GAAR Part 1: what makes an arrangement impermissible under section 96

In one line. The General Anti-Avoidance Rule is a short chapter with a heavy burden attached, and almost every case decided under it so far has turned not on whether the taxpayer saved tax but on whether the Revenue proved the ingredients Parliament actually enacted.

The Telangana High Court quashed directions of the GAAR Approving Panel in M/s. EKGE Retail LLP v. Assistant Commissioner of Income-Tax, W.P. No. 21210 of 2022, reserved on 6 August 2026 and pronounced on 13 October 2026 by P. Sam Koshy ACJ and Narsing Rao Nandikonda J. A retail business trading under a well known apparel brand was sold as a going concern on 13 June 2017 for Rs 2,10,46,21,110, producing long-term capital gains of Rs 1,75,91,50,508 on a net worth of Rs 33,70,26,715 computed under section 50B. The seller has since continued as a franchisee of the buyer on a commission of about eighteen per cent, so the commercial relationship did not end with the sale. Part of the surplus was invested in listed equities, shares were bought before bonus issues and an equivalent quantity sold afterwards, and the resulting short-term capital loss of Rs 49,32,71,673 was set against the slump sale gains under section 70(2). The Approving Panel, by directions dated 23 February 2022 under section 144BA(6), declared that an impermissible avoidance arrangement. The Court held the Revenue had failed on every ingredient: no arrangement within section 102(1), no main purpose of obtaining a tax benefit within section 96(1), and neither clause (b) nor clause (d) made out.

One structural feature of the facts is worth having in mind throughout, because the order compresses it and a Revenue officer will not. The company that sold the business, New Meena Bazar International Pvt. Ltd., was amalgamated into Madevsag Retail Pvt. Ltd. by an order of the National Company Law Tribunal dated 18 June 2018 with an appointed date of 1 April 2017, so the transfer was deemed to have been effected by the transferee; Madevsag then merged into EKGE Retail by an order dated 24 December 2018 effective from 1 April 2018. The entity that sold the business, the entity that bought the shares and the entity that claimed the set-off are therefore not the same on the face of the record, and only a retrospective appointed date makes them one.

That is the third GAAR decision of the same High Court on capital loss harvesting in twenty-eight months, and the second of them on bonus stripping in listed shares; whether the middle one was a bonus case at all could not be established, and that uncertainty, which is about what can be read rather than about the law, is dealt with in Part 3. The three do not sit comfortably together. This is not a note on one of them. It sets out Chapter X-A as a whole, and alongside each provision its counterpart in the Income-tax Act, 2025, which has been in force since 1 April 2026 and governs tax year 2026-27 onwards; the 2025 Act number is given in brackets at the first substantive mention of each provision, and the full concordance is in Part 3. It takes the chapter as a whole because the pattern running through all three is that the cases are won and lost on the structure of section 96(1) rather than on the merits of the planning, and because the chapter is short enough that a practitioner can hold the whole of it in mind.

This is Part 1 of three. This Part is the substantive test: what has to be shown before an arrangement is impermissible at all, which is where the Revenue has lost every case it has lost. Part 2 takes up section 99, which belongs to the test rather than to its consequences, and then what follows a finding: the section 98 consequences, the section 100 question, the section 144BA machinery from the reference to the Approving Panel’s directions, what remedy survives, and what the three Telangana High Court decisions actually decide. Part 3 is the judicial doctrine alongside the statute, Circular No. 7 of 2017, the Form 3CD clause 30C question a chartered accountant has usually answered already, how a reference is defended, how a transaction is built so that the question does not arise, and the full concordance with the Income-tax Act, 2025.

Key points, across all three Parts. The Part that works each one out is named at the end of the point.

• Rule 10U keeps Chapter X-A out altogether where the aggregate tax benefit to all the parties in the relevant assessment year is three crore rupees or less, and where the income arises from the transfer of investments made before 1 April 2017. Both are computed on definitions elsewhere in the chapter, and both should be worked before anything else on the file. (Part 1)
• Section 96(1) has two requirements and they are cumulative: the main purpose must be to obtain a tax benefit, and the arrangement must additionally answer one of the four descriptions in clauses (a) to (d). The four descriptions are alternatives to each other, not a further list of cumulative conditions. Getting that structure wrong is the commonest error on both sides. (Part 1)
• Before section 96 is reached there must be an “arrangement” within section 102(1). That is a threshold, not a formality, and in two of the three Telangana cases it is where the Revenue lost. (Part 1)
• Section 99 lets the Revenue collapse connected persons, disregard an accommodating party and look through a corporate structure for the purpose of deciding whether a tax benefit exists at all. That is how a benefit is found in a taxpayer who has none, and how the three crore threshold is reached. (Part 2)
• Section 100 says the Chapter applies “in addition to, or in lieu of, any other basis for determination of tax liability”. Whether that lets the Revenue reach conduct Parliament chose to address by a specific rule is the central unresolved question in this field. It was not argued in EKGE. (Part 2)
• The section 144BA machinery has three stages and two of them are the Commissioner’s. A taxpayer who does not answer the section 144BA(2) notice loses the Approving Panel stage altogether, because the reference to the Panel arises only where the assessee has objected. (Part 2)
• No appeal lies against the Panel’s directions, by section 144BA(14). The assessment order made in consequence is appealable, but it bypasses the Commissioner (Appeals) and goes straight to the Tribunal under section 253(1)(e). The directions themselves can only be attacked under Article 226. (Part 2)
• A Chapter X-A adjustment carries penalty exposure, and a tax treaty is not an answer to it: section 90(2A) disapplies the beneficial-provision rule so far as the Chapter is concerned. (Part 2)
• Bonus stripping in shares was outside section 94(8) until the Finance Act 2022 extended it to securities with effect from assessment year 2023-24. Every GAAR case on the technique so far concerns a year before that. (Part 1 and Part 2)
• The same presiding judge upheld GAAR on related-party facts in June 2024 and has since quashed it twice on exchange-traded facts. The 2024 decision is before the Supreme Court, leave has been granted, and it has not been disposed of. The Supreme Court decided its own GAAR matter on 15 January 2026. Neither is mentioned in the October 2026 order. (Part 2)

What GAAR is, and when it switched on

Chapter X-A is eight sections long, and the Income-tax Act, 2025 re-enacts it as Chapter XI, sections 178 to 184, in substantially the same words. Section 95 applies it (2025 Act, section 178), section 96 defines the target (section 179), section 97 deems certain arrangements to lack commercial substance (section 180), section 98 sets out the consequences (section 181), section 99 allows connected persons and accommodating parties to be collapsed (section 182), section 100 states how the Chapter sits alongside the rest of the Act and section 101 is the rule-making hook (the two merged into section 183), and section 102 defines the terms (section 184). Everything else, the Approving Panel and the forms and the time limits, is in section 144BA (2025 Act, section 274) and in the rules made under section 101 (Income-tax Rules, 1962, rules 10U to 10UC; Income-tax Rules, 2026, rules 127 to 130).

Section 95(1) provides that “Notwithstanding anything contained in the Act, an arrangement entered into by an assessee may be declared to be an impermissible avoidance arrangement and the consequence in relation to tax arising therefrom may be determined subject to the provisions of this Chapter.” Sub-section (2) provides that “This Chapter shall apply in respect of any assessment year beginning on or after the 1st day of April, 2018.” An Explanation adds that the Chapter may be applied to any step in, or a part of, an arrangement as it applies to the arrangement.

The non obstante clause is the whole of the Revenue’s answer on the SAAR question. It is the textual basis on which the Telangana High Court held in 2024 that GAAR can reach conduct which a specific provision did not then cover. It is dealt with at length in Part 2.

The Explanation makes a step a target in its own right. Read with the presumption in section 96(2), it means the Revenue does not have to characterise the whole of a commercial transaction as avoidance; it can isolate one leg. That is how a genuine business sale and a genuine investment programme can still generate a GAAR reference on one trade inside them, and it is the reason a planner has to give each step its own reason rather than relying on the commerciality of the whole.

And the commencement history is worth getting right, because it is frequently misstated. Chapter X-A was first inserted by section 41 of the Finance Act 2012 with effect from 1 April 2014. It never came into operation: section 25 of the Finance Act 2013 omitted that Chapter with effect from the very date it would have taken effect, and section 26 of the same Act inserted the present Chapter with effect from 1 April 2016. Section 26 of the Finance Act 2015 then renumbered section 95 as section 95(1) and added sub-section (2), pushing application out to assessment year 2018-19. So the Chapter has been on the statute book since 1 April 2016 but bites only from assessment year 2018-19, and the 2012 version is of historical interest only. The 2015 amendment is often dated to 1 April 2016. The better view is 1 April 2015: clause 26 of the Finance Bill 2015 renumbers section 95 and inserts sub-section (2) without a separate commencement date, and section 1(2) of the Finance Act 2015 brings the relevant sections into force on 1 April 2015 save as otherwise provided. On that reading the amendment to section 95 took effect a year before the Chapter it amends, which is odd but is what the Act says.

Section 101, which the 2025 Act folds into section 183(b), deserves a line of its own because it is usually passed over. It provides that the Chapter “shall be applied in accordance with such guidelines and subject to such conditions, as may be prescribed”, and it is the authority for Rules 10U onwards. The consequence matters: those rules are not administrative guidance of the kind that can be attacked as inconsistent with the Act, they are the conditions subject to which the Chapter itself operates. That is why Rule 10U(1)(a) is an exclusion from the Chapter rather than a concession the Revenue can decline to apply.

One consequence of the commencement date deserves a warning rather than a footnote. Because GAAR applies only from assessment year 2018-19, a good deal of the reasoning that looks like GAAR reasoning in earlier years is not. In ACIT v. Vallabh Roopchand Bhanshali, ITA No. 889/Mum/2024, decided on 6 August 2024 for assessment year 2017-18, an off-market sale of listed scrips at a loss to a limited liability partnership in which the assessee held sixty per cent, the Tribunal at Mumbai expressly noted that GAAR applies from assessment year 2018-19 onwards and was therefore inapplicable, said that an inference could nonetheless be drawn from section 95, and decided the matter on McDowell and Vodafone. The defence of that course is that the Tribunal was using section 95 as an indication of Parliament’s view of devices of that kind rather than as a source of power, which is a conventional if weak use of later legislation. Even so, importing the statutory standard into a year Parliament excluded from it is not a step that should pass without argument.

Rule 10U: the three crore threshold and the 1 April 2017 grandfathering

Before any of the doctrine matters, Chapter X-A has to apply at all, and Rule 10U of the Income-tax Rules, 1962 keeps it out of four situations. Its counterpart under the Income-tax Act, 2025 is rule 128 of the Income-tax Rules, 2026, which carries all four forward in the same terms, with “previous year” becoming “tax year” and the definitions pointing to the new Act. Two of them are structural and two are specific to foreign portfolio investment.

The rule as it governed every year so far decided, that is before 31 March 2026, provided:

“(1) The provisions of Chapter X-A shall not apply to—

(a) an arrangement where the tax benefit in the relevant assessment year arising, in aggregate, to all the parties to the arrangement does not exceed a sum of rupees three crore;

(b) a Foreign Institutional Investor,—(i) who is an assessee under the Act; (ii) who has not taken benefit of an agreement referred to in section 90 or section 90A as the case may be; and (iii) who has invested in listed securities, or unlisted securities, with the prior permission of the competent authority, in accordance with the Securities and Exchange Board of India (Foreign Institutional Investor) Regulations, 1995 and such other regulations as may be applicable, in relation to such investments;

(c) a person, being a non-resident, in relation to investment made by him by way of offshore derivative instruments or otherwise, directly or indirectly, in a Foreign Institutional Investor;

(d) any income accruing or arising to, or deemed to accrue or arise to, or received or deemed to be received by, any person from transfer of investments made before the 1st day of April, 2017 by such person.

(2) Without prejudice to the provisions of clause (d) of sub-rule (1), the provisions of Chapter X-A shall apply to any arrangement, irrespective of the date on which it has been entered into, in respect of the tax benefit obtained from the arrangement on or after the 1st day of April, 2017.”

The three crore gate. Read the words rather than the summary. It is the tax benefit “in the relevant assessment year”, it is computed “in aggregate, to all the parties to the arrangement”, and the test is whether it “does not exceed” three crore rupees. So it is tested year by year, it is aggregated across parties rather than applied per taxpayer, and it is an exclusion the Revenue must clear rather than a threshold the taxpayer must prove. How the benefit is quantified is in Rule 10U(3)(iv) (2026 Rules, rule 128(3)), which defines “tax benefit” for this purpose by reference to section 102(10) and provides that for sub-clauses (a) to (e) of that definition it is the amount of tax, and for sub-clause (f), an increase in loss, it is “the tax that would have been chargeable had the increase in loss referred to therein been the total income”. That last limb is the operative one in a loss-harvesting case: the benefit is measured by the tax on the loss, not by the loss.

Two features of it are easy to get wrong in both directions. Because the test aggregates across the parties, a taxpayer whose own benefit is modest can still be inside the Chapter; and because section 99 permits connected persons to be collapsed for the purpose of deciding whether a tax benefit exists, the aggregation can reach further than the parties the taxpayer would have named. Because the test is annual, an arrangement spread across years is tested year by year and a single year below the line does not protect the arrangement.

The grandfathering gate. Clause (d) excludes income from the transfer of investments made before 1 April 2017. The exclusion is keyed to when the investment was made, not to when the arrangement was entered into or when the income arose. Sub-rule (2) then applies the Chapter to an arrangement whenever entered into, in respect of a tax benefit obtained on or after 1 April 2017, and the opening words “without prejudice to the provisions of clause (d) of sub-rule (1)” are the pivot on which the whole grandfathering dispute turns. The Delhi High Court read them as confining sub-rule (2) so that it could not defeat the clause (d) grandfathering; the Supreme Court set that judgment aside on 15 January 2026, and that litigation is dealt with in Part 2.

On the EKGE facts neither gate was available, and the order does not mention Rule 10U at all. The short-term capital loss of Rs 49,32,71,673 sheltered long-term gains charged at twenty per cent, so the benefit measured on the Rule 10U(3)(iv) basis was of the order of ten crore rupees, several times the line. And the pre-bonus shares were bought during the financial year 2017-18, after 1 April 2017, out of the proceeds of a business sold in June 2017, so clause (d) had nothing to attach to. The year was assessment year 2018-19, the very first in which the Chapter applied. The silence of the order is a reminder that these gates are not applied by the officer of his own motion: they are raised by the taxpayer, in the objections, with the computation annexed.

Both provisions have since been rewritten. Clause (d) and sub-rule (2) were substituted by the Income-tax (Tenth Amendment) Rules 2026 with effect from 31 March 2026, the substitution dropping the “without prejudice” opening and writing the grandfathering exception into the operative words of sub-rule (2) instead; a companion amendment made the corresponding change to the rule under the Income-tax Act, 2025 with effect from 1 April 2026. The amendment is recorded in the footnotes to the Department’s own published text of the rules; the notifications are reported as No. 54/2026 for rule 10U and No. 55/2026 for rule 128, both of 31 March 2026, which could not be read in an official copy for this article. Anyone working a live file must therefore check which text governs the year in question: the version set out above governs every year any decided case concerns, and the substituted version governs from 31 March 2026 onwards. One further textual point for anyone quoting from the Department’s web page rather than the Gazette: the page prints “accrues of arises” in sub-rule (2) where the notification reads “accrues or arises”.

These gates are not theoretical, though the one decided case on them is more qualified than it is usually described. In Reverse Age Health Services Pte Ltd v. DCIT, ITA No. 1867/Del/2022, decided by the Tribunal at Delhi on 17 February 2023 for assessment year 2018-19, the Revenue had taxed short-term capital gains of Rs 1,92,63,473 on a conduit finding against a Singapore company. The Tribunal held both exclusions made out, the benefit being below three crore rupees and the shares acquired before 1 April 2017, but framed that reasoning on the assumption that GAAR applied and decided the appeal on other grounds. It is a useful illustration of how the exclusions are applied and it is not a case decided on Rule 10U alone, so it should not be cited as one.

A third gate sits outside Rule 10U and is worth a line. Chapter X-A applies only from assessment year 2018-19. For an earlier year the Revenue is thrown back on the judicial doctrine, which is a materially different and generally weaker position.

Who the parties are for the purpose of that aggregate is not answered in Rule 10U. Section 99 permits connected persons to be treated as one person and an accommodating party to be disregarded for the purpose of deciding whether a tax benefit exists at all, and it is therefore capable of carrying a case over the three crore line that would not cross it on the assessee’s own figures. It is taken up in Part 2.

How bonus stripping produces a loss: nil cost, first-in-first-out and section 70(2)

Before the law, the mechanics, because a practitioner has to be able to show on paper where the loss comes from. The Revenue’s instinct is to describe it as artificial; the answer is that it is the arithmetic the Act itself prescribes.

The three provisions that produce the loss are given below in their 1961 numbering only; their counterparts in the Income-tax Act, 2025 were not traced for this article and are not guessed at, for the reason given in the concordance in Part 3.

The commercial idea is simple and the EKGE order sets it out in the taxpayer’s own illustration. If a share trades at Rs 1,000 before a bonus is declared and the bonus is allotted one for one, the price ought in theory to halve to Rs 500 once the enlarged capital is in issue. In practice it was believed it would not fall quite that far, trading at Rs 501 or above, and that the difference was an arbitrage available with little additional risk. On those figures an ex-bonus price of Rs 501 yields about two per cent within a fortnight, which annualises to about fifty per cent. That is the commercial rationale the taxpayer advanced, and whether it is made out on the evidence is a different question from whether it is a rational thing for an investor to attempt.

The gain being sheltered is computed separately, and the mechanics of a slump sale under section 50B are dealt with elsewhere. The tax consequence of the share trades comes from three provisions acting together.

The cost of a bonus share is nil. Section 55(2)(aa)(iiia) provides that the cost of acquisition of a financial asset “allotted to the assessee without any payment and on the basis of holding of any other financial asset, shall be taken to be nil in the case of such assessee”. It has applied since 1 April 1996, the clause having been inserted by the Finance Act 1995; the Department’s current page for the section carries no footnote recording that, so the attribution rests on the commencement date rather than on the Department’s own note. Clause (aa) is expressed to operate “subject to the provisions of sub-clauses (i) and (ii) of clause (b)”, so a bonus share that became the property of the assessee before 1 April 2001 is not forced to nil and the fair market value option survives. Citing section 55(2)(b) for the nil cost rule, as is sometimes done and as the Revenue’s submission recorded in the EKGE order appears to do, is wrong: clause (b) is the residuary provision and is relevant here only as the carve-out.

The shares sold are identified by first-in-first-out. For shares held in dematerialised form this is statutory, not administrative: section 45(2A) provides that for the purposes of section 48 and the proviso to section 2(42A), “the cost of acquisition and the period of holding of any securities shall be determined on the basis of the first-in-first-out method”. The circular that explains it is Circular No. 768 dated 24 June 1998, which also supplies the working rules that matter in practice, that first-in-first-out applies only to the dematerialised holding, that it is applied account-wise where an investor has more than one account, and that where physical shares are dematerialised later the basis is the date of entry into the account. Circular No. 704 dated 28 April 1995 is a different circular: it deals with the date of transfer by reference to the broker’s note, and its paragraph 3 applies first-in-first-out only “to reckon the period of the holding of the security, in cases where the dates of purchase and sale could not be correlated through specific numbers of the scrips”, which presupposes physical certificates. The Revenue’s submission in EKGE attributed the first-in-first-out requirement to Circular No. 704. That attribution has the wrong source. It appears in the Court’s summary of counsel’s submission rather than in the directions themselves, so nothing can safely be said from it about how the directions were written; but it is worth checking the Panel’s own order on any live file for the same error, because the two circulars do different work and the wrong one does not support the proposition.

And the set-off is unrestricted in the direction that matters. Section 70(2) entitles an assessee to set a short-term capital loss “against the income, if any, as arrived at under a similar computation made for the assessment year in respect of any other capital asset”, with no qualification. Section 70(3) confines a long-term loss to income “in respect of any other capital asset not being a short-term capital asset”. Section 74(1)(a) and (b) repeat the asymmetry for carried-forward losses. So a short-term capital loss can shelter a long-term capital gain and the converse is not true. Both sections were recast by the Finance Act 2002 with effect from 1 April 2003; between assessment years 1988-89 and 2002-03 the Finance Act 1987 had omitted section 70(2) altogether and both directions were open.

Put the three together and the loss is not an artefact of planning but of the statute. The original shares carry their full cost; the bonus shares carry nil; first-in-first-out treats the original shares as the ones sold; the market price after a bonus reflects the enlarged capital. The seller therefore realises a loss on the original shares and holds bonus shares at nil cost whose eventual sale will produce a gain of the whole consideration. The loss is real in the year and the gain is deferred to the year of the later sale.

That last sentence is the most useful thing a defender can say to a Panel, because it reframes the whole dispute. What the taxpayer obtains from bonus stripping is, in substance, a timing benefit. Section 94(8), where it applies, recognises exactly this: it does not condemn the transaction, it ignores the loss and adds it to the cost of the bonus units, so that the deferral is removed and nothing else is. Whether a timing benefit is the kind of mischief that justifies the machinery of Chapter X-A, when Parliament has addressed the identical mischief elsewhere by a rule that merely reverses the timing, is a question the Revenue should have to answer.

The two conditions in section 96(1), and how they are misread

Section 96(1), re-enacted without change of substance as section 179(1) of the Income-tax Act, 2025, is the heart of the chapter and it is short enough to set out in full:

“An impermissible avoidance arrangement means an arrangement, the main purpose of which is to obtain a tax benefit, and it—

(a) creates rights, or obligations, which are not ordinarily created between persons dealing at arm’s length;

(b) results, directly or indirectly, in the misuse, or abuse, of the provisions of this Act;

(c) lacks commercial substance or is deemed to lack commercial substance under section 97, in whole or in part; or

(d) is entered into, or carried out, by means, or in a manner, which are not ordinarily employed for bona fide purposes.”

Read the structure before the content. There is a stem, “the main purpose of which is to obtain a tax benefit”, and there is a tail, “and it”, followed by four descriptions. The word “or” appears once, at the end of clause (c), which is the ordinary drafting convention for a list of alternatives. So the provision requires the main purpose and any one of the four. It does not require all four, and it is not satisfied by any one of the four alone.

The EKGE court began from a canon of construction that a defender should quote before anything else:

“The scheme of Chapter X-A of the Act introduced w.e.f. 01.04.2018 is one of the considerable rigour and for that very reason its conditions must be construed strictly.”

It then stated the structure, unusually cleanly:

“Section 96(1) of the Act declares an arrangement to be an impermissible avoidance arrangement only where two conditions are cumulatively satisfied: firstly, that the main purpose of the arrangement is to obtain a tax benefit; and secondly, that the arrangement additionally answers any one of the four descriptions in clauses (a) to (d). The threshold requirement, preceding both limbs, is that there must exist an ‘arrangement’ as defined in Section 102(1).”

And on burden:

“The burden of establishing each of these ingredients rests upon the Revenue, for Chapter X-A of the Act, visits the assessee with consequences of the gravest kind, including disregard or re-characterisation of transactions and denial of treaty benefits, none of which can follow upon a mere suspicion or upon the Revenue’s disapproval of the commercial course an assessee has chosen to adopt.”

The strict construction holding is the answer to the Revenue’s best textual argument, which is that the definitions in Chapter X-A are drawn as widely as language permits and should be given their width. The Revenue’s reply, which should be anticipated, is that a rule aimed at avoidance which is construed strictly against the Revenue is a rule that defeats its own purpose, and that no decision of the Supreme Court yet applies a strict construction canon to this Chapter. At present the proposition rests on this judgment.

Both errors available on section 96(1) are live in the decided cases, and they run in opposite directions.

The Revenue’s error is to treat the four limbs as a description of the result rather than as conditions. If a tax benefit resulted, the reasoning runs, the arrangement must have been abusive and the manner must not have been bona fide. That is the reasoning the EKGE court identified and rejected, and it is taken up under clauses (b) and (d) below.

The assessee’s error, or at least the assessee’s good fortune, is the opposite. The Telangana High Court is reported, in Smt. Anvida Bandi v. DCIT, to have treated the four limbs as cumulative, effectively reading an implicit “and” into a provision that uses “or”. If that is what the Court held, it is a generous reading and it should not be built on: it would require the Revenue to prove non-arm’s-length terms, misuse, absence of commercial substance and an unusual manner in every case, which would make clauses (a) to (d) a single compound test rather than four alternatives, and would make the “or” at the end of clause (c) do no work. There is a defence of the judgment available, and because nobody has read it in primary text the defence may well be the right reading: the Court may have been recording that the Revenue had pleaded all four limbs and failed on each, rather than laying down a cumulative test. Until the certified copy is available, treat EKGE as the authority on the structure of section 96(1) and Anvida Bandi as authority on the facts.

A word on Anvida Bandi generally, because it is cited a great deal and will be cited to you. The judgment is reported at [2025] 177 taxmann.com 726 (Telangana), W.P. No. 3201 of 2023, and is dated 22 August 2025 before P. Sam Koshy and Suddala Chalapathi Rao JJ, for assessment year 2020-21. Its primary text is not publicly available: although the free judgment databases carry Telangana decisions up to the present, this one is absent from them, and the High Court’s own judgment search is behind a security check. Everything located for this article is law-firm paraphrase, and no such note reproduces a sentence of the judgment. The writ number and the amounts are corroborated across several of those notes; the date of 22 August 2025 and the composition of the bench are given by the Taxmann report and could not be confirmed elsewhere. The particulars and holdings described in this article are taken from those reproductions rather than from a primary text, and nothing from it should be put inside quotation marks in a submission without a certified copy.

The threshold question: is there an arrangement at all?

Section 102(1), which becomes section 184(2) of the Income-tax Act, 2025, defines “arrangement” to mean “any step in, or a part or whole of, any transaction, operation, scheme, agreement or understanding, whether enforceable or not, and includes the alienation of any property in such transaction, operation, scheme, agreement or understanding.”

On its face that is about as wide as a definition can be made. It covers a step, it covers a part, it covers an understanding that is not enforceable, and it expressly brings in the alienation of property. The Approving Panel in EKGE held that buying before a bonus and selling after it was an “operation” and a “scheme” within those words, which is not an obviously wrong reading of the text.

The Court disagreed, and the reasoning repays close reading:

“The words ‘transaction’, ‘operation’, ‘scheme’, ‘agreement or understanding’ employed in Section 102(1) of the Act must be read in their setting, and each of those expressions read in the company of the others, connotes a degree of concert, design or arrangement between identifiable parties, or at the very least a pre-ordained set of steps whose outcome is within the control of the person said to have entered into it.”

Applied to the facts:

“Whereas here, the quantum of gain or loss is not and cannot be pre-ordained, being wholly dependent upon the movement of market prices over which the petitioner exercised no control whatsoever, it is difficult to characterise the petitioner’s trade as an ‘operation’ or ‘scheme’ within the meaning of the definition.”

And the observation that gives it force:

“That the petitioner’s expectation of a favourable price movement post-bonus happened to be borne out in some instances and evidently not in others is a consequence of market behaviour and not of any design of the petitioner’s making.”

This is a noscitur a sociis argument: the wide words take colour from their neighbours. It is a respectable argument and it produces a sensible result. It is also doing a great deal of work, and three difficulties should be acknowledged by anyone who intends to rely on it.

The statute does not say “two or more persons”. The reported holding in Anvida Bandi that an arrangement must be between two or more persons has no textual anchor in section 102(1), which speaks of a step in a transaction and of an understanding that need not be enforceable. A unilateral series of steps is comfortably within the ordinary meaning of “operation”. The counter is that “agreement or understanding” plainly imports bilaterality and that “scheme” and “operation” should be read with them, which is the EKGE reasoning; but the Revenue has the easier textual case here and should be expected to make it.

That matters on the EKGE facts in a way the order does not develop. The transferor was amalgamated into the transferee by an order of the National Company Law Tribunal with an appointed date retrospective to a point before the business sale, so the identity of the person who made the gain, the person who bought the shares and the person who claimed the set-off is supplied by the scheme rather than by the trades. A Revenue officer with a freer hand would say that the scheme is itself a step in an arrangement within section 102(1) read with the definition of “step” in section 102(9). The taxpayer’s answers are that the appointed date was fixed for company-law reasons unconnected with the shares, that the Board has said GAAR will not apply where a court has explicitly and adequately considered the tax implications in sanctioning an arrangement, and that the amalgamation preceded both the share purchases and any tax benefit. None of that was argued, and it is the line of attack this order leaves open for the next case.

The definition expressly includes a single step, and “step” is itself defined. If an arrangement can be “any step in” a transaction, it is hard to see why one leg of an investment programme cannot be an arrangement merely because the counterparty to that leg is anonymous. The answer, if there is one, is that the step must still be a step in something that is itself an arrangement.

And the Court’s reasoning turns on the outcome rather than on the steps, which is the Revenue’s best point. Section 102(1) requires a transaction, operation, scheme, agreement or understanding, or a step in one. It nowhere requires that the outcome be determinate. The steps here were entirely pre-ordained: buy before the record date, take the allotment, sell an equivalent quantity afterwards. What was not pre-ordained was how much the trade would make or lose. On the Revenue’s case the Court has read an outcome requirement into a definition that contains only a conduct requirement, and the consequence of that reading, if it is right, is that no unilateral multi-step restructuring carried out by a single taxpayer at market prices can ever be an arrangement, which would take a large part of Chapter X-A out of use. That is the strongest objection to this part of the judgment and it will be made on appeal.

What the anonymity of the counterparty really establishes is that the price was not set by concert, which goes to clause (a) and clause (d). It does not by itself show there was no design on the part of the taxpayer.

None of that means the Court was wrong. It means the holding is a construction of wide words against the Revenue in a penal context, which is a legitimate approach but is not the only available one, and it has not yet been tested on appeal.

The definitions that decide the case: section 102

Section 102, which becomes section 184 of the Income-tax Act, 2025 with every clause number shifted by one from “arrangement” onwards, does more work than its position at the end of the Chapter suggests. Four of its eleven definitions decide who is inside an arrangement, and therefore whose tax benefit is aggregated for the three crore gate and in whose hands a benefit can be found at all. An adviser who has not read them cannot do the Rule 10U arithmetic the rest of this article recommends.

“Party”, in clause (6) (2025 Act, section 184(7)), “includes a person or a permanent establishment which participates or takes part in an arrangement”. Note that a permanent establishment is a party in its own right, and that participation is the test rather than contracting. The three crore aggregation under Rule 10U(1)(a) runs across “all the parties to the arrangement”, so this definition fixes the denominator of the calculation.

“Connected person”, in clause (4) (2025 Act, section 184(5)), is very wide and is the provision that does the most damage in a family or group file. It means any person connected directly or indirectly to another person and includes: a relative of an individual; a director of a company or a relative of such director; a partner or member of a firm, association of persons or body of individuals, or a relative of such partner or member; a member of a Hindu undivided family or a relative of such member; any individual having a substantial interest in the business of the person, or a relative of such individual; a company, firm, association, body or family having a substantial interest in the business of the person, together with its directors, partners or members and their relatives; the converse case where the person’s own director, partner or member has a substantial interest in the other; and any other person carrying on a business in which the person, or its director, partner or member, or a relative, has a substantial interest. “Relative” is separately defined by section 102(7), which takes the meaning in the Explanation to section 56(2)(vi); and two further definitions complete the set, “asset” in clause (2), which includes property or a right of any kind, and “benefit” in clause (3), which includes a payment of any kind whether tangible or intangible and is the root of “tax benefit”.

“Substantial interest”, in clause (8) (2025 Act, section 184(9)), is the twenty per cent test that governs most of those limbs: a person has a substantial interest in a business if, at any time during the financial year, he is the beneficial owner of equity shares carrying twenty per cent or more of the voting power where the business is carried on by a company, or is beneficially entitled to twenty per cent or more of the profits in any other case.

“Step”, in clause (9) (2025 Act, section 184(10)), “includes a measure or an action, particularly one of a series taken in order to deal with or achieve a particular thing or object in the arrangement”. The Explanation to section 95 and the presumption in section 96(2) both key off that word, which is how a single leg of a larger transaction becomes the unit of analysis.

Two further definitions matter in specific contexts. “Tax benefit”, in clause (10) (2025 Act, section 184(11)), includes a reduction, avoidance or deferral of tax or other amount payable under the Act, an increase in a refund, the same two things arising as a result of a tax treaty, a reduction in total income, and an increase in loss, in each case “in the relevant previous year or any other previous year”; those closing words govern the whole list, which is what allows a benefit obtained in a later year to be attributed to an arrangement entered into earlier. And “fund”, in clause (5) (2025 Act, section 184(6)), includes cash, cash equivalents and any right or obligation to receive or pay cash, which is the definition that gives round trip financing under section 97(2) its reach.

The practical instruction is short. Before accepting that the aggregate tax benefit is below three crore rupees, work out who the parties are under clause (6) and who is connected under clause (4) read with clause (8), because the Revenue will.

Main purpose, and the error of reasoning backwards

Suppose the threshold is crossed. The Revenue must then show that the main purpose of the arrangement was to obtain a tax benefit. The statute says “main purpose”, not “a purpose”, and not “one of the main purposes”, which is the formulation used in the principal purpose test in the multilateral instrument and in several foreign general anti-avoidance rules. India chose the narrower test in 2013 and did not change it when the chapter was re-enacted in the Income-tax Act, 2025. The point is stronger than it looks: the Chapter as first enacted in 2012 read “the main purpose or one of the main purposes”, and the words “or one of the main purposes” were dropped when the present Chapter was substituted in 2013. The EKGE court put the point in a sentence worth carrying into a hearing:

“The statute uses the words ‘main purpose’, not ‘a purpose’ or ‘an incidental consequence’. That an assessee, having incurred a loss, thereafter claims the set-off which the statute itself expressly permits, cannot by itself convert the earning of that loss into an exercise whose main purpose was the obtaining of a tax benefit.”

“Tax benefit” is widely defined and expressly includes a reduction in total income and an increase in loss, so a loss claim is squarely a tax benefit. That is not the difficulty. The difficulty is purpose, and the Court identified the characteristic failure in the Revenue’s reasoning and named it:

“Instead of asking whether the material established that the dominant purpose of the petitioner’s investment decisions was the obtaining of a tax benefit, the impugned directions proceed from the premise that because a tax benefit resulted, the purpose must have been to obtain it. That is to reason backwards from effect to intention and it is not the enquiry that Section 96(1) mandates.”

The same paragraph then states, inside Chapter X-A and in the Court’s own voice, the proposition that is usually traced through Azadi Bachao and the English cases:

“It is well settled that an assessee is entitled to so arrange its affairs as to attract the least incidence of tax, provided it does so within the four corners of the law, and that the mere circumstance that a lawful course adopted by an assessee also happens to be tax-efficient does not render that course impermissible.”

That is the single most useful sentence in the order for a defender, because it is a statement of the principle made by a court applying the very chapter said to displace it.

That is a real error and not merely a rhetorical one. Section 96(1) is not satisfied by an arrangement that produces a tax benefit; if it were, the stem would be surplus and every loss claim would be within the chapter. The stem exists to separate arrangements undertaken for a commercial reason that happen to be tax-efficient from arrangements undertaken because they are tax-efficient. That separation can only be made on evidence about why the thing was done, and the Revenue has to gather it.

In EKGE there was evidence, and the Court set it out at length: a surplus of Rs 151.06 crore after discharging Rs 59.40 crore of liabilities out of the consideration of Rs 210.46 crore, of which about Rs 20 crore arrived only a year after the transfer, leaving about Rs 130 crore then available for investment; engagement with the representatives of most of the leading financial institutions in the country; an internal consultant retained to advise on deployment, the taxpayer having no prior experience of dealing with investment managers; the surplus parked meanwhile in liquid mutual funds, which earned Rs 2,09,27,881 in the year and was returned as short-term capital gains; and a considered allocation of about one-third to listed equities and two-thirds to bonds and debt funds. The Court described that as “a commercial narrative that respondent No.3 has left substantially unaddressed”, which is the criticism that matters: not that the Panel reached the wrong conclusion on the evidence, but that it did not engage with it.

Two cautions about using that narrative as a template, because a practitioner who copies it without noticing them will be caught.

The allocation and the execution did not match. The stated plan was about Rs 50 crore into equities. The aggregate actually invested in pre-bonus shares was Rs 106.31 crore, more than double that figure and over eighty per cent of the corpus. The order records both numbers without reconciling them. That is the most obvious line of attack on the commercial narrative and it should be expected in the next case. A written asset allocation is worth having; a written asset allocation that the execution then departs from by a factor of two is worth rather less, and the file should either follow the plan or record why it changed.

And the retained position proves less than it looks. The Court said of the Rs 49.07 crore retained at cost:

“That retention is itself telling; an assessee whose sole object was the manufacture of a capital loss would have had no reason to remain invested to that extent in the underlying scrip.”

The Revenue’s answer, which the judgment does not address, is that retaining stock is a definitional feature of the technique rather than evidence against it. Bonus stripping works by selling the original holding and keeping the nil-cost bonus shares, and section 94(8)(c), where it applies, requires as a condition of the specific rule that the person sells the original units “while continuing to hold all or any of the additional units”. Parliament legislated on the assumption that a bonus stripper retains stock. So retention as such is neutral. What is not neutral, and what a taxpayer should be able to show, is retention of the original scrip beyond what the technique requires, at a quantum that can be explained on commercial grounds. A taxpayer who exits entirely is in much the weaker position; a taxpayer who stays invested has a real answer, provided he can say why he held what he held.

The group comparison, and what the Revenue’s own assessments show

There is a third ground in the judgment which is neither purpose nor clause, and it is the one most transferable to a search practice. The Court said:

“We are also unable to overlook two features of the record that the impugned Directions leave unexplained. The search and seizure operation conducted on 27.11.2018 and 28.11.2018 yielded no incriminating material whatsoever against the petitioner, and although the statement of its Director came to be recorded, nothing that could be described as incriminating emerged. Further, of the approximately 105 assessments of the petitioner’s group reopened in consequence of that search, 104 came to be completed with nil or minimal adjustment to the returned income.”

And drew the conclusion:

“Against this backdrop, the singling out of the present assessment for the invocation of the extraordinary machinery of Chapter X-A required, at the very least, a careful and reasoned demonstration of each ingredient of Section 96(1) and such a demonstration is conspicuously absent from the impugned Directions.”

This is a comparative argument: the Revenue’s own treatment of the group is evidence against its characterisation of this one assessment. In any file arising out of a search the group’s assessment history is known to the taxpayer and should be tabulated and put on the record before the Commissioner, with the dates and the adjustments. A Chapter X-A reference that is the sole adverse outcome of a group-wide exercise invites exactly the question the Court asked.

The Revenue’s answer is available and should be anticipated: each assessment stands on its own facts, and a hundred clean assessments say nothing about the one transaction that nobody else in the group entered into. That answer is a good one where the transaction really is unique to the assessee. It is a much weaker one where the Chapter X-A reference is the residue of an exercise that found nothing.

The presumption in section 96(2), and the order of the enquiry

Section 96(2), re-enacted as section 179(2) of the Income-tax Act, 2025 with “notwithstanding” becoming “irrespective of”, provides:

“An arrangement shall be presumed, unless it is proved to the contrary by the assessee, to have been entered into, or carried out, for the main purpose of obtaining a tax benefit, if the main purpose of a step in, or a part of, the arrangement is to obtain a tax benefit, notwithstanding the fact that the main purpose of the whole arrangement is not to obtain a tax benefit.”

It is narrower than it is usually described. It is not a general presumption that an arrangement is an impermissible avoidance arrangement. It is a presumption about the relationship between a step and the whole: once the main purpose of a step is shown to be a tax benefit, the whole arrangement is presumed to have that main purpose, and it is for the assessee to prove the contrary. Its practical effect is to stop a taxpayer from sheltering a tainted step inside a large commercial transaction by pointing to the commerciality of the transaction as a whole.

The presumption has a precondition and the Revenue has to satisfy it. Something must first be shown about the main purpose of a step. The presumption cannot be used to establish that very thing, and it cannot be used to establish the separate ingredients in clauses (a) to (d) at all. That sequential reading is reported to be the one the Telangana High Court adopted in Anvida Bandi, holding that the presumption arises only once a section 96(1) condition is met and that the Revenue must establish the conditions with concrete evidence rather than assertions of intent.

But the opposite reading is on the books and is from the same Court. In Ayodhya Rami Reddy Alla v. PCIT (Central), W.P. Nos. 46510 and 46467 of 2022, decided on 7 June 2024 by P. Sam Koshy and Laxmi Narayana Alishetty JJ, the Court contrasted the position under GAAR with Vodafone and said that section 96(2) places the responsibility on the taxpayer. That is a reading on which the presumption does rather more work, and the Revenue will cite it. It is also the decision now before the Supreme Court.

For a practitioner the safe course is to argue the sequence, not the burden in the abstract. The question to put to the Panel is: which step have you identified, and what is the material from which you say its main purpose was a tax benefit? If the answer is that the loss was claimed, that is reasoning backwards again. If there is a real answer, the presumption engages and the file turns on the commercial narrative.

Clause (a): rights and obligations not ordinarily created at arm’s length

Clause (a) catches an arrangement that “creates rights, or obligations, which are not ordinarily created between persons dealing at arm’s length”. It was not invoked in EKGE and is not discussed in any of the three Telangana decisions, which is unfortunate, because in practice it is the limb with the widest reach. Most real GAAR exposure is in related-party restructuring, intra-group funding, family settlements, share swap ratios and buy-backs, and clause (a) is the limb that bites there.

It looks at the content of the rights and obligations, not at the price. A transaction can be at an arm’s length price, certified as such under Chapter X, and still create rights no independent party would accept: a tenure no lender would grant, an option exercisable only on the occurrence of a tax event, a put that leaves the holder without economic risk, a waiver or subordination that an unconnected creditor would never give, a ratio in a scheme that no outside shareholder would have voted for. The question is not what the parties paid but whether they would have agreed to this at all.

It is the only limb with an express comparator, and the comparator has to be identified. “Not ordinarily created between persons dealing at arm’s length” requires the Revenue to say what persons dealing at arm’s length would ordinarily have done. An order that asserts the conclusion without identifying the comparator transaction has not applied the clause, and that is a point to take in the objections rather than at the hearing.

And it overlaps with the transfer pricing code and with section 40A(2), which cuts both ways. The taxpayer’s argument is that where the terms have been tested and documented under section 92 and found to be at arm’s length, that is strong evidence under clause (a) too. The Revenue’s answer is that section 92 prices a transaction that the parties have chosen to enter into, whereas clause (a) asks whether parties at arm’s length would have entered into it in that form at all, so an arm’s length price is not an answer to a non-arm’s length right. Both propositions are respectable and neither has been decided.

On the facts of a case like EKGE, clause (a) is very difficult to run, and saying why is cleaner than relying on the general proposition that exchange execution is ordinary, which belongs to clause (d). Where a taxpayer buys and sells listed shares on an exchange at market prices, there is no identifiable person with whom non-standard rights could have been created, so the comparator has little to bite on.

The Revenue’s answer has not been run in any of the decided cases and should be anticipated. An exchange trade does create rights and obligations: the contract note creates them, and the trade is novated to the clearing corporation, which is an identifiable counterparty in law even though the opposite client is not. On that footing the question clause (a) asks is whether those rights are ones ordinarily created between persons dealing at arm’s length, to which the answer on screen-based dealing is plainly yes. That is a less elegant answer than saying the clause has nothing to bite on, and it is the more durable one, because it does not depend on a proposition about anonymity that the statute nowhere requires.

Clause (b): misuse or abuse

Clause (b) catches an arrangement that “results, directly or indirectly, in the misuse, or abuse, of the provisions of this Act”. The EKGE court dealt with it in a short passage and the reasoning is the most doctrinally important part of the judgment:

“There has been no misuse and no abuse. The petitioner invoked Section 70(2) according to its terms, computed the cost of bonus shares as the statute commanded under Section 55, applied the First-In-First-Out method as the Board’s Circular required, and computed its slump sale gains under Section 50B. Abuse connotes the employment of a provision for a purpose alien to that for which it was enacted. Set-off under Section 70(2) exists precisely so that losses may be set against gains in arriving at total income, and the petitioner used it for no other purpose.”

That gives clause (b) a workable content: abuse means using a provision for a purpose alien to the purpose for which it was enacted. It is the orthodox formulation of abuse in anti-avoidance law and it imports a purposive enquiry into the provision relied on, not into the taxpayer’s motive. The distinction matters, because it means the Revenue cannot establish clause (b) merely by establishing the stem. If the stem were enough, clause (b) would add nothing.

The Court reinforced the point with a structural argument about the nature of the charge:

“Income tax is a single, unified and annual levy. It is not an aggregate of discrete taxes upon discrete receipts. The total income of an assessee for a previous year can be arrived at only after the losses sustained in some activities are set off against the gains derived from others; that is not a concession granted to the assessee but an intrinsic feature of the manner in which taxable income is computed under the Act.”

And drew the conclusion:

“To describe the working out of a computation provision according to its own terms as the obtaining of a ‘tax benefit’ is to misapprehend the nature of the levy. A ‘tax benefit’ within the contemplation of Chapter X-A must be something that the assessee secures by circumventing or subverting the scheme of the Act and not something that flows from the direct and unavoidable operation of a computation provision upon a loss that was genuinely and commercially sustained.”

This passage deserves both approval and a caveat, and a practitioner should carry both.

The approval is that it is right about set-off. Sections 70 and 74 are not reliefs. Section 70(2) entitles an assessee to set a short-term capital loss against the income computed in respect of any other capital asset, and the words carry no restriction. Treating the exercise of that entitlement as itself the abuse would be to treat the computation of total income as presumptively suspect.

The caveat is that the sentence about a tax benefit proves too much if read widely. Section 102(10) defines tax benefit to include “an increase in loss” and “a reduction in total income”, so a loss set-off is a tax benefit as the statute uses that term; what the Court must mean is that the existence of a tax benefit is not itself the misuse required by clause (b), which is a point about clause (b) and not about the definition. Read as a gloss on the definition it would put most of Chapter X-A out of action, because almost every tax benefit arises from the operation of some provision according to its terms. Read as a point about abuse it is sound. Quote it with that distinction made, or it will be turned around.

Clause (c): commercial substance under section 97, the limb nobody argued

Clause (c) of section 96(1) catches an arrangement that “lacks commercial substance or is deemed to lack commercial substance under section 97, in whole or in part”. It is the most heavily drafted of the four limbs and, on these facts, the most dangerous, which makes it striking that the EKGE order does not mention section 97 at all. The Approving Panel had invoked clauses (b) and (d), and the Court decided the case on those. Anyone reading the judgment as authority that bonus stripping has commercial substance, or that clause (a) could not be made out on such facts, is reading into it something it does not contain.

Section 97(1), which becomes section 180(1) of the Income-tax Act, 2025, deems an arrangement to lack commercial substance if:

“(a) the substance or effect of the arrangement as a whole, is inconsistent with, or differs significantly from, the form of its individual steps or a part; or

(b) it involves or includes—(i) round trip financing; (ii) an accommodating party; (iii) elements that have effect of offsetting or cancelling each other; or (iv) a transaction which is conducted through one or more persons and disguises the value, location, source, ownership or control of funds which is the subject matter of such transaction; or

(c) it involves the location of an asset or of a transaction or of the place of residence of any party which is without any substantial commercial purpose other than obtaining a tax benefit (but for the provisions of this Chapter) for a party; or

(d) it does not have a significant effect upon the business risks or net cash flows of any party to the arrangement apart from any effect attributable to the tax benefit that would be obtained (but for the provisions of this Chapter).”

Sub-section (2) provides that round trip financing “includes” any arrangement in which, through a series of transactions, funds are transferred among the parties and those transactions have no substantial commercial purpose other than the tax benefit, and it directs that this be determined without regard to whether the funds can be traced, to the sequence in which they moved, or to the means by which they moved. The word is “includes”, not “means”, so the category is open, which is the Revenue’s point and is worth noticing before the provision is dismissed as narrow. Sub-section (3) defines an accommodating party as one whose main purpose in participating, in whole or in part, is to obtain a tax benefit for the assessee, whether or not it is a connected person; the 2025 Act moves that definition out of the section entirely, to section 184(1).

Sub-section (4), which becomes section 180(3) of the 2025 Act, is the one that wins cases for taxpayers and is the least quoted:

“For the removal of doubts, it is hereby clarified that the following may be relevant but shall not be sufficient for determining whether an arrangement lacks commercial substance or not, namely:—(i) the period or time for which the arrangement (including operations therein) exists; (ii) the fact of payment of taxes, directly or indirectly, under the arrangement; (iii) the fact that an exit route (including transfer of any activity or business or operations) is provided by the arrangement.”

Notice the force of “shall not be sufficient”. Parliament has legislated that short duration cannot by itself establish absence of commercial substance. In a bonus stripping case the Revenue’s instinct is to point at the holding period: bought days before the record date, sold days after. Clause (i) of section 97(4) is the direct answer to that instinct. The Telangana High Court is reported to have held in Anvida Bandi that timing alone cannot bring a transaction within GAAR, though the accounts of that decision attribute the holding to the Expert Committee’s report rather than to section 97(4); the statutory provision is the better foundation and is available whether or not the Court used it.

Two limbs would nonetheless repay a careful Revenue’s attention on a bonus stripping file.

Clause (b)(iii), offsetting or cancelling elements. Buying a block of shares and selling an equivalent quantity of the same scrip within weeks looks, on one view, like two elements that cancel each other. The answer is that they do not cancel: the taxpayer is left holding the bonus shares and remains exposed to the scrip. The qualification made above applies here too, that the bonus parcel is retained as a feature of the technique, so the answer is stronger where the taxpayer also keeps original scrip he was not required to keep.

Clause (d), no significant effect on business risks or net cash flows. This is the serious one. A bonus stripping trade does move cash and does carry price risk between purchase and sale, which is real market risk and is the risk the EKGE court treated as decisive at the threshold. But the Revenue’s point would be that the risk is incidental and that the net effect on the taxpayer’s position, apart from the tax benefit, is slight against a loss sheltering a gain many times its size. That is an argument about degree, it has never been tested on exchange-traded facts, and it is the argument the Revenue should have been making.

In Ayodhya Rami Reddy Alla the Court did decide the case on commercial substance, holding the arrangement “devoid of commercial substance as per Section 97”. The facts there were different in kind, and the difference is set out in Part 2.

Clause (d): the means and the manner

Clause (d) catches an arrangement “entered into, or carried out, by means, or in a manner, which are not ordinarily employed for bona fide purposes”. The EKGE court’s treatment of it is the easiest passage in the judgment to apply:

“The clause is directed at the means and the manner, not at the outcome.”

Then the application:

“Purchases and sales of listed securities on a recognised stock exchange regulated by the Securities and Exchange Board of India, with securities transaction tax and service tax duly discharged, represent the most ordinary, conventional and universally employed means of transacting in securities in this country. Transactions of enormous aggregate volume are executed through this very channel every trading day. There is no means more ordinary and no manner more bona-fide available to an investor wishing to deal in equities. To hold that recourse to a regulated stock exchange is a means not ordinarily employed for bona-fide purposes would be to stand the clause upon its head.”

The sentence “directed at the means and the manner, not at the outcome” is the sentence to take into a hearing. Clause (d) is about how the thing was done. It is not a second route to the same conclusion as the stem, and an order that establishes clause (d) by pointing to the tax result has not applied it at all.

The limit of the point should also be stated plainly, because an adviser who overstates it will mislead a client. Clause (d) is not satisfied merely because the venue was ordinary, and it is not answered by showing that the venue was ordinary either. An arrangement can be carried out through entirely conventional channels and still be caught by clause (b) or clause (c). Where a taxpayer has dealt on exchange, clause (d) will usually be the Revenue’s weakest limb and clause (c) its strongest.

Where this goes next

That is the test, and on the decided cases it is where the Revenue has lost. Part 2 takes up section 99 and the three crore aggregate, what may be done once the test is satisfied, the machinery through which it has to be done, the remedy against it, and what the three Telangana High Court decisions actually decide. Part 3 takes up the doctrine that runs beside the statute, the Board’s circular, the Form 3CD clause 30C exposure, the practice of answering a reference and structuring around one, and the concordance with the Income-tax Act, 2025.

This note is general commentary on the law as at 18 October 2026 and is not advice on any matter. The position in a particular case depends on its own facts.