notice of deficiency to a taxpayer whose taxes have not been paid in full.” (emphasis added)). Our sister circuits have read these provisions as creating a categorical, nonnegotiable notice requirement, denial of which can never be harmless.3
Thus, while we all agree on the ultimate result in this case, I take strong issue with the path the majority takes to get there. First, I disagree with the majority’s construction of “computational adjustment” in I.R.C. § 6230 and § 6231 because it reads these sections to unambiguously contain an additional requirement — change in a partnership item — contrary to the statutes’ plain language, as well as common sense. Second, I object to the majority’s treatment of binding circuit precedent, namely, this court’s decision in Olson v. United States, 172 F.3d 1311 (Fed.Cir.1999). The majority interprets the same statutory phrase at issue in Olson but reaches a far different, conflicting construction. Substantively, I believe Olson is correct and the majority is wrong. Pragmatically, I doubt these opinions can coexist: they leave this court and future litigants the impossible task of reconciling decisions that reach contrary results on nearly identical facts. The majority’s gossamer distinction from Olson dazzles only at a distance; scrutiny reveals it to overlook the heart of the (conflicting) analysis and result.
Finally, I strenuously object to the majority’s creation of a new, sweeping harmless error rule. This is a rule no party sought or supported — not even the government, its beneficiary.4 More importantly, *1368the rule is legally unsound and practically harmful. The majority derives its new rule from 28 U.S.C. § 2111. There is little reason to think, however, that this “general” harmless error rule trumps the more specific provisions of the tax code, which creates no such exception and arguably forecloses the majority’s approach. In effect, the majority’s new harmless error rule largely nullifies one of the most basic, important protections a taxpayer has against the government: the right to notice. This right is not simply a procedural formality; it is how a taxpayer knows the IRS is planning to collect taxes. I.R.C. §§ 6211-6213. A “notice of deficiency” tells the taxpayer how much he owes and why the IRS thinks he owes it. Id.; see also Abrams v. Comm’r, 814 F.2d 1356, 1357 (9th Cir.1987). Under the majority’s new harmless error rule, the IRS can proceed with collection without telling the taxpayer what it plans to do. Because receiving this notice is a prerequisite for filing suit in Tax Court, the majority’s rule also deprives taxpayers of the only pre-payment, pre-assessment forum in which to challenge the IRS. See I.R.C. § 6213(a); Laing v. United States, 423 U.S. 161, 165 n. 4, 96 S.Ct. 473, 46 L.Ed.2d 416 (“A deficiency notice is of import primarily because it is a jurisdictional prerequisite to a taxpayer’s suit in the Tax Court .... ”); see also Desmet v. Comm’r, 581 F.3d 297, 302-03 (6th Cir.2009); Abrams, 814 F.2d at 1357; Gaska v. Comm’r, 800 F.2d 633, 635 (6th Cir.1986). Whether a taxpayer has to pay the taxes allegedly owed before getting to challenge whether he owes anything may not matter if the amount is small, but may matter a great deal if the sum is large or the taxpayer poor. Even if some errors in a notice of deficiency may be subject to harmless error analysis under § 2111, it is hard to see how the error alleged here — failure to issue any notice— can ever be harmless.
These concerns are discussed in more detail below.
I. THE STATUTE
My first disagreement is with the majority’s construction of “computational adjustment” in I.R.C. § 6230. In my view, the Court of Federal Claims got it right.
Section 6230 specifies when the IRS must issue a notice of deficiency to the taxpayer before assessing and collecting taxes. I.R.C. §§ 6213(a), 6230. It also states when the IRS is not required to issue a notice of deficiency. For our purposes, the relevant exception is for “computational adjustments.” If the taxes the IRS seeks to assess are “computational adjustments,” the IRS may assess and collect the taxes without issuing a notice of deficiency. Id. § 6230(a)(1). The tax code defines computational adjustment as “the change in the tax liability of a partner which properly reflects the treatment under this subchapter [C] of a partnership item.” I.R.C. § 6231(a)(6) (emphasis added).
I would affirm the Court of Federal Claims’ careful, well-reasoned construction of “computational adjustment.” See Bush v. United States, 78 Fed.Cl. 76, 81-84 (2007). I would hold that a computational adjustment includes any (1) change in a partner’s tax liability (2) which correctly applies subchapter C(3) to a partnership item. I.R.C. § 6231(a)(6); see also 26 C.F.R. § 301.6231(a)(6)-1 T. That is what § 6231(a)(6) says, and I see no reason to depart from it.
