In February 2007, two companies incorporated outside India agreed the sale of a single share in a third company incorporated in the Cayman Islands, and on the face of the documents no asset located in India changed hands at all. Whether the Income-tax Act, 1961 could nevertheless reach the gain on that offshore exchange is the interpretive question from which the Indian law of retrospective taxation, in the form that made it internationally notorious, descends. The Supreme Court answered it one way in January 2012. Parliament answered it the other way within months, writing its answer back to the first day of April 1962, the day the Act itself had commenced half a century earlier. Two international tribunals then held that the second answer breached treaties India had signed, and in 2021 Parliament withdrew the demand it had so carefully constructed, on conditions, without conceding that the construction had been unlawful.

The episode is usually told as a morality play, and the moral is usually drawn too large. Retrospective taxation was not abolished in India in 2021, because nothing in the statute book ever forbade it and nothing in the withdrawing statute forbids it either. What happened is narrower and more instructive. A court read a charging provision and found that it did not extend to a class of transaction. A legislature, constitutionally entitled to legislate backwards in fiscal matters, extended it backwards over a decided case. The extension collided not with the Constitution but with investment treaties, and the eventual retreat was a settlement priced in refunds, forgone interest and abandoned claims rather than a confession of legal error. Tracing that sequence accurately, with the demand stated in its components rather than in the headline figures that newspapers preferred, is the work of this article.

A preliminary discipline governs everything that follows. The figures in this dispute were reported differently at different stages because the demand itself changed composition as interest accrued and penalty was threatened, and any account that offers a single total without its components is misleading. The original quantified demand of October 2010 was Rs 11,218 crore, of which about Rs 7,900 crore was tax and the remainder interest. A penalty of up to one hundred per cent of the tax was separately threatened in 2011, which would have carried the exposure beyond Rs 20,000 crore, and by the time an arbitral tribunal came to record the claim, reported aggregates ranged from about Rs 22,100 crore to about Rs 27,900 crore depending on how interest and penalty were counted. Each figure in this article is therefore tied to the stage at which it was the operative one.

Three ways the episode is conventionally misread

Because the dispute was reported for a decade in the compressed vocabulary of business journalism, three misreadings have hardened into the standard summary, and each one distorts a different part of the record. The first is that the Supreme Court held retrospective taxation to be unconstitutional, or held that India could not tax offshore transfers of Indian businesses. It held neither. It construed a single clause of the Act, found the charge absent from it, and expressly left Parliament free to enact the charge by legislation; the constitutional validity of doing so retrospectively was a question the Court was never asked to decide in the case and never decided. The second misreading is that the Taxation Laws (Amendment) Act, 2021 repealed retrospective taxation, or repealed Explanation 5. It did neither. It disapplied the 2012 amendment for indirect transfers undertaken before 28 May 2012 and nullified the resulting demands on conditions, leaving the charge intact for every later transaction and leaving Parliament’s retrospective power exactly where the Constitution had placed it. The third misreading is that the arbitral tribunals held India’s tax law to be unlawful. They did not examine the amendment’s validity under Indian law at all. They held that the manner of its application to two specific investors breached a treaty standard of fair and equitable treatment, a conclusion about India’s conduct under international law that coexisted, without logical difficulty, with the amendment’s complete validity under domestic law. Keeping the three corrections in view is not pedantry. Each misreading flatters a different constituency, the judiciary, the legislature or the taxpayer respectively, and each conceals the actual distribution of responsibility for the outcome, which the remainder of this article sets out.

Retrospective taxation and the words of section 9(1)(i)

The charging provision at the centre of the dispute is the first limb of section 9(1) of the Income-tax Act, the provision that deems income to accrue or arise in India in defined situations even where the transaction that produces it happens elsewhere. The complete architecture of the Act, its charging sections, its computation heads and its assessment machinery, is set out in the complete guide to the Income Tax Act 1961, and only the single clause that matters here needs close reading. Section 9(1)(i) deems income to accrue or arise in India where it arises, directly or indirectly, through or from any business connection in India, or through or from any property in India, or through or from any asset or source of income in India, or through the transfer of a capital asset situate in India. The territorial hook is the last of these. If the capital asset transferred is situate in India, the gain is Indian income in the hands of a non-resident, and the buyer who pays for the asset without deducting tax at source can be pursued for the tax.

The provision had a history that both sides mined. The source rule in section 9 descends from the corresponding provisions of the Income-tax Act, 1922, and its organising idea was territorial: income is deemed to arise in India where its source, its payer connection or its asset is located in India. The neighbouring limb of the same clause, directed at income arising through a “business connection” in India, had been construed by the Supreme Court in Commissioner of Income-tax v R. D. Aggarwal in 1965 as requiring a real and intimate relation between the trading activities abroad and those in India, a relationship of continuity rather than a casual contact. A later generation of authority, of which Ishikawajima-Harima Heavy Industries v Director of Income-tax in 2007 is the leading instance, insisted that a deeming provision of this kind be applied with territorial discipline: the fiction extends Indian taxing reach only to the extent the provision specifies, and income with no territorial nexus to India is not drawn in by implication. That jurisprudence formed the interpretive weather of 2007. It told advisers that section 9 was a provision with edges, that its fictions were to be read strictly, and that the situs of the asset transferred was the hinge of the capital gains limb. The revenue’s contrary reading in Vodafone was therefore not an eccentric position; it was a request to move the edge, made first to the courts and then, successfully, to Parliament.

Two features of the text framed everything that followed. The first is that the clause fastens on the capital asset transferred, and asks where that asset is situated. The share that changed hands in 2007 was a share in a Cayman Islands company, and a share is ordinarily situated where the company is incorporated and its register kept, which was not India. The second feature is that the clause contains no language of indirection. It does not say that income arising through the transfer of an asset that derives its value from another asset in India is deemed to arise in India. It does not mention shares in foreign companies, holding structures, or underlying value at all. The revenue’s case therefore had to be built out of the general words, principally the word “through”, and out of an argument about substance: that what was really transferred, beneath the Cayman share, was a controlling interest in an Indian telecommunications business, and that the court should read the provision as reaching the economic reality.

Which words of section 9(1)(i) did the dispute turn on?

The dispute turned on the word “through” and on the phrase “capital asset situate in India”. The revenue argued that “through” was wide enough to carry the charge across an offshore holding company to the Indian assets beneath it. The taxpayer answered that the asset actually transferred was a foreign share, situated abroad, and that a provision written for direct transfers could not be stretched by one preposition into a general tax on indirect transfers.

The textual point mattered because Parliament had elsewhere shown that it knew how to write an indirect transfer charge when it wanted one. The Direct Taxes Code Bill, the long-gestating replacement for the 1961 Act which was under preparation in the same period, contained an express provision taxing the transfer of shares in a foreign company whose value derived substantially from Indian assets. The existence of that draft language cut both ways politically, but legally it supported the taxpayer: if the existing Act had already contained the charge, the draft code would not have needed to create it. The Supreme Court would later make exactly this use of the comparison. The interpretive question was thus a genuine one about the reach of enacted words, not a contrived ambiguity, and it is why the case is properly classified as a judicial interpretation dispute rather than as a contest about whether offshore tax planning is respectable. The Court was asked what the provision meant. Its answer, and Parliament’s counter-answer, are the twin subjects of the sections that follow.

There is one further textual feature worth recording, because it explains the shape of the 2012 amendment. Section 9(1)(i) is a deeming provision. It does not tax all income with an Indian flavour; it enumerates the connections that Parliament treats as sufficient to deem income Indian. A deeming provision of that kind is ordinarily read strictly, because each limb is a deliberate extension of territorial reach beyond the ordinary source rules, and a court asked to extend a fiction beyond the cases the legislature has specified is being asked to legislate. That was the gravitational pull on the Supreme Court’s reasoning. The counter-pull, which the revenue pressed throughout, is that deeming provisions in tax statutes are remedial and should be read to capture the income Parliament plainly wanted to capture. The tension between those two pulls, neither of them disreputable, is the doctrinal core of the case.

