Summary: The article discusses the retrospective amendments introduced by the Finance Act, 2026, including the insertion of Sections 92CA(3AA) and 147A and amendments to Sections 144C, 153, 153B and 292BA, and examines whether each amendment satisfies the Supreme Court’s settled test for validating legislation. It explains that Parliament may retrospectively remove the basis of a judicial decision but cannot merely override a judgment without curing the defect identified by the Court. The article distinguishes between defects arising from statutory absence and defects arising from non-compliance with an existing legislative mandate, suggesting that the amendments require a provision-by-provision analysis rather than a common approach. It notes that constitutional challenges to these amendments are pending before various High Courts and that the Supreme Court has permitted assessees to amend pending writ petitions to challenge their validity while remanding reassessment matters. The author advises practitioners to preserve jurisdictional grounds in appeals, challenge the retrospective amendments where appropriate, and recognise that the validity of these amendments remains unresolved.
Not All Retrospective Amendments Are Equal
There is a peculiar difficulty that has arrived on the files of most litigators this year, and it is one we have not had to deal with before in quite this form. A ground of appeal that was good in March is, in April, apparently no longer available — not because a court has rejected it, but because Parliament has legislated it out of existence, with retrospective effect, in some cases reaching back nearly two decades.
The grounds in question were not ingenious ones. A transfer pricing order passed beyond the sixty-day period. A reassessment notice issued under Section 148A by the Jurisdictional Assessing Officer when Section 151A and the scheme framed under it had made the process faceless. These were jurisdictional objections, taken carefully and argued at length, and in a number of matters they succeeded before the High Courts on reasoning that did no more than enforce the statute as Parliament had written it.
The Finance Act, 2026 has now inserted Section 92CA(3AA) with effect from 1 June 2007, Section 147A with effect from 1 April 2021, and has amended Sections 144C, 153, 153B and 292BA, each with retrospective operation. Orders that had been held bad now stand validated. The internal analysis in most offices concluded, in substance, that the appeals resting on these grounds are dead.
I would not accept that so quickly, and the purpose of this folio is to set out why. Whether Parliament may validate in this manner is not an open-ended question, and it is not a political one. It has a settled test, laid down by the Supreme Court and restated as recently as 2023. When that test is applied honestly, provision by provision, the amendments do not all come out the same way. That distinction is what a practitioner needs, and it is precisely what a single note on “the retrospective amendments” cannot give.
The question is also live. The Revenue Bar Association’s challenge to these amendments is listed before the Madras High Court on 21 July 2026 and many other High Courts and the High Courts hearing the reassessment matters have been asked to decide by 30 September 2026.
I. The test the Court will apply
The position may be stated in three propositions, each supported by authority.
First, Parliament may legislate retrospectively, and it may do so in a manner that removes the basis of a judgment. This is settled and is not in controversy [Shri Prithvi Cotton Mills Ltd. v. Broach Borough Municipality, (1969) 2 SCC 283 (Constitution Bench)].
Second, what Parliament may not do is declare the judgment to be ineffective while leaving its foundation untouched. The distinction is between removing the substratum of a decision and legislatively overruling it. In Prithvi Cotton Mills the Gujarat legislature supplied a definition of “rate” that the parent statute had never contained — the term was undefined, and the vacuum was filled. In Indian Aluminium Co. v. State of Kerala, (1996) 7 SCC 637, the legislature supplied the statutory authority for a levy that had been imposed by executive action without any such authority.
In each case the defect cured was a defect of absence. In neither did the legislature reverse a mandate it had itself previously and clearly expressed.
Third, and this is the proposition that governs the present amendments, a validating law that does not cure the defect identified by the Court is ultra vires and colourable. The Supreme Court through Justice Nagarathna has put this so eloquently and beautifully, I cannot resit to reproduce both the important paragraphs from the decision of NHPC Ltd – SC – 2023INSC810, dated 06-09-2023:
“13………..The power of judicial review is a part of the basic feature of our Constitution which is premised on the rule of law. Unless a judgment has been set aside by a competent court in an appropriate proceeding, finality and binding nature of a judgment are essential facets of the rule of law informing the power of judicial review. In that context, we observe that while it may be open to the legislature to alter the law retrospectively, so as to remove the basis of a judgment declaring such law to be invalid, it is essential that the alteration is made only so as to bring the law in line with the decision of the Court. The defects in the legislation, as it stood before the Amendment and Validation Act of 1997 was enacted, must be cured by way of the amendments introduced retrospectively. Simply setting at naught a decision of a court without removing the defects pointed out in the said decision, would sound the death knell for the rule of law. The rule of law would cease to have any meaning if the legislature is at liberty to defy a judgment of a court by simply passing a validating legislation, without removing the defects forming the substratum of the judgment by use of a non-obstante clause as a technique to do so.