Based on this construction, I would further hold that the changes in the tax lia*1369bility of Bush and Shelton were computational adjustments because the changes reflected how Bush and Shelton had to report certain partnership losses (a partnership item) on their returns. See 26 C.F.R. § 301.6231(a)(3)-l (defining “partnership item”). This change in how partnership losses should be reported, or “treat[ed],” resulted from the taxpayers’ settlement agreements with the IRS. Because no additional fact finding was required to compute their tax liability, I agree with the Court of Federal Claims that a notice of deficiency was not required, meaning the tax assessments were valid. In holding that the taxpayers were thus not entitled to a tax refund, I would rely on what the statute says, not on a new harmless error exception, as the majority does.
The majority reads the definition of computational adjustment to mean that only “change[s] ... [to] a partnership item” are computational adjustments. Maj. Op. at 1360 (“[T]here were no changes to their partnership items ... [thus, t]he assessments were not computational adjustments ...(emphases added)).5 I part ways with the majority because that is not what the statute says. Grammatically, the majority is correct that “of a partnership item” could modify “change” in § 6231(a)(6), but context suggests this interpretation makes little sense. A rule of grammar is that subordinate clauses “should be placed near the words they modify.” Margaret Shertzer, The Elements of Grammar 47 (1986). Similarly, a principle of statutory construction is that “[r]eferential and qualifying phrases, where no contrary intention appears, refer solely to the last antecedent.” 2A Norman J. Singer & Shambie Singer, Sutherland Statutes and Statutory Construction § 47.33 (7th ed.2009). That is why we say, “Flying over Washington, I saw the Lincoln Memorial,” not “I saw the Lincoln Memorial flying over Washington.”
The definition of “computational adjustment” is certainly not the model of clarity. To blame are the numerous qualifying phrases — “in the tax liability,” “of a partner,” “which properly reflects the treatment,” “under this subchapter,” and “of a partnership item.” I.R.C. § 6231(a)(6). The challenge is to determine which phrase belongs to what. When the above guidelines are applied to § 6231(a)(6), they suggest the “change” is to “the tax liability of a partner,” not in a “partnership item.” Tellingly, the phrase “partnership item” appears nowhere near the word “change”; it is part of a phrase qualifying something else entirely, “treatment.”
It is the dependent clause, “which properly reflects the treatment under this subchapter of a partnership item,” that requires the change (in tax liability) to accord with how partnership items are “treat[ed]” under subchapter C. See I.R.C. §§ 6221 (requiring the “tax treatment of any partnership item” to be determined “at the partnership level”), 6229, 6231; see also Keener v. United States, 551 F.3d 1358, 1364 (Fed.Cir.2009) (noting *1370that subchapter C, “Tax Treatment of Partnership Items,” contains rules for “dealing with” partnership items).
Requiring the proper “treatment” of a partnership item, as the statute says, is not the same thing as requiring a “change” in the value of a partnership item, as the majority holds. Maj. Op. at 1360-61. There are many instances where a TEFRA proceeding or settlement may not change the value of partnership items, such as the partnership’s income or loss. As occurred here, however, the TEFRA proceeding or settlement may affect how an individual partner reports, or “treats,” partnership items on his individual return, resulting in a change in tax liability. In this case, the partner’s individual, at-risk amounts affected how much of the partnership losses Bush and Shelton could claim on their returns. The amount of the partnership item, here partnership losses, remained the same; what changed was how those losses were “treated” (reported) on Bush’s and Shelton’s individual returns, resulting in a corresponding change to them tax liability.
Disregarding what the statute says, the majority limits the meaning of “treatment” by equating it to “change.” But that is not the word Congress chose.6 The word “treatment” is broad, but understandably so, given that the tax code performs an enormous, mixed bag of functions — from categorizing items as “income” or “loss,” to determining whether an item warrants a tax credit, deduction, or additional tax. It is thus not surprising that Congress has used “treatment” often, but not because it really meant “change.” See, e.g., I.R.C. §§ 1361-1379 (subchapter S, “Tax Treatment of S Corporations and Their Shareholders”), 4462, 6211-6234 (subchapter C, “Tax Treatment of Partnership Items”). Here, there is no evidence that Congress intended the phrase “treatment under this subchapter of a partnership item” to mean less than its naturally broad and inclusive meaning. Cf. Reiter v. Sonotone Corp., 442 U.S. 330, 338-39, 99 S.Ct. 2326, 60 L.Ed.2d 931 (1979). This court has consistently construed “treatment” in tax law broadly, to encompass everything from the procedures for calculating income or loss, to the criteria for deciding whether a company is subject to double or single taxation.7
The majority fails to give proper weight to the words Congress chose, or to the *1371distinction Congress drew by using different words in I.R.C. § 6231(a)(6). Badaracco v. Comm’r, 464 U.S. 386, 398, 104 S.Ct. 756, 78 L.Ed.2d 549 (1984) (“Courts are not authorized to rewrite a statute because they might deem its effect susceptible of improvement.”); Kirkendall v. Dep’t of the Army, 479 F.3d 830, 857-58 (Fed.Cir.2007); see also Sosa v. Alvarez-Machain, 542 U.S. 692, 711 n. 9, 124 S.Ct. 2739, 159 L.Ed.2d 718 (2004) (“[W]hen the legislature uses certain language in one part of the statute and different language in another, the court assumes different meanings were intended.” (quotation marks omitted)); 2A Norman J. Singer & Shambie Singer, Sutherland Statutes and Statutory Construction § 46:6 (7th ed.2009). Here, there is no reason to think that Congress used “treatment” to mean “change.” Russello v. United States, 464 U.S. 16, 23, 104 S.Ct. 296, 78 L.Ed.2d 17 (1983) (“We would not presume to ascribe this difference [in language between subsections] to a simple mistake in draftsmanship.”); see also Barnhart, 534 U.S. at 454, 122 S.Ct. 941. Had Congress intended to limit “computational adjustments” to tax liability changes arising from changes in a partnership item, it could have used “change” again, rather than “treatment.” The majority errs by rewriting the definition of “computational adjustment” to include this new limitation.
Neighboring provisions in § 6231 provide similarly strong evidence that “treatment” in § 6231(a)(6)’s definition of computational adjustment does not necessarily require “change” in a partnership item. Subsection (a)(4) of § 6231 defines “non-partnership item” as any “item which is (or is treated as) not a partnership item.” Subsection (a)(12) of § 6231 provides that “[e]xcept to the extent otherwise provided in the regulations, a husband and wife who have a joint interest in a partnership shall be treated as 1 person.” In the context of (a)(4) and (a)(12), it would be nonsensical to read “treated” to mean “changed.” The “normal rule” of statutory interpretation is that “identical words used in different parts of the same statute are generally presumed to have the same meaning.” IBP, Inc. v. Alvarez, 546 U.S. 21, 34, 126 S.Ct. 514, 163 L.Ed.2d 288 (2005); see also Nijhawan v. Holder, — U.S. -, -, 129 S.Ct. 2294, 2301, 174 L.Ed.2d 22 (2009); Sullivan v. Stroop, 496 U.S. 478, 484, 110 S.Ct. 2499, 110 L.Ed.2d 438 (1990). As used in (a)(4) and (a)(12) of § 6231, “treated” is better read to mean how tax liability should be computed under the tax code — further evidence that “computational adjustment” in (a)(6) of § 6231 does not require any change in a partnership item, but only correct application of subchapter C. See Finnegan v. Leu, 456 U.S. 431, 438 & n. 9, 102 S.Ct. 1867, 72 L.Ed.2d 239 (1982) (reading “virtually identical language” to have the same meaning); Tunik v. Merit Sys. Prot. Bd., 407 F.3d 1326, 1347-48 (Fed.Cir.2005).
II. OLSON V. UNITED STATES
An independent, though equally compelling, reason why I concur only in the result is the conflict the majority creates with this court’s decision in Olson v. United States, 172 F.3d 1311 (Fed.Cir.1999). Though the majority says otherwise, Olson already considered and decided what constitutes a “computational adjustment” un*1372der I.R.C. §§ 6230 and 6231. See Maj. Op. at 1360-61; Olson, 172 F.3d at 1317-18.