The transaction and the revenue’s claim

The transaction was agreed on 11 February 2007 and completed in May 2007. Hutchison Telecommunications International Limited, the Hong Kong based arm of the Hutchison Whampoa group, sold the entire share capital of CGP Investments (Holdings) Limited, a company incorporated in the Cayman Islands, to Vodafone International Holdings BV, a Netherlands company in the Vodafone group. The price was about 11.1 billion US dollars, a figure reported at the time with slight variations between about 11.08 and 11.2 billion dollars depending on the treatment of assumed loans and closing adjustments. CGP was not an operating business. It was the apex of a holding structure through which the Hutchison group held its Indian telecommunications investment: beneath it sat a chain of intermediate companies which together held about 52 per cent of Hutchison Essar Limited, the Indian operating company, with options and allied arrangements over roughly a further 15 per cent, giving the structure an aggregate interest of about 67 per cent in the Indian business. Hutchison Essar, later renamed Vodafone Essar, ran one of India’s largest mobile networks.

What Vodafone bought, in legal form, was one foreign company from another foreign seller, for a price paid outside India, under documents executed outside India. What it acquired, in economic substance, was control of an Indian business with Indian licences, Indian spectrum, Indian subscribers and Indian employees. The whole dispute is contained in the distance between those two sentences. The structure itself was not invented for the sale. Hutchison had held its Indian investment through offshore holding companies for years, for the ordinary reasons that multinational groups hold emerging market investments through intermediate jurisdictions: regulatory approvals, financing, joint venture arrangements with Indian partners, and the ability to sell the whole investment at the holding level. That history mattered later, because the Supreme Court’s analysis of whether the structure was a device turned substantially on whether it had a commercial life independent of the exit.

The revenue’s claim was not, in form, an assessment on the seller. Hutchison’s gain, if taxable at all, was the seller’s income, and the seller was a non-resident outside the practical reach of an Indian assessment. The claim was instead built on the withholding architecture of the Act. Section 195 obliges any person responsible for paying a non-resident a sum chargeable under the Act to deduct tax at source. Section 201 treats a payer who fails to deduct as an assessee in default, liable for the tax itself together with interest. In September 2007 the department served a show cause notice on Vodafone contending that the price it had paid Hutchison was chargeable in India, that Vodafone had been obliged to withhold tax from it, and that Vodafone was in default for failing to do so. The sum then in contemplation was about 2.1 billion US dollars. The theory made the buyer the collection point for the seller’s alleged liability, and it had the practical virtue, from the department’s perspective, that Vodafone, unlike Hutchison, had a vast and continuing Indian business against which a demand could be enforced.

Why did the revenue proceed against the buyer and not the seller?

The revenue proceeded against Vodafone as buyer because the Act’s withholding machinery made the buyer the collectible party. Section 195 required tax to be deducted from any payment to a non-resident that was chargeable in India, and section 201 made a payer who failed to deduct an assessee in default for the tax, with interest. The seller had exited and was offshore; the buyer had acquired an Indian business and remained fully present within the jurisdiction.

This procedural choice had a doctrinal consequence that shaped the litigation. Because the proceedings were framed as a default in withholding, the threshold question was not what tax was in the end payable after a full assessment of the seller, but whether the sum paid was chargeable under the Act at all, since a payment that is not chargeable attracts no withholding obligation. The entire merits of the indirect transfer question were therefore litigated inside a jurisdictional inquiry: did section 9(1)(i), as it stood, deem the gain on a foreign share sale to arise in India? If it did not, the notice collapsed at the root. The department’s contrary position, that the withholding inquiry was preliminary and should not decide the substantive charge, was pressed at every stage and rejected at every stage, and the case accordingly became the vehicle for a definitive ruling on the substantive provision. It is a pattern familiar in Indian tax litigation, where the largest questions of chargeability are frequently settled in proceedings that began as recovery machinery directed at the party within reach.

The revenue’s affirmative case, as distinct from its procedural route, was an argument about what had really been sold. The department contended that the CGP share was a cipher: that the commercial reality of the February 2007 agreement was the transfer of Hutchison’s Indian telecommunications business, together with the bundle of rights by which that business was controlled, including interests held through subsidiaries, call options over further stakes held by Indian partners, and rights of management and control over Hutchison Essar. On that view, the consideration of about 11.1 billion dollars was paid overwhelmingly for assets and rights situated in India, the gain was in substance a gain on Indian assets, and section 9(1)(i) applied to it because the transfer operated, in the language of the clause, “through” the offshore instrument to the Indian business beneath. The argument had genuine forensic strength at the level of economics, which is why it persuaded the Bombay High Court in substance, and why the debate it generated was never a debate between honesty and evasion. Its weakness was legal rather than factual: it required the court to identify the capital asset transferred not as the asset the parties had in law conveyed, but as the aggregate of everything the conveyance economically delivered, and to find in the word “through” a warrant for that substitution. Everything in the decade that followed is contained in that single contested move. The Supreme Court refused it as a matter of construction; Parliament enacted it as a matter of legislation; the tribunals condemned its retrospective application as a matter of treaty law; and the 2021 Act withdrew it as a matter of settlement. The argument itself, that the value was Indian and should bear Indian tax, was never refuted, and in its prospective form it is the law.

The road through the courts

Vodafone challenged the show cause notice in the Bombay High Court, arguing that the department had no jurisdiction even to inquire, because the transaction lay wholly outside the Act. In December 2008 the High Court dismissed the writ petition and allowed the department to proceed, holding at that threshold stage that the notice was not barred and that the department was entitled to examine the transaction. Vodafone carried the matter to the Supreme Court, which in January 2009 declined to shut the inquiry down but protected the taxpayer’s position: the department could decide the jurisdictional question, and if it decided against Vodafone, Vodafone could challenge that decision in the High Court. The department duly decided. By an order in May 2010 it held that it had jurisdiction, reasoning that the transaction, viewed in substance, effected a transfer of capital assets situated in India and of controlling rights over the Indian company.

Vodafone returned to the Bombay High Court, and on 8 September 2010 the High Court upheld the department in substance. The Court held that the transaction could not be characterised as a simple sale of one foreign share; that the bundle of rights transferred included, in substance, interests in the Indian company; and that the department had jurisdiction to tax the gain to the extent attributable to Indian assets, leaving quantification and apportionment for the assessment stage. The judgment was careful and it was not a blank cheque for the revenue: it accepted that part of the consideration might reflect value outside India, and it framed the department’s power as extending to the Indian element. But on the question that mattered, whether the Act reached the transaction at all, the revenue had won.

The quantified demand followed in October 2010: Rs 11,218 crore, comprising about Rs 7,900 crore of tax and the balance of interest. Vodafone appealed to the Supreme Court, which in November 2010 required it, as the price of the appeal being heard, to deposit Rs 2,500 crore and to furnish a bank guarantee of Rs 8,500 crore. In March 2011 the department issued a notice proposing a penalty of up to one hundred per cent of the tax for the failure to deduct, a step that would have more than doubled the exposure; the Supreme Court stayed the penalty proceedings until the appeal was decided. The appeal was then heard at length, and judgment was delivered on 20 January 2012. The revenue also filed a review petition after the judgment, which the Court dismissed on 20 March 2012, days after the Budget had already proposed the legislative reversal described below. The sequence is worth fixing precisely, because later commentary frequently misorders it: the government did not legislate after exhausting its remedies and then lose again; it legislated while the judgment was still warm, and the dismissal of the review and the introduction of the Finance Bill fell within the same week.