14. The legislative device of abrogation by enacting retrospective amendments to a legislation, as a means to remove the basis of a judgment and validate the legislation set aside or declared inoperative by a Court, must be employed only with a view to bring the law in line with the judicial pronouncement. Abrogation is not a device to circumvent any and all unfavourable judicial decisions. If enacted solely with the intention to defy judicial pronouncement, such an amendment Act may be declared to be ultra-vires and as a piece of ‘colourable legislation.’ The device of abrogation, by way of introducing retrospective amendments to remove the basis of a judgment, may be employed when a legislature is under the bonafide belief that a defect that crept into the legislation as it initially stood, may be remedied by abrogation. An act of abrogation is permissible only in the interests of justice, effectiveness and good governance, and not to serve the oblique agenda of defying a court’s order, or stripping it of its binding nature.”
The operative test that emerges is therefore not whether Parliament has used the word “clarified”.
It is whether, after the amendment, the Court could still arrive at the same verdict on the same reasoning. If the foundation has genuinely shifted, the validation holds. If the foundation is intact and only the outcome has been declared unacceptable, it does not.
Nor is the legislature’s own description of its amendment conclusive. Where an amendment is substantive in character — creating new obligations, impairing accrued rights, or altering the legal character of past transactions — it does not become retrospective merely because it is labelled clarificatory. The Court will examine the true nature and effect of the provision [CIT v. Vatika Township (P) Ltd., (2015) 1 SCC 1 : (2014) 367 ITR 466 (SC) (Constitution Bench)].
Parliament may remove the ground on which a judgment stands. It may not leave the ground in place and declare that the judgment has fallen.
II. Applying the test, provision by provision
Here is where the general note goes wrong. It treats the Finance Act, 2026 amendments as a single event. They are not. The question to be asked of each is the Prithvi Cotton Mills question: was the defect the Court identified a defect of absence, or a defect of non-compliance?
Section 92CA(3AA) — the transfer pricing limitation. The amendment prescribes the manner in which the sixty-day period for the TPO’s order is to be reckoned against the limitation date, with effect from 1 June 2007. The defect the courts had addressed here was arguably one of absence: the statute prescribed the period but did not spell out with precision how it was to be computed, and the courts supplied a construction. Parliament has now supplied its own. On the doctrine, this is closer to the Prithvi Cotton Mills pattern of filling a computational vacuum than to a defiance of expressed legislative intent.
The transfer pricing limitation ground, whatever its appeal, is the weakest of the three on the doctrine — precisely because there was no earlier and clearly expressed parliamentary mandate that the amendment now contradicts.
Section 147A — the JAO/faceless question. The provision states, notwithstanding any judgment, order or decree of any court and notwithstanding Section 151A or any scheme framed thereunder, that the Assessing Officer for the purposes of Sections 148 and 148A shall mean, and shall always be deemed to have meant, an officer other than the National Faceless Assessment Centre or an assessment unit under Section 144B(3).
Test that against the doctrine. What did the High Courts find? Not that the statute was silent. They found that Parliament, by Section 151A, had expressly mandated a faceless process for the issue of these notices; that the scheme framed under that section said so in terms; and that a notice issued outside the mechanism was consequently issued without jurisdiction. The defect was therefore not one of absence. It was non-compliance with a clear legislative mandate.
And the amendment does not supply anything that was missing. It declares that the mandate always meant its opposite. Its non-obstante clause reaches expressly past Section 151A and past the judgments themselves — which is the very technique the Supreme Court has identified as a colourable device and condemned. On the doctrine, this is the most vulnerable of the amendments, and by some margin.
Sections 144C, 153 and 153B — the DRP and limitation amendments. These lie between the two, and no common answer is available. The question in each case remains the same — did Parliament supply what was missing, or reverse what it had already said? — but the answer turns on which judgment the particular amendment is directed at, and on how clear the language was that the judgment construed. Each must be tested on its own footing. The analysis is provision-by-provision, and a single verdict on “the retrospective amendments” is not an analysis at all.
III. The question that arrives on the file
The doctrinal position matters only because it answers a practical question. You are holding a ground — a JAO ground under Section 148A, or a limitation ground — which the statute appears to have taken away. Do you press it, preserve it, or abandon it?
For the present, preserve it, and the reason is that the Supreme Court has itself indicated that the question is open. In April 2026, disposing of the batch of JAO reassessment appeals, the Supreme Court did not decide the merits. It set aside the judgments under challenge, remanded the matters to the High Courts, and expressly permitted assessees to amend their pending writ petitions to challenge the constitutional validity of the retrospective amendment, including on the ground of impermissible retrospectivity. It stayed the reassessment proceedings meanwhile and requested the High Courts to decide expeditiously, preferably by 30 September 2026. A Court that regarded the amendment as concluding the matter would not have done so.
The practical discipline that follows is straightforward:
Take the ground in the memorandum of appeal. Do not omit it because a validating amendment appears to have closed it. A ground not taken is a ground lost, and the validity of the amendment relied upon is itself sub judice.
Plead it in two stages. First, that the notice or order was without jurisdiction on the law as it stood. Second, in the alternative, that the amendment now relied upon to cure that defect is itself open to challenge as an impermissible legislative overruling on the principles in Prithvi Cotton Mills and Indian Aluminium (supra).
Consider amending a pending writ. Where a writ petition is already before the High Court, the Supreme Court’s April order invites precisely this course.
Be realistic before the Tribunal. The Tribunal cannot strike down a statutory provision. Where the matter is listed there, the ground must be taken and kept alive for the forum competent to decide it.
Advise the client accurately. The position is unresolved; the September horizon is real; and nothing should be conceded in a submission that the High Courts have not yet conceded in a judgment.
- The caveat
None of this is decided, and I would not have it read otherwise.
The Revenue Bar Association’s petition is listed before the Madras High Court on 21 July 2026, the challenge resting on Articles 14, 19, 245 and 265, Entry 82 of List I, the separation of powers and the basic structure. The submission is the one set out above — that Parliament cannot get over a judicial decision by inserting a non-obstante clause without curing the defect the judgment identified.
The Revenue’s answer is not frivolous. It will be said that these amendments are genuinely clarificatory; that the High Courts had themselves divided on the construction of Section 151A; and that a legislature is entitled to state the meaning of its own enactment. Whether Section 147A crosses the line from validation into reconstruction is a question on which senior and careful lawyers will differ, and on which no court has yet ruled.
It is more useful, I think, to write about a question that is open than about one that is closed. The value of the doctrine here is not that it supplies the answer. It is that it tells you which question the Court will ask — and therefore which of your grounds is worth the fight, which needs a different foundation, and which should no longer be described to a client as a certainty.
The next folio reaches you in a fortnight.
From the File
This came up while we were settling a memorandum of appeal. The reassessment notice had been issued by the Jurisdictional Assessing Officer, and we had drafted the ground in February 2026, thinking it was about as safe as a ground gets. Then the Finance Act came in, before the appeal was even filed, and the junior who had drafted it asked whether we should take it out.
We kept it in, but everything behind it had to be rewritten. One jurisdictional objection has become an objection in two parts — that the notice was bad on the law as it stood, and that the provision now relied on to cure it is itself open to challenge. The second part is not about our client’s assessment at all. It is about the limits of the validating power, and it will not be decided by the Tribunal hearing this appeal.
The ground has not become pointless. It has become bigger, and slower. The client is entitled to be told both.
****
Marginalia is written by CA Piyush P. Bafna, Advocate & Chartered Accountant, founder of Piyush Bafna & Associates, Pune. He is the author of AI in Income Tax Litigation (Bharat Law House, 1st Edition, April 2026). Comments, disagreements, and corrections are welcomed at piyush@piyushbafna.com.
Author Bio