Olson holds that assessments are “computational adjustments” when they require “no individualized factual determinations” as to the correctness of the original partnership items or “any other factual matters such as the state of mind of the taxpayer upon filing.” 172 F.3d at 1318. Under Olson, when the “critical” questions of fact “ha[ve] already been resolved,” then “application of that stipulated fact to the tax returns in question requires only computational action.” Id. Facts can be resolved, for example, by a taxpayer’s settlement agreement with the IRS, which “concede[s] that they ha[ve] no entitlement” to certain tax credits. Id.
Under Olson, the correct result in this case is exactly the opposite of what the majority reaches. As in Olson, Bush and Shelton entered into settlement agreements with the IRS. These agreements provided that Bush and Shelton could not claim losses from certain investments. The IRS subsequently recomputed the taxes owed by Bush and Shelton. These adjustments to their tax liability were based on the settlement agreements, “entail[ing] nothing more than reviewing the taxpayers’ returns for the years in question, striking out the tax credits that had been improperly claimed, and re-summing the remaining figures.” Id. Olson holds that applying stipulated facts in this fashion is a prototypical “computational adjustment” under §§ 6230 and 6231. Giving Olson the controlling weight it is due, we should be affirming the Court of Federal Claims’ conclusion that the assessments here were “computational adjustments,” exempt from the general notice of deficiency requirement.
Instead, the majority claims Olson does not apply because the settlement agreement in Olson concerned partnership items, whereas the settlement agreements in this case do not. Maj. Op. at 1360-61. This is a distinction that makes no difference, because the relevant reasoning in Olson is not so limited.
Olson set out the analytical framework for deciding what constitutes a “computational adjustment” whenever there is a settlement agreement, not only when that agreement pertains to particular items. 172 F.3d at 1317-18. The majority’s rule undercuts the purpose of TEFRA in exactly the way Olson feared. See id. at 1318. By requiring a notice of deficiency, the majority does the opposite of streamlining partnership proceedings, instead giving the taxpayer an unnecessary, unwise opportunity to attack the taxes assessed. The opportunity is unnecessary because any assessments are based on a settlement the taxpayer signed, which the taxpayer acknowledges may give rise to future tax liabilities. The opportunity is unwise because it allows the taxpayer to selectively attack settlement provisions he dislikes by challenging the resulting assessments. Such attacks are possible because a notice of deficiency allows the taxpayer to go to Tax Court, where he can relitigate the amount of tax owed and the reason he owes it, even when the settlement fully resolves the issue. See I.R.C. §§ 6211-6213.
The decisions of other courts agree with Olson and conflict with the majority. None read § 6231(a)(6) as the majority does, instead defining “computational adjustment” as tax assessments that require no additional factual determinations to determine the individual’s tax liability. See, e.g., Desmet, 581 F.3d at 303-04; Calla-*1373way v. Comm’r, 231 F.3d 106, 109-10 & n. 4 (2d Cir.2000) (holding that “where no further factual determinations are necessary at the partner level, an assessment attributable to an ‘affected item’ may also be made by computational adjustment” because determining the change in tax liability “is a mathematical calculation and requires no further factfinding”); see also Bob Hamric Chevrolet, Inc. v. United States, 849 F.Supp. 500, 510 (W.D.Tex.1994) (holding that “a settlement is usually applied to a partner by means of a computational adjustment and not under the ordinary deficiency and refund procedures”); Harris v. Comm’r, 99 T.C. 121, 126, 1992 WL 176438 (1992); Powell v. Comm’r, 96 T.C. 707, 712, 1991 WL 80646 (1991); N.C.F. Energy Partners v. Comm’r, 89 T.C. 741, 1987 WL 45298 (1987) (superseded on other grounds by the Taxpayer Relief Act of 1997, Pub.L. No. 105-34, § 1238(a), 11 Stat. 126).
These decisions reaffirm my conviction that the interpretation of “computational adjustment” in Olson is correct. The majority errs by failing to follow Olson and by creating a distinction with no basis in the statute.
III. NEW HARMLESS ERROR EXCEPTION
The majority only reaches the right result because it creates a new harmless error exception. This rule was not advocated by either party in this court or below, even though the government stands to benefit from it. The majority holds that even when a notice of deficiency is required under § 6230, the IRS’s failure to issue such a notice will be excused if this failure was not “prejudicial.” Maj. Op. at 1361-62, 1364. The majority defines “prejudicial” as an error that could affect the “outcome,” meaning it results in an incorrect assessment or collection of taxes. Id. at 1364-65.
In my view, the majority errs by inventing a harmless error exception for “procedural” errors by the IRS. Maj. Op. at 1364-65. Though the majority may intend its holding to be “limited,” I question whether the test can, or will, be limited to the facts of this case. Cf. Maj. Op. at 1365-66. Though the majority does not discuss them, the implications of this new rule are sweeping: So long as the IRS is right that more taxes are owed, it can always bypass the notice of deficiency procedures in I.R.C. §§ 6211-6216. As a practical matter, taxpayers will rarely, if ever, be able to satisfy the majority’s test, which requires showing that litigating in Tax Court “could have resulted in a different outcome” than a refund suit in district court or the Court of Federal Claims. Maj. Op. at 1365. Indeed, by design the primary difference between a Tax Court action and a refund suit is whether the taxpayer has to pay the disputed taxes before litigating. Compare I.R.C. §§ 6213, 6512, with I.R.C. §§ 6227, 7422. This is one of the rare cases where the exception may indeed consume the rule, namely the procedures set out in I.R.C. §§ 6211-6216.8
More importantly, the majority’s exception has no textual basis in the tax code. *1374Though the majority tenuously relies on the general harmless error standard in 28 U.S.C. § 2111,9 properly applied, § 2111 changes nothing. Maj. Op. at 1363-65. It does not change the result here because I.R.C. §§ 6211-6216 make clear that a taxpayer has an unqualified right to a notice of deficiency. Until the IRS issues the taxpayer such a notice, the IRS cannot assess or collect that deficiency. I.R.C. § 6213(a); see also Frontone, 383 F.3d at 658; Eschweiler v. United States, 946 F.2d 45, 48 (7th Cir.1991) (noting the IRS’s “statutory obligation to send notice of the deficiency”).
The majority’s mistake is to treat the notice of deficiency as an empty procedural gesture, like using the right size paper or filing enough copies of an appellate brief. See Maj. Op. at 1363-65; see, e.g., Fed. R.App. P. 32; Ninth Cir. R. 32-2. As with most tax laws, here the “substantial right” at stake is the taxpayer’s right to have the IRS follow particular procedures before assessing or collecting taxes. See I.R.C. §§ 6212-6213. The majority incorrectly assumes that the IRS’s failure to follow the procedures Congress established only matters if that failure could affect the outcome, the tax assessed. Maj. Op. at 1361-65. Focusing on whether the outcome (tax assessed) is right or wrong ignores the nature of the right at stake.
Tax laws are technical laws. They are not subject to the general principles of equity as the majority now holds; instead, they require strict adherence to the explicit procedures they establish. See United States v. Dalm, 494 U.S. 596, 608, 110 S.Ct. 1361, 108 L.Ed.2d 548 (1990); Lewyt Corp. v. Comm’r, 349 U.S. 237, 249, 75 S.Ct. 736, 99 L.Ed. 1029 (1955); Oropallo v. United States, 994 F.2d 25, 28 n. 3 (1st Cir.1993); In re Graham, 981 F.2d 1135, 1138 (10th Cir.1992); Ewing v. United States, 914 F.2d 499, 501 (4th Cir.1990); Richardson v. Smith, 301 F.2d 305, 306 (3d Cir.1962) (noting that “taxation is a game which must be played strictly in accordance with the rules”). Because the right at issue is partially procedural, much of the “harm” is deprivation of that procedure. Cf. Hughes v. Rowe, 449 U.S. 5, 13 & n. 2, 101 S.Ct. 173, 66 L.Ed.2d 163 (1980); Carey v. Piphus, 435 U.S. 247, 266, 98 S.Ct. 1042, 55 L.Ed.2d 252 (1978) (holding that deprivation of a party’s procedural due process rights is actionable, even without injury).
Here, however, the harm is greater than just a loss of procedure because receiving a notice of deficiency is prerequisite to filing in Tax Court. When taxpayers are entitled to a notice of deficiency and receive none, they lose their statutory right to a prepayment, pre-assessment forum (Tax Court) due to the IRS’s error, thus depriving them of a substantial right. See I.R.C. § 6213(a). The majority misses the point by framing the issue as whether “there were any issues that [the taxpayers] were unable to litigate in the context of the *1375refund suit that they would have been able to litigate in a Tax Court suit.” Maj. Op. at 1364-65. The tax code does not make a taxpayer’s right to sue in Tax Court depend on whether the outcome could have been different, as the majority suggests. Id. at 1364-65.
The majority attempts to support its new harmless error rule through misplaced reliance on Elings v. Commissioner, 324 F.3d 1110 (9th Cir.2003), and Smith v. Commissioner, 275 F.3d 912 (10th Cir.2001). These cases do not help the majority, because none involve the failure to give the taxpayer notice or follow the tax code’s procedures. Maj. Op. at 1364-65. In those cases, unlike here, the IRS followed the required procedure: it issued a notice of deficiency. See Elings, 324 F.3d at 1111; Smith 275 F.3d at 913-14. The only question was whether errors in those notices were harmless. The tax code expressly provides that errors in a notice of deficiency will generally not invalidate that notice. I.R.C. § 7522(a). Because of I.R.C. § 7522(a), Elings and Smith are quite different cases from one where the taxpayer received no notice, as occurred here. Whatever its faults, an error-filled notice still alerts the taxpayer that the IRS thinks he owes more taxes. No court has held, as the majority does here, that the IRS’s failure to issue any notice is harmless error.
The majority also diverges from Supreme Court precedent by failing to strictly and narrowly construe the tax procedures that parties (including the IRS) must follow. The proper course is to tack as close to the plain language as common sense and precedent permit, not create a new harmless error “exception” to excuse the IRS’s failings. See, e.g., Helvering v. Nw. Steel Rolling Mills, Inc. 311 U.S. 46, 49, 61 S.Ct. 109, 85 L.Ed. 29 (1940); Deputy v. du Pont, 308 U.S. 488, 493, 60 S.Ct. 363, 84 L.Ed. 416 (1940). The provisions at issue are characteristically narrow; they do not create a harmless error exception for when the IRS exceeds its authority and fails to follow the tax code’s deficiency procedures. See I.R.C. §§ 6211-6216. As in Botany Worsted Mills v. United States, 278 U.S. 282, 288-89, 49 S.Ct. 129, 73 L.Ed. 379 (1929), here Congress has prescribed the “exclusive method” by which the IRS can assess and collect deficiencies, I.R.C. §§ 6211-6216. See O’Bryant v. United States, 49 F.3d 340, 346-47 (7th Cir.1995) (“Congress gave the IRS specifically delineated collection authority, and the IRS must act within that authority.”). Part of this prescription is the degree of formality required — the statute requires the IRS to issue a notice of deficiency; it is not enough that the ultimate assessment turns out to be right in the absolute sense.10 Id.; cf. Botany Worsted, 278 U.S. at 289, 49 S.Ct. 129; Raleigh & Gaston R.R. Co. v. Reid, 13 Wall. 269, 270, 20 L.Ed. 570 (1871) (“When a statute limits a thing to be done in a particular mode, it includes the negative of any other mode.”).
The majority is simply wrong to equate a taxpayer’s right to a refund suit or a collection due process hearing (“CDP hearing”) with a taxpayer’s pre-payment, pre-assessment right to challenge the alleged deficiency in Tax Court. Maj. Op. at 1365-66. There are many differences between a CDP hearing and Tax Court, not *1376the least of which is the timing and the existence of an assessment. A taxpayer only has a right to a CDP hearing after taxes have been assessed, the taxpayer has refused to pay, and the IRS has finally sought to collect the unpaid taxes by levying the taxpayer’s property. I.R.C. §§ 6330, 6331(a). By then, the taxpayer’s credit score has been affected and his property subject to liens. I.R.C. § 6322; see United States v. Nat’l Bank of Commerce, 472 U.S. 713, 719-20, 105 S.Ct. 2919, 86 L.Ed.2d 565 (1985); Bronson v. United States, 46 F.3d 1573, 1577 n. 10 (Fed.Cir.1995); In re DeAngelis, 373 F.2d 755, 757 (3d Cir.1967). While it is true that a CDP hearing provides the taxpayer an opportunity to dispute the “underlying tax liability” if he “did not receive any statutory notice of deficiency for such tax liability or otherwise did not have an opportunity to dispute any such tax liability,” the legislative history of CDP hearings suggests Congress wanted to provide “greater due process to taxpayers who are trying to comply with our complex tax laws,” not a substitute for notices of deficiency. 144 Cong. Rec. S4147, 4182 (1998) (emphasis added). Tellingly, even the Tax Court does not consider CDP hearings a substitute for the IRS following the tax code’s deficiency procedures. See Freije v. Comm’r, 125 T.C. 14, 36-37, 2005 WL 1649142 (2005); see also I.R.C. § 6330. This is not surprising, given that CDP hearings are the eleventh-hour option for opposing collection. Similarly, the majority is wrong to equate a taxpayer’s right to file a refund suit with the right to file in Tax Court, before paying anything. Maj. Op. at 1364-65, 1366. As the name suggests, refund suits can only be filed after the taxpayer has paid the amount of taxes the IRS says he owes.
Not surprisingly, the majority’s view also conflicts with the views of other courts. Other courts that have considered the issue do not agree that a CDP hearing or refund suit substitutes for the opportunity to bring a pre-payment suit in Tax Court. See, e.g., Phila. & Reading Corp. v. Beck, 676 F.2d 1159, 1163-64 (7th Cir.1982); Hoyle, 131 T.C. 13, 2008 WL 5156596, at *6 (“[T]his Court has held that ‘petitioner’s opportunity in a section 6330 proceeding [CDP hearing] to dispute the underlying tax liability does not cure an assessment made in derogation of his right under section 6213(a) to a deficiency proceeding.’ ” (emphasis added)).
For these reasons, I cannot agree that the IRS’s failure to issue a notice of deficiency, when such a notice is required, is harmless error.
IV. CONCLUSION
In holding that “computational adjustment” in I.R.C. §§ 6230 and 6231 is limited to situations where there is a change in a partnership item, the majority reaches a conclusion at odds with the statute, our precedent, and common sense. It compounds this error by creating a sweeping harmless error exception, to the detriment of the taxpayer and at odds with every circuit to consider the issue.
For these reasons, I respectfully concur only in the result.
. From the statement of related cases, we already know that these are test cases, and that the majority's decision will immediately affect the approximately thirty tax refund actions brought by partners of related partnerships, which were stayed pending the outcome of this appeal. Of course, these are not the only cases where the question will turn on the definition of "computational adjustment.” Nor is it likely the government will decline to avail itself of the majority’s new harmless error rule, by arguing that other errors the IRS commits are harmless.
. Despite the majority’s protestations to the contrary, under its view of the statute and without the newly minted harmless error rule, Bush and Shelton would be able to raise issues resolved by the settlement agreement in Tax Court. Maj. Op. at 1361-62, n. 9. The majority thus undermines the finality of IRS settlement agreements, the goal of the Tax *1367Equity and Fiscal Responsibility Act of 1982 ("TEFRA”) to streamline partnership proceedings, and this court's decision in Olson v. United States, 172 F.3d 1311 (Fed.Cir.1999), which all bar such repetitive litigation. See I.R.C. §§ 6222, 6224; Olson, 172 F.3d at 1318.
. See, e.g., United States v. Frontone, 383 F.3d 656, 658 (7th Cir.2004) ("The Internal Revenue Service is forbidden, with immaterial exceptions, to assess a deficiency until it has issued the taxpayer a notice of deficiency.”); Clark v. United States, 63 F.3d 83, 84 n. 1 (1st Cir.1995); Phila. & Reading Corp. v. United States, 944 F.2d 1063, 1072 (3d Cir.1991) ("An assessment of taxes that is not preceded by the statutorily required notice of deficiency or a validly executed and accepted waiver of notice of deficiency is illegal, and we so hold.”); see also Ulrich v. Comm’r, 585 F.3d 1235, 1236-37 (9th Cir.2009); Pagonis v. United States, 575 F.3d 809, 813 (8th Cir.2009); Keado v. United States, 853 F.2d 1209, 1211-12 (5th Cir.1988); Hoyle v. Comm’r, 131 T.C. No. 13, 2008 WL 5156596, at *6 (2008) (holding that if an assessment "was not preceded by a notice of deficiency as required by section 6213(a), the assessment is invalid,” meaning no lien could have arisen and the collection was illegal).
. The majority concedes that the government never used the words "harmless error” or argued that the harmless error rule in 28 U.S.C. § 2111 applied. Maj. Op. at 1363 n. 10. Perhaps that is why the majority resorts to extracting a harmless error "essence” from isolated sentences in the government's brief.
. The majority suggests that our disagreement is about the circumstances in which "changes in the tax liability" under I.R.C. § 6230 and § 6231 must occur. Maj. Op. at 1360 ("Judge Prost's opinion suggests that ... the 'change in tax liability' need not be the result of a change ... in the TEFRA proceedings [or settlement].”) This assertion is incorrect. I agree with the majority that such determinations must occur during a proceeding or settlement under TEFRA. See I.R.C. §§ 6224, 6231; see also Olson, 172 F.3d at 1317. The real disagreement is about what the word "change” refers to in the definition of computational adjustment. See I.R.C. § 6231(a)(6).
. There is no question Congress could have used "change” instead of "treatment,” had the intention been to limit computational adjustments to "changes” in partnership items. See Kucana v. Holder, - U.S. -, 130 S.Ct. 827, 837, - L.Ed.2d - (2010); Nken v. Holder, - U.S. -, 129 S.Ct. 1749, 1758, 173 L.Ed.2d 550 (2009); Carcieri v. Salazar, - U.S. -, 129 S.Ct. 1058, 1065, 172 L.Ed.2d 791 (2009); Barnhart v. Sigmon Coal Co., 534 U.S. 438, 452, 122 S.Ct. 941, 151 L.Ed.2d 908 (2002) ("[W]hen Congress includes particular language in one section of a statute but omits it in another section of the same Act, it is generally presumed that Congress acts intentionally and purposely in the disparate inclusion or exclusion.” (quotation marks omitted)).
. See, e.g., Keener, 551 F.3d at 1364; Info. Sys. & Networks Corp. v. United States, 437 F.3d 1173, 1175 (Fed.Cir.2006) (using "treat” to describe whether S corporations are subject to double or single taxation); The Falconwood Corp. v. United States, 422 F.3d 1339, 1341 (Fed.Cir.2005) (using "treat” to describe whether a corporation elects a certain type of taxation); SKF USA Inc. v. United States, 263 F.3d 1369, 1382 (Fed.Cir.2001) (using "treatment” to describe how overpayments are dealt with for tax refund purposes); Stone Container Corp. v. United States, 229 F.3d 1345, 1350 (Fed.Cir.2000) (using "treated” to describe whether a particular tax gives the Court of International Trade jurisdiction); Culley v. United States, 222 F.3d 1331, 1333 (Fed.Cir.2000) (using "treatment” to describe whether a taxpayer could compute his tax *1371liability under I.R.C. § 1341); Holiday Vill. Shopping Ctr. v. United States, 773 F.2d 276, 279 (Fed.Cir.1985) (using “treatment” to describe whether a tax is computed by viewing the business as a partnership or aggregate of individuals).
. The majority compounds the harm of this new rule by subsequently trying to limit it, decreeing that ''[w]hat the taxpayer may not do is forgo his right to resist collection, voluntarily pay the tax, and then secure a refund of tax admittedly owed without showing prejudice,” because there is none. Maj. Op. at 1366. With this sentence, the majority rewrites the due process protections available to the taxpayer. Unless he is prepared to show prejudice, a taxpayer must now fight collection every step of the way, even if he ultimately shows that the IRS was wrong. The major*1374ity cites no authority for its new requirement, which suggests taxpayers may no longer file refund suits unless they previously "resisted] collection.” This rule ignores the century we live in and what the tax code says. We are no longer colonists opposing the Stamp Act or a tax on tea. Congress gave taxpayers die right to do exactly what the majority now forbids: “voluntarily pay the tax,” then sue for a “refund of the tax,” without showing prejudice. I.R.C. § 7422.
. Section 2111 provides that, ”[o]n the hearing of any appeal ... the court shall give judgment after an examination of the record without regard to errors or defects which do not affect the substantial rights of the parties.”
. The majority's analogy to jury trials is thus inapposite; trials are not like tax collections, for which the tax code limits how the IRS may go about its business. See Maj. Op. at 1364-65. No statute similarly prescribes how parties must resolve their disputes.