The Vodafone judgment

The judgment of the Supreme Court in Vodafone International Holdings BV v Union of India, delivered on 20 January 2012 and reported at (2012) 6 SCC 613, was the product of a three judge bench of Chief Justice S. H. Kapadia, Justice K. S. Radhakrishnan and Justice Swatanter Kumar. The Court set aside the Bombay High Court’s judgment of 8 September 2010 and held for the taxpayer on every element of the claim. Its holdings can be stated with precision, and it is a disservice to the judgment, committed by supporters and critics alike, to state them any more broadly.

First, the transaction was characterised as what its documents said it was: a transfer of the share capital of CGP, a Cayman Islands company, by a non-resident seller to a non-resident buyer, outside India. The subject matter of the transfer was the CGP share, and a share in a company is situated where the company is incorporated. It followed that no capital asset situate in India had been transferred, and the deeming limb of section 9(1)(i) directed at such transfers was not engaged. Secondly, the Court held that section 9(1)(i) is not a look through provision. The revenue had invited the Court to disregard the holding structure and treat the sale as a direct transfer of the underlying Indian shares and rights. The Court declined, holding that the statutory language did not authorise that exercise and that the department’s approach would require the Court to rewrite the charging provision. Thirdly, the Court affirmed the discipline that has governed this area since the Azadi Bachao Andolan decision of 2003, departing from the broader anti-avoidance rhetoric of the McDowell decision of 1985: a taxpayer is entitled to arrange its affairs within the law, a holding structure with commercial substance is not a colourable device merely because it is tax efficient, and the substance over form doctrine is reserved for arrangements shown to be shams or lacking in commercial reality. On the facts, the Hutchison structure was a longstanding investment platform, not a contrivance assembled for the exit, and the transaction was characterised by the Court as a bona fide structured foreign direct investment falling outside India’s territorial tax jurisdiction.

Fourthly, and with consequences beyond the case, the Court reasoned from the structure of fiscal legislation itself. The judgment emphasised certainty and predictability as constituents of the rule of law in taxation, particularly for foreign investment, and it pointed to the draft Direct Taxes Code, in which Parliament’s own draftsmen had proposed an express charge on indirect transfers, as evidence that such a charge did not form part of the existing Act. A court, in other words, would not supply by interpretation a provision that the legislature had recognised it needed to enact. Justice Radhakrishnan’s concurring judgment ranged more widely over the use of offshore financial centres and holding jurisdictions, treating them as a normal feature of cross-border investment rather than a badge of avoidance, and insisted that the inquiry into a structure is an inquiry into its genuineness and commercial purpose, not into the motives of investors who choose recognised routes.

The reasoning by which the Court reached those holdings rewards attention, because it is the reasoning, rather than the result alone, that fixed the terms on which Parliament’s response would be judged. The majority’s method was to begin with the situs of the asset. A share is situated where the company is incorporated and its register kept, and the asset transferred was the share capital of a Cayman Islands company, carrying with it the indirect benefit of the companies beneath it but legally distinct from each of them. From that starting point the question became whether section 9(1)(i) authorised the Court to disregard the identity of the asset transferred and treat the transaction as if the Indian assets at the bottom of the chain had themselves been sold. The Court held that it did not, and that a provision deeming income to arise in India on the transfer of an asset situate in India could not be converted, by the presence of the word “through”, into a charge on the transfer of an asset situated abroad. The deeming fiction had to be confined to its stated purpose, a discipline the Court drew from the long line of authority on statutory fictions, under which a fiction is carried only as far as the legislature’s purpose requires and no further.

Running through the judgment is a distinction between looking at a transaction and looking through it, which the Court was careful to preserve. A court may, and where avoidance is alleged must, look at a transaction as a whole, examining its documents, its commercial purpose and the conduct of the parties, in order to understand what was actually done. That is the enduring and legitimate part of the substance doctrine. What a court may not do, absent statutory warrant, is look through the legal forms the parties have genuinely adopted and recharacterise the transaction as a different one that Parliament has not taxed. The revenue’s case required the second operation: the CGP share, genuinely held and genuinely sold, had to be treated as if it were the Hutchison Essar shareholding itself. The Court’s refusal was therefore not a refusal to examine substance. It was a refusal to let examination substitute for enactment, and it explains why the judgment, although delivered in a case about an offshore holding company, is cited far beyond tax law as an authority on the limits of purposive construction in fiscal statutes.

The treatment of the avoidance authorities completes the reasoning. The Court accepted the revenue’s invitation to consider whether the structure was a device, and answered it on the facts: the Hutchison group had held its Indian investment through the CGP chain for years before any sale was contemplated, Vodafone was an operating telecommunications group acquiring a business it then ran, and the exit was of a kind the structure had always contemplated. On those findings the arrangement stood on the legitimate side of the line that McDowell had drawn and that Azadi Bachao Andolan had reaffirmed in the treaty context: tax planning within the framework of the law is not impermissible merely because it saves tax, and the colourable device doctrine is directed at arrangements that are artificial in fact, not at arrangements that are merely tax efficient. The judgment’s restraint on this point is what gives the subsequent history its edge. Parliament was not reversing a court that had closed its eyes to avoidance. It was reversing a court that had looked at the structure, found it genuine, applied the enacted words, and held for the taxpayer on both limbs.

The limits of the decision deserve equal care. The Court did not hold that indirect transfers are incapable of being taxed by India. It did not hold that Parliament lacks the power to tax them, retrospectively or otherwise. It did not constitutionalise any right to a frozen tax law. It held that the Act as enacted did not contain the charge, and it left the legislative door expressly open: Parliament could amend the law, and the judgment’s discussion of the draft code indicates that the Court expected any such charge to arrive by legislation. What the Court could not have prevented, and did not purport to prevent, was the choice Parliament then made about the temporal direction of that legislation. The dispute that followed was therefore not about judicial supremacy in any simple sense. It was about the difference between a legislature changing the law for the future after losing a case, which is an ordinary incident of parliamentary government, and a legislature changing the law for the past fifty years so that the case it lost is deemed never to have been correctly decided, which raises a distinct question about the security of adjudication itself.

What happened to the money Vodafone deposited with the Supreme Court?

The Supreme Court had required Vodafone to deposit Rs 2,500 crore and to furnish a bank guarantee of Rs 8,500 crore as a condition of its appeal being heard. Having allowed the appeal, the Court directed the department to return the Rs 2,500 crore deposit with interest at four per cent, and directed the return of the bank guarantee. The revenue thereby retained nothing from the largest tax demand it had then raised against a single transaction.

The operative order also set aside the penalty shadow: since the sum paid was not chargeable, there was no default in withholding, and the proceedings founded on sections 195 and 201 fell with the charge itself. Within weeks the deposit had been repaid, as the company confirmed after the review petition was dismissed. The financial position in the spring of 2012 was therefore complete and, on the law as the Supreme Court had declared it, indefeasible: no tax, no interest, no penalty, and the deposit returned with interest. Everything that happened afterwards happened because Parliament changed the law on which that position rested, and changed it backwards.

The reversal: the Finance Act 2012

The Finance Bill 2012 was introduced with the Budget on 16 March 2012 by the Finance Minister, Pranab Mukherjee. Its direct tax proposals contained a package of amendments directed, with unusual candour in the explanatory material and in ministerial statements, at the class of transaction the Supreme Court had just held to be outside the charge. The Finance Act, 2012 received the assent of the President on 28 May 2012. Its relevant provisions did four things.

The first was to insert two explanations into section 9(1)(i), each expressed to operate with retrospective effect from 1 April 1962, the date of commencement of the Income-tax Act itself. Explanation 4 clarified that the expression “through” in the clause means and has always meant “by means of”, “in consequence of” or “by reason of”. Explanation 5, the operative provision, clarified that an asset or capital asset, being any share or interest in a company or entity registered or incorporated outside India, is deemed to be and has always been deemed to be situated in India if the share or interest derives, directly or indirectly, its value substantially from assets located in India. The second was to make companion retrospective amendments to the definitions in section 2, expanding “capital asset” and “transfer” so that the indirect transfer theory ran through the Act’s charging and computation provisions and not merely through section 9, and to amend section 195 so that the withholding obligation applied to payments between non-residents regardless of whether the payer had any presence in India. The third is described in the next paragraph. The fourth, quieter but significant, was administrative: on 29 May 2012 the Central Board of Direct Taxes instructed that assessments completed before 1 April 2012 would not be reopened on the strength of the amendments, a concession to finality at the margins which confirmed, by its necessity, how far the amendments otherwise reached.

The third element was the validation clause, section 119 of the Finance Act, 2012. A retrospective amendment of the charging provision would by itself have faced an argument that demands and notices issued under the unamended law, and struck down by the courts, could not be revived merely because the substantive law had changed. The validation clause addressed that argument directly, providing that notices, assessments and demands founded on the indirect transfer theory would be deemed valid notwithstanding any judgment to the contrary. It was the clause that converted a general change in the law into the resurrection of the Vodafone demand itself, and it is the clause that the Taxation Laws (Amendment) Act 2021 would eventually have to disapply in terms. Any account of the episode that describes the 2012 Act as a mere clarification of section 9 has not reckoned with section 119, whose only function was to overcome adjudication.

What is the difference between a validation clause and an ordinary amendment?

An ordinary amendment changes the law for cases governed by it, and a retrospective amendment changes the law for past cases as well, but neither by itself disturbs a judgment that has already struck down a specific demand. A validation clause goes the further step of deeming the invalidated acts valid notwithstanding the judgment, curing the defect the court identified so that the original demand revives in the hands of the authority that issued it.

The drafting technique of the explanations deserves a moment’s attention, because it is the point at which the constitutional argument, examined below, is located. The amendments were not framed as new charging provisions operating from 1962. They were framed as clarifications, introduced by the formula “for the removal of doubts, it is hereby clarified”, and Explanation 5 was written in the language of deemed always to have been. The legislative theory was therefore that the Act had always meant what the amendment said it meant, and that the Supreme Court had simply misread it. That theory had a long pedigree in Indian fiscal legislation, where clarificatory explanations are a standard device for resolving interpretive disputes, and it had a real legal consequence: a genuinely clarificatory amendment is conventionally treated as declaratory of the law as it always stood, and its retrospective operation is correspondingly unremarkable. The controversy of 2012 turned on whether that description could honestly be applied to a provision that created a charge on a class of transaction which the Supreme Court, after the fullest argument, had held to be outside the Act. The government of the day, speaking through the Finance Minister, maintained that the amendments restored the original legislative intent and protected the Indian tax base against the erosion of source taxation. The opposition, in Parliament and outside it, characterised the package as a retrospective overruling of the Court dressed in clarificatory language. Both descriptions are recorded here because the article returns to the distinction when it states the two cases at full strength.

The scale of what was at stake was also stated at the time. The government indicated in the parliamentary debates of March 2012 that transactions of the Vodafone type, involving indirect transfers of Indian assets, represented a potential tax base of about Rs 40,000 crore across the cases within the department’s contemplation. The amendments were accordingly not a single-case measure in design, whatever their immediate target; they were a general reopening of a class, reaching back to 1962, under which any completed offshore transfer of a foreign holding company with substantial Indian assets became potentially chargeable, subject to the limitation provisions of the Act and to the Board’s self-denying instruction on completed assessments. Fresh demands followed, including a revived and re-quantified demand on Vodafone in 2013 as interest continued to accrue on the original figure.

The assurances with which the amendment was carried through Parliament belong in the record, because later events tested each of them. Moving the Finance Bill, the Finance Minister, Pranab Mukherjee, maintained that the amendments were clarificatory of the law as it had always stood, that they were directed at transactions in which tax had not been paid on value created in India, and that they would not override India’s double taxation avoidance agreements, since where a treaty protected a transaction the treaty position would prevail. He further indicated that the amendments were not intended to reopen cases in which assessments had been completed and taxes paid. The Central Board of Direct Taxes gave administrative effect to part of that assurance by an instruction of 29 May 2012 directing that completed assessments not be reopened on the strength of the retrospective changes. The residue, however, was considerable: every case in which a demand was unpaid, contested or unassessed stood exposed, and it was within that residue that the seventeen demands later counted by the 2021 Act were raised. The treaty assurance, too, proved accurate only in the forum the Minister had in mind. The amendments did not override the tax treaties as a matter of domestic law, as the Sanofi Pasteur judgment later confirmed they could not. What they overrode, in the tribunals’ analysis, was a different treaty altogether: the investment treaties, whose fair and equitable treatment standard the retrospective application was held to breach. The assurance and the breach are both part of the record, and the distance between them is a measure of how incompletely the consequences of the drafting had been mapped in March 2012.

The amendment’s wider family

Explanation 5 did not travel alone, and a reader who confines attention to it will misunderstand the engineering of the Finance Act, 2012. The Act amended the charging and definitional provisions of the Income-tax Act as a set, so that the indirect transfer theory would hold at every point at which the Vodafone litigation had exposed a joint. The definition of “capital asset” in section 2(14) was amended by an Explanation declaring it to include any rights in or in relation to an Indian company, including rights of management or control, held directly or indirectly. The definition of “transfer” in section 2(47) was amended to declare that a transfer includes the disposing of or parting with an asset or any interest in it, or the creation of an interest in any asset, in any manner whatsoever, directly or indirectly, on an absolute or a conditional basis, voluntarily or involuntarily. Section 195, the withholding provision on which the default case against Vodafone had been built, was amended to declare that its obligation applied and had always applied in respect of income chargeable under the Act, whether or not the payer or the payee had a residence, place of business or any other presence in India. Each of these amendments was likewise expressed to operate from 1 April 1962.

The pattern repays a moment’s study, because it shows what a legislature must do when it reverses a judgment that rested on several independent grounds. The Supreme Court’s decision had turned on the situs of the asset, on the character of the charge, on the genuineness of the structure and on the absence of a look through warrant. An amendment confined to Explanation 5 would have left the revenue exposed on the definitions at the next hearing: a transferor could have argued that what passed was not a capital asset within section 2(14), or that no transfer within section 2(47) had occurred in India, or that a foreign payer with no presence in India stood outside section 195 altogether. The 2012 Act therefore rewrote the definitions in parallel and retrospectively, so that each element of the charge would be found in the Act as amended for every year since 1962. The drafting is, on its own terms, thorough. It is also the clearest internal evidence against the clarificatory characterisation. A clarification resolves a doubt about existing language; it does not ordinarily need to reconstruct four provisions at once and validate the demands founded on the reconstruction. The companion amendments are the amendment’s confession that the law of 1962 to 2012, read as a whole, did not contain the charge, and that containing it had to be manufactured across the architecture of the Act.

The same Finance Act carried a second measure whose fate throws the first into relief. The general anti-avoidance rule in Chapter X-A, long proposed as the principled instrument against abusive structures, was enacted in 2012 but deferred, and deferred again after the Shome Committee’s intervention, taking effect only from assessment year 2018-19. The contrast is instructive for the series thesis. The instrument that would have tested structures prospectively, by criteria of substance, was postponed for half a decade out of concern for the investment climate. The instrument that reversed a single decided case with fifty years of retrospectivity was brought into force at once. A state confident in its anti-avoidance doctrine does not need the second instrument at all, since any abusive indirect transfer could be met under the first. The sequence in which the two were commenced records a choice about which kind of power the state preferred to hold, and it is the choice that the arbitral tribunals later priced.

The wider 2012 pattern

The indirect transfer amendments were the most consequential of the 2012 Budget’s backwards steps, but they were not the only ones, and the pattern matters because it shows that the Vodafone reversal was an instance of a governing technique rather than an isolated reaction to a single defeat. The same Finance Act rewrote the definition of “royalty” in section 9(1)(vi) by Explanations declared to operate from 1 June 1976, the date that provision had commenced, so as to include payments for the use of computer software and for transmission by satellite and similar processes, whatever the terms on which the software or the process was supplied. That amendment, too, reversed the drift of decided cases that had treated payments for shrink-wrapped software as payments for a copyrighted article rather than for the copyright itself. Its afterlife ran on a parallel track to Vodafone’s. In Engineering Analysis Centre of Excellence v Commissioner of Income-tax in 2021, the Supreme Court held that amounts paid by Indian distributors and end users to foreign suppliers for computer software under end user licence agreements were not royalty under the applicable tax treaties, and that the domestic definition, however retrospectively widened, could not enlarge the treaty charge. The technique met the same boundary in both sagas: retrospective words could rewrite the Act, but they could not rewrite India’s treaties, and where a treaty allocated the taxing right away from India the amendment stopped at the treaty line.

Other amendments in the same budget season carried shorter retrospectivity and attracted less notice, but the cumulative impression on investors and advisers was of a state reaching for the past as a first resort: transfer pricing adjustments extended to specified domestic transactions, reporting and documentation obligations enlarged, and the general anti-avoidance rule announced for commencement in the following year before being deferred. The Shome Committee’s work has to be read against that whole canvas, since it was appointed in the first instance on the anti-avoidance proposals and asked to consult on the climate the Budget had created. When the committee laid down its rarest of rare test for retrospective legislation, it was answering not one amendment but a season of them, and its central insight generalises: a legislature that legislates backwards as a routine instrument teaches its taxpayers to price that risk into every transaction, and the price is paid by the jurisdiction on investments that have nothing to do with the original dispute. The Vodafone amendment is remembered because the bill for that lesson was presented in public, in foreign courts, in a currency other than rupees.

The two cases, stated at full strength

The stance this article takes toward the argument of 2012 is that both sides were making a real point, and that the subsequent history vindicated each side in a different forum. The cases deserve to be stated as their makers stated them.

The government’s case began with the economics of the transaction. A business built entirely in India, on Indian spectrum allocated by the Indian state, serving Indian subscribers and employing Indian staff, had changed hands for about 11.1 billion dollars, and the gain embedded in that change of control accrued, in every economic sense, from Indian assets and Indian policy. To hold that the gain escaped Indian tax because the share that formally moved was registered in the Cayman Islands was, on this view, to subordinate the tax base of a developing economy to the geography of incorporation. Source based taxation of gains on the disposal of domestic businesses is not an Indian eccentricity; most capital importing states assert it, and the question was only whether India’s statute had asserted it in terms. The revenue’s lawyers added that the Act had always been intended to reach such transfers, that the explanations of 2012 declared rather than created the law, and that Parliament’s competence to enact fiscal legislation retrospectively, including legislation that alters the basis of a decided case, was settled constitutional doctrine and not a novelty invented for Vodafone. On the policy plane, ministers argued that a contrary rule invited every future exit from an Indian investment to be routed through an offshore holding company, hollowing out the capital gains base precisely where the gains were largest. The strongest form of the government’s case, and the form in which its own expert committee would later engage with it, accepts that certainty matters but insists that a statute should be judged by whether the charge it imposes is substantively justified, and that a charge on the transfer of an Indian business is justified wherever the documents happen to be signed.

The critics’ case, made by the taxpayer, by much of the Indian and international business press, and in substance by the government’s own expert committee, began from the decided case. Whatever the merits of taxing indirect transfers prospectively, the Supreme Court had heard the revenue’s arguments in full and had rejected them in January 2012, in a judgment that turned on the enacted text. To respond by deeming the text to have meant the opposite since 1962 was not, on this view, clarification but reversal under a false label, and the validation clause removed any doubt about the object: it existed to revive a specific demand that the Court had annulled. The critics added that the burden fell on conduct that was lawful when undertaken and incapable of being rearranged after the event; that interest and penalty for a failure to withhold tax that the Supreme Court had held was never due compounded the unfairness; and that the cost would be paid not by Vodafone alone but by India’s reputation as a jurisdiction in which a final judgment of the highest court settles a dispute. The strongest form of this case concedes Parliament’s formal competence and attacks the exercise: a power may be constitutionally available and still be ruinously expensive to use, and the measure of the 2012 amendment should be what it collected in the end against what it cost in the end.

The government’s case had a further limb that is easily forgotten because it was addressed to Parliament rather than to the press. The judgment, ministers argued, had not merely decided one dispute; it had published a template. If the transfer of an Indian business through an offshore holding company lay outside the charge, then every exit from every Indian investment could be routed through a holding jurisdiction, and the capital gains base on the sale of Indian businesses would become voluntary for any investor sophisticated enough to interpose a foreign company at acquisition. The revenue’s figure of about Rs 40,000 crore, given to Parliament in the March 2012 debates as the scale of the transactions of this type, was offered as the measure of that exposure. On this limb of the case, delay was itself a fiscal event: each year spent legislating prospectively would be a year in which pending exits were restructured to beat the new charge, and the amendment’s reach back to 1962 was defended as the only reach long enough to catch the template at its source. The argument was not ignoble, and versions of it have persuaded legislatures elsewhere to act against disclosed avoidance with effect from the date of an announcement. Its fatal weakness in 2012 was that the announcement it relied on was a judgment for the taxpayer, delivered on the law as it stood when the transactions closed; and a state that treats its own Supreme Court’s exposition of the law as the starting gun for reversing that exposition backwards should not be surprised when investors reprice the value of winning in its courts.

Political attribution requires the same care as legal attribution, because the episode spanned three governments in the telling and two in fact. The retrospective amendment was enacted in 2012 by the United Progressive Alliance government, in a Budget presented by Pranab Mukherjee. The committee whose recommendations might have softened it was appointed by the same government later that year and its report was not acted on by that government. The National Democratic Alliance government that took office in 2014 had criticised retrospective taxation in opposition and gave administrative assurances against fresh retrospective demands, but it left the 2012 amendment and the pending Vodafone and Cairn demands in place for seven years, defended the demands in arbitration, and settled only after losing both arbitrations. The withdrawal of 2021 was enacted by that second government, in a Bill introduced by the Finance Minister, Nirmala Sitharaman. Credit and blame assigned to a single administration, in either direction, do not survive the chronology.

The committee that was not followed

In 2012 the government constituted an expert committee under the chairmanship of Parthasarathi Shome, initially to consult on the general anti-avoidance rules and then to examine the retrospective amendments relating to indirect transfers. The committee’s draft report on the retrospective amendments was released in October 2012, and its reasoning is the most authoritative internal critique the amendment ever received, because it proceeded from within the revenue tradition rather than from the taxpayer’s camp. The committee concluded that retrospective application of a tax law should occur only in exceptional cases, in the rarest of rare situations: to correct apparent mistakes or anomalies in the statute, to give effect to provisions that are genuinely clarificatory, or to protect the tax base from highly abusive schemes lacking economic substance. It should not be used to expand the tax base. Applying that test, the committee held that the indirect transfer provisions introduced by the Finance Act, 2012 were not clarificatory in nature and would tend to widen the tax base, and it recommended that, once the operative definitions were clarified, the provisions should be applied prospectively.

The committee also addressed the position of taxpayers caught by the amendment if the government nevertheless insisted on retrospectivity. It recommended that no person be treated as an assessee in default under section 201, or as a representative assessee of a non-resident, in respect of an indirect transfer undertaken on the law as it had been understood, and that neither interest nor penalty should attach to demands raised by reason of the retrospective amendment. Each of those recommendations engages a real feature of the 2012 package: the default theory under section 201 was the very mechanism by which Vodafone had been pursued, and interest and penalty formed the majority of the eventual exposure in several of the affected cases. The report was not acted on. The finance ministry of the day received it, the government changed in 2014, and the amendment survived both, defended thereafter on the narrower administrative assurance that it would not be invoked to open fresh cases without high level approval. When the withdrawal finally came in 2021, its architecture vindicated the committee’s analysis in almost every particular: prospective operation restored, default consequences removed for the covered class, and the sums collected returned. The nine year interval between the recommendation and its adoption is the cleanest measure in the entire episode of the distance between the state’s expert knowledge and its litigation interest.

The committee’s comparative survey, recorded in its report, placed the Indian practice in a wider frame that has lost none of its force. Examining the practice of other jurisdictions, the report treated retrospective changes to tax law as exceptional in comparable systems, generally confined to forestalling avoidance disclosed after the event and frequently tied in their effect to the date on which the change was announced, so that a taxpayer on notice could not complain of surprise. The Indian amendments of 2012 failed each element of that pattern. They were not confined to a disclosed avoidance scheme, since the structure they caught had been examined and upheld by the Supreme Court on its facts. They were not tied to any announcement, the first public indication of the change being the Finance Bill itself, introduced after the judgment it reversed. And their reach, fifty years to the commencement of the parent Act, exceeded anything the comparative survey could parallel. The report’s restraint in stating these conclusions, in the flat prose of an official committee, should not disguise what they amount to: a finding, by the revenue’s own experts, that the 2012 exercise lay outside the accepted practice of the states whose investment India was competing to attract. Successive governments filed that finding and proceeded anyway, which is why the committee section of this record is the one that most damages every administration that handled the dispute.

The first attempted settlement, 2013 to 2014

The dispute came within reach of a negotiated end once, long before the arbitrations, and the failure of that attempt is part of the record of how the amendment was administered. On 3 January 2013 the department wrote to Vodafone reviving the demand in an updated amount of Rs 14,200 crore, the original tax with the interest element recomputed. Within months, however, the same government that had enacted the amendment was exploring a way out of it. The Union Cabinet in June 2013 approved a Finance Ministry proposal for a non-binding conciliation with Vodafone, Finance Minister P. Chidambaram explaining that any outcome acceptable to the government would have to return to the Cabinet and then to Parliament, since settling the liability required amending the Act. The Attorney General’s advice in favour of exploring conciliation, given after the Law Ministry changed hands, marked a formal softening inside the government: the executive that had defended the amendment as indispensable was, within a year, prepared to legislate again to be rid of its consequences in the flagship case.

The subsidiary dispute over which the conciliation foundered deserves a paragraph of its own, because it illustrates how widely the retrospective season had spread across the group’s affairs. The transfer pricing case concerned Vodafone India Services, the group’s Indian services company, which had issued shares to its Mauritius parent in a rights issue and held call options over shares in the Indian operating structure. The department’s transfer pricing officer treated the share issue as undervalued and the options as an international transaction, producing an adjustment reported at about Rs 3,700 crore. In October 2013 the Bombay High Court held for Vodafone on the threshold question: the issue of shares by an Indian company to its parent did not give rise to income or to an international transaction chargeable under the transfer pricing code, and the jurisdiction assumed over it failed. The ruling was one of a series in those years in which the courts confined the revenue’s expansive readings of the Act’s machinery provisions, and it strengthened Vodafone’s conviction that its disputes were better fought in court and in arbitration than settled at a discount. That conviction, the government’s insistence that the transfer pricing case stood outside any settlement of the capital gains demand, and the disagreement over the conciliation framework together ensured that the 2013 opening closed unused.

The conciliation never began. The parties divided over the framework, Vodafone preferring conciliation under the United Nations Commission on International Trade Law rules and the government insisting on the Indian Arbitration and Conciliation Act, and over scope, Vodafone seeking to include a separate transfer pricing dispute of about Rs 3,700 crore concerning its Indian services company and the Finance Ministry refusing to club the two. By early 2014 the government had moved to withdraw the offer, the revenue department resumed pursuit of the demand with interest and penalty, and Vodafone issued its supplementary notice invoking the investment treaty, commencing the arbitration that the conciliation had been designed to forestall. The episode matters to the analysis for a reason beyond chronology. It demonstrates that the costs of the amendment were visible to its own authors within a year of enactment, that an exit was attempted while the amendment was still young, and that the exit failed not on the principle of settlement but on forum and scope. The nine years that followed were therefore not nine years of uniform official conviction. They were nine years in which successive governments, each aware of the price, declined to pay the political price of legislating a withdrawal, until an arbitral enforcement campaign against Indian sovereign property abroad made withdrawal the cheaper option.

The constitutional doctrine on legislating backwards

The Vodafone argument is often misreported as having established that retrospective taxation is unconstitutional in India. It established nothing of the kind, and the constitutional position before, during and after the episode is uniform enough to be stated in a short sequence of propositions. Article 265 of the Constitution requires that no tax be levied or collected except by authority of law, and the Finance Act 2012 was law. Article 20(1), which prohibits retrospective criminal liability, has no application to a fiscal charge, which is civil in character. The Supreme Court had accepted as early as Rai Ramkrishna v State of Bihar in 1963 that a legislature may impose a tax with retrospective effect, the enquiry being not whether retrospectivity is permissible but whether the particular law offends some other constitutional limitation. And a legislature that loses a case on the meaning of its statute may respond by amending the statute, including retrospectively, provided it genuinely alters the legal basis on which the judgment rested rather than purporting simply to declare the judgment wrong: that validating technique, articulated in cases such as Shri Prithvi Cotton Mills v Broach Borough Municipality in 1969 and systematised in the Indian Aluminium case in 1996, is a recognised incident of legislative power in India.

A further strand, less noticed, concerns the territorial competence that underlay the whole dispute. The charge on an indirect transfer is by design a charge with extra-territorial operation: it fastens on a transaction between two foreign persons, executed abroad, in respect of a foreign asset, because of the Indian value beneath it. In G.V.K. Industries Ltd v Income Tax Officer, a Constitution Bench held in 2011 that Parliament may legislate with respect to extra-territorial aspects or causes only where those aspects have a nexus with India, a connection that must be real and not illusory, and that Parliament has no power to legislate for a foreign territory as such. The Vodafone charge comfortably satisfied that requirement once enacted, since the value taxed was value located in India and realised from an Indian business. But the G.V.K. discipline explains the shape the litigation took before 2012. The revenue’s difficulty was never nexus in the constitutional sense; the Indian connection was obvious. Its difficulty was statutory: the nexus that the Constitution permits Parliament to use still has to be carried into the charging provision by words that reach it, and the unamended section 9(1)(i), as the Supreme Court read it, had not done so. Competence to legislate and legislation enacted are different things, and the distance between them is where the Vodafone case was lost and won.

The genuinely contestable ground lay elsewhere, in the doctrine of clarificatory amendments. Indian courts accept that an amendment which genuinely clarifies or supplies an evident omission may be read as declaratory of the law as it always stood, and such amendments routinely operate for past years. But the label is not conclusive. The question is one of substance: whether the provision, before amendment, could bear the meaning the amendment attributes to it. A Constitution Bench in Vatika Township v Union of India in 2014 restated the allied presumption that legislation is prospective unless retrospectivity is expressed or necessarily implied, and the jurisprudence on explanations treats the legislature’s description of its own amendment as clarificatory as a factor, not a verdict. It was open to a court, consistently with that jurisprudence, to hold that Explanation 5 was in substance a new charge dressed as a clarification, because the Supreme Court had held in 2012, on the fullest consideration, that the unamended words would not bear the revenue’s meaning. No Indian court was in the event required to decide that question in the Vodafone matter itself, since the dispute migrated to treaty arbitration and then to settlement, and it remains formally undecided at the level of the Supreme Court. That is a precise and significant sense in which the episode settled less than is commonly supposed: the constitutional argument at its heart was constructed, briefed and publicly debated for a decade without ever receiving an authoritative answer.

The same question of temporal reach recurs across this cluster of statutes, which is why the episode is best read as one instance of a general Indian legislative habit rather than as a singularity. The Black Money Act 2015 taxes undisclosed foreign income and assets by reference to a valuation date and reaches assets acquired in years before the Act commenced, raising in a different register the question of how far a charging statute may travel backwards. A third instance, over the prosecution and attachment provisions directed at transactions completed before the operative amendment, runs through the benami transactions legislation. In each case the legislature’s instinct is to make the new law speak to the old facts, and in each case the courts are asked to locate the line between legitimate retrospectivity and the unsettling of closed transactions. Vodafone remains the largest and most expensive of these disputes, but it is not doctrinally alone.

Two governments, one amendment

The amendment outlived the government that enacted it, and the conduct of its successor is as much a part of the record as its enactment. The government that took office in 2014 had criticised retrospective taxation in opposition and arrived with a manifesto commitment to a non-adversarial tax regime. Its first Finance Minister, Arun Jaitley, stated in his 2014 Budget speech that the government would not ordinarily bring about a change retrospectively which creates a fresh liability, and that disputes already in existence would be allowed to reach their logical conclusion through the legal process. The formulation was carefully drawn. It disowned the practice for the future while expressly preserving the 2012 amendment and the demands under it, which were left to the very arbitral process the new government then defended. For seven years that position held: the amendment stood unrepealed, the Vodafone demand was maintained and its interest computed, the Cairn recovery was retained and defended, and India’s counsel argued before both tribunals that the retrospective tax was a legitimate exercise of sovereign power with which the treaties did not interfere.

The texture of those seven years was not quiet. In February 2016, on the eve of a flagship investment promotion week, Vodafone confirmed that it had received a fresh notice from the department in respect of the revived demand, by then stated at Rs 14,200 crore with interest, warning that its Indian assets could be seized if the sum went unpaid. The company replied publicly that the dispute was in arbitration and that the notice contradicted the government’s own assurances of a stable tax regime. Ministers responded that the notice was a routine step in an existing proceeding, which in the narrow sense it was, since the demand had never been withdrawn. The exchange captured the equilibrium the successor government had chosen: the amendment and its demands would be neither enforced to a conclusion nor surrendered, but held, with the arbitrations left to decide what domestic politics would not. Vodafone’s Indian operating business, by then one of the largest in the group, traded on under the shadow of a liability that belonged legally to its parent and commercially to no one, and in 2018 the Indian business was merged with Idea Cellular in a consolidation driven by the price war in Indian telecommunications, a transaction with no connection to the tax dispute beyond sharing a corporate family with its most famous member.

The even-handedness the stance charter requires is simply the record. The amendment was enacted by a Congress-led government against the advice it had itself commissioned, and that government attempted within a year to conciliate its way out of the flagship case. It was retained by a Bharatiya Janata Party-led government that had condemned the practice, defended through two arbitrations by that government, and withdrawn by that government in 2021 under the pressure of enforcement proceedings, on settlement terms that required the taxpayers to abandon their awards and accept refunds without interest. Neither government’s record supports the partisan simplification in either direction. The lesson the sequence teaches is institutional rather than political: retrospective legislation of this kind creates commitments that bind successors of every party, because repeal concedes revenue and invites the charge of favouring the taxpayer, while retention transfers the contest to treaty tribunals whose awards bind the state regardless of which party holds office. The amendment’s longevity is explained less by anyone’s continuing belief in it than by the asymmetry of exit costs, and that asymmetry is itself a reason for legislatures to hesitate before enacting such measures: they are easy to pass and, as nine years demonstrated, exceptionally hard to leave.

The nine-year round trip

The sequence from the Supreme Court’s judgment to the statutory withdrawal occupied nine years, from January 2012 to the settlements that followed the Taxation Laws (Amendment) Act in the autumn of 2021. Including the transaction and the domestic litigation that preceded the judgment, the full arc runs to fourteen years. The table below is the episode in one view. Its final row records the outcome the whole sequence produced: the taxpayer kept the treatment the Supreme Court had awarded it in 2012, the revenue collected nothing on the original transaction, and each side had borne a decade of costs to arrive at the position the judgment had already established.

Stage	What happened	The taxpayer’s position	The revenue’s position
February to May 2007	Hutchison sells CGP, a Cayman Islands holding company, to Vodafone’s Netherlands company for about 11.1 billion US dollars, transferring control of Hutchison Essar in India	The sale is of a foreign share, outside Indian tax jurisdiction	No position yet taken
September 2007	The department issues a show cause notice alleging a failure to withhold tax under sections 195 and 201, contemplating about 2.1 billion US dollars	The notice is without jurisdiction and is challenged in the Bombay High Court	The substance of the deal is a transfer of Indian assets and the buyer was bound to withhold
December 2008 to May 2010	The Bombay High Court lets the inquiry proceed; the department holds in May 2010 that it has jurisdiction	The jurisdictional objection is renewed after the department’s order	Jurisdiction affirmed at the administrative stage
September to October 2010	The Bombay High Court upholds the department in substance on 8 September 2010; a demand of Rs 11,218 crore follows, about Rs 7,900 crore tax and the rest interest	Appeal to the Supreme Court; deposit of Rs 2,500 crore and a guarantee of Rs 8,500 crore furnished	The gain is taxable to the extent attributable to Indian assets
20 January 2012	The Supreme Court holds for Vodafone: section 9(1)(i) is not a look through provision and the transaction is outside the charge; the deposit is ordered back with interest at four per cent	Complete vindication on the law as enacted	Total defeat; a review petition is filed
March to May 2012	The review is dismissed on 20 March 2012; the Finance Act, 2012 inserts Explanations 4 and 5 into section 9(1)(i) with effect from 1 April 1962 and validates demands by section 119	The decided case is reversed by legislation reaching back fifty years	The law is clarified to state what it always meant
October 2012	The Shome Committee recommends that the indirect transfer provisions operate prospectively and that no default, interest or penalty attach if they do not	The government’s own experts endorse the taxpayer’s critique	The recommendations are received and not acted on
2013 to 2016	A fresh demand is served on Vodafone as interest accrues; Vodafone invokes the India and Netherlands treaty in 2014; Cairn receives a demand of Rs 10,247 crore in 2014 on its 2006 reorganisation and invokes the India and United Kingdom treaty in 2015	The dispute moves from domestic law to treaty protection	Both demands are defended; about Rs 7,900 crore is recovered from Cairn by selling its shares and seizing dividends and refunds
25 September 2020	A tribunal under the India and Netherlands treaty holds unanimously that imposing the tax with interest and penalties breached fair and equitable treatment, and directs India to cease enforcement	The retrospective demand is internationally unlawful even if domestically valid	The award is challenged at the seat of the arbitration
21 December 2020	A tribunal under the India and United Kingdom treaty awards Cairn damages of 1.2 billion US dollars plus interest and costs, seated at The Hague	The tax recovered must be restored with damages	A setting aside application is filed at The Hague in March 2021
June to November 2021	Cairn obtains orders freezing Indian government properties in Paris and sues Air India in New York; the Taxation Laws (Amendment) Act, 2021 receives assent on 13 August 2021; Cairn gives the statutory undertakings in November and a refund of about Rs 7,900 crore follows	Enforceable rights under the awards are surrendered in exchange for refunds and nullified demands, without interest	The demands are withdrawn as a settlement to restore investor confidence, not as an admission that the amendment was unlawful
The end position	No tax stands collected on the 2007 transaction; the Cairn recovery is refunded without interest; the arbitral claims are abandoned	Substantially where the Supreme Court left it in 2012, after nine further years of litigation	Substantially where the Supreme Court left it in 2012, having borne the costs of the arbitrations and the reputational damage recorded against the amendment

Practitioners who have to explain this sequence to a board or a classroom usually reduce it to the table and a single slide, and the chronology travels well in that form; a reader who receives it as a presentation file can open it in a PowerPoint viewer in the browser and check each stage against the sections above. What the compression cannot show is the distribution of the costs. Vodafone’s exposure was a liability contingent on proceedings it kept winning; it never paid the tax. Cairn’s position was worse in kind: the department had actually collected about Rs 7,900 crore from it by coercive recovery before the award, so Cairn litigated for the return of its own money, and recovered it in the end without interest for the years of its detention. The revenue’s costs were of a third kind: the legal costs of two arbitrations it lost, a reported direction to reimburse part of Vodafone’s costs put at 4.3 million pounds, and the less quantifiable but more durable cost of a decade in which every investor presentation on India carried a slide about this dispute.

The international round: two arbitrations

The migration of the dispute from Indian courts to international tribunals was possible because India’s bilateral investment treaties promise foreign investors fair and equitable treatment and provide for arbitration of disputes between the investor and the host state. The treaties do not prohibit taxation, and they do not prohibit retrospective legislation in the abstract. What they supply is a standard against which the manner of a state’s conduct toward a specific investment can be measured, and an enforcement mechanism outside the state’s own courts. Both Vodafone and Cairn invoked that machinery, and the difference between their cases is instructive: Vodafone’s complaint was about a demand revived after a final judgment in its favour, while Cairn’s was about a demand raised for the first time under the retrospective amendment against a transaction that predated even the Vodafone judgment.

Why could the dispute leave the Indian courts for an arbitral tribunal?

India’s investment treaties with the Netherlands and the United Kingdom allowed investors of those states to submit disputes with India to international arbitration rather than relying only on Indian courts. The tribunals constituted under those treaties could not strike down the Finance Act, and did not purport to. They could decide whether India’s conduct toward the investment breached the treaty standard of fair and equitable treatment, and award damages or direct that enforcement cease.

Vodafone invoked the India and Netherlands treaty, commencing arbitration in 2014 after the revived demand had been served, and a second invocation under the India and United Kingdom treaty produced proceedings that the Delhi High Court restrained as duplicative. The principal tribunal, administered by the Permanent Court of Arbitration, delivered a unanimous award on 25 September 2020. It held that the imposition of the tax liability on Vodafone, together with interest and penalties, through the retrospective amendment, breached the fair and equitable treatment guarantee in Article 4(1) of the treaty, and that India was obliged to cease enforcement of the demand, any attempt at which would engage its international responsibility. The tribunal did not hold that India could not tax indirect transfers, and it did not hold that retrospective legislation is contrary to international law in itself. Its reasoning turned on the treatment of this investor: a liability imposed backwards over a transaction that the Supreme Court of the host state had held to be outside the statute, pursued with interest and penalty, fell short of the stability and predictability the treaty standard protects. India challenged the award at the seat of the arbitration.

Cairn’s case arose from a different transaction inside the same statutory amendment. In 2006, as part of a reorganisation preceding the public listing of its Indian subsidiary, Cairn’s United Kingdom parent transferred the shares of the group’s Indian holding company down the chain to the Indian company that was to be listed. No outside buyer was involved and no value left the group; the transfer positioned the Indian assets under the listing vehicle. In January 2014 the department issued a draft assessment contending that the 2006 reorganisation had generated capital gains chargeable under the Act as retrospectively amended, and raised a demand of Rs 10,247 crore. Cairn disputed the demand, and the department proceeded to recover a large part of it by force of the state’s collection powers: it sold Cairn’s residual shareholding in the Indian company, appropriated dividends payable to Cairn and withheld tax refunds, assembling in aggregate about Rs 7,900 crore. Cairn invoked the India and United Kingdom treaty in 2015. The tribunal, seated at The Hague, delivered a unanimous award on 21 December 2020 holding that India had breached fair and equitable treatment, that the tax demand was not a fair or equitable application of the amended law to Cairn’s investment, and awarding damages of 1.2 billion US dollars together with interest and costs, the interest carrying the enforceable total to about 1.72 billion dollars as Cairn later computed it. India applied to set the award aside at The Hague in March 2021.

Two features of the Cairn recovery distinguish it from every other episode in this article and explain why Cairn, rather than Vodafone, supplied the enforcement drama of 2021. The first is that the department had restrained Cairn from disposing of its residual shareholding in Cairn India, its former listed subsidiary, while the claim was pursued, so the company’s principal remaining asset in India was frozen at the outset. The second is that the recovery, when it came, was executed rather than merely demanded. Over the years that followed, the department sold the seized shareholding in tranches, appropriated dividends declared on it and applied tax refunds due to Cairn against the demand, assembling about Rs 7,900 crore in all. Cairn was therefore in the position that Vodafone had avoided throughout: it had actually borne the tax, on a transaction internal to its own group, under a law enacted eight years after the transaction closed, and it litigated at The Hague not to resist a liability but to recover its own money with compensation for its detention. The tribunal’s award of damages, set at 1.2 billion dollars with interest, was restitutionary in substance even though treaty law frames it as compensation for breach. That character mattered when the settlement came, because the 2021 Act required Cairn to surrender the award and its accrued interest in exchange for a refund of the same principal without interest: the state returned what it had taken, kept the time value of it for the years of detention, and extinguished a larger enforceable liability at a discount measured by the interest waived on both sides.

The enforcement phase converted an abstract legal defeat into a diplomatic embarrassment of a concrete kind. Cairn registered the award in multiple jurisdictions and moved against Indian state assets abroad: in June 2021 a French court authorised the freezing, by way of judicial mortgage, of about twenty Indian government properties in Paris, and in New York Cairn sued Air India as an alter ego of the Indian state against whose assets the award might be executed. Whatever the ultimate legal fate of those manoeuvres, their political meaning was plain. The retrospective amendment, defended in 2012 as a protection of the Indian tax base, had produced a position in which Indian sovereign property abroad was under attachment at the suit of a company from which the tax had already been collected. It was against that background, and not in a fit of abstract reconsideration, that the withdrawal legislation was introduced in August 2021.

The withdrawal of 2021

The Taxation Laws (Amendment) Bill was introduced in the Lok Sabha on 5 August 2021, passed both Houses within days, and received the President’s assent on 13 August 2021 as Act 34 of 2021. Its operative provisions are short and precisely targeted. It amended section 9 of the Income-tax Act to provide that no tax demand shall be raised in future on the basis of the Finance Act, 2012 amendments in respect of an indirect transfer of Indian assets undertaken before 28 May 2012, that date being the day the Finance Act, 2012 itself received assent. It provided that demands already raised for such transfers, including demands sustained by the validation clause in section 119 of the Finance Act, 2012, would be nullified on fulfilment of specified conditions. And it provided that amounts paid or collected in such cases would be refunded, without interest. The statement of objects and reasons recorded that demands had been raised in seventeen cases under the retrospective amendment, that arbitration had been invoked in four of them under India’s investment treaties, and that tribunals had ruled against the government in two. The government presented the measure publicly as one to restore investor confidence and to end litigation that damaged India’s credibility, and the revenue secretary of the day was explicit that the state was not accepting the arbitral awards or conceding that the 2012 amendment had been beyond Parliament’s power.

The parliamentary passage of the withdrawal was swift and, by the standards of 2012, strikingly consensual in outcome if not in tone. Introducing the Bill, the Finance Minister, Nirmala Sitharaman, defended the sovereign right to tax retrospectively while characterising the 2012 amendment’s application as ill-advised, and presented the measure as the fulfilment of her government’s standing commitment against retrospective taxation that creates fresh liability. Opposition members supported the withdrawal while pressing the point that it had arrived only after India had lost both arbitrations and faced enforcement against its property abroad, a sequence they characterised as retreat rather than reform. The government did not accept that characterisation, and went out of its way to record that the sovereign right itself was untouched: the amendment was being disapplied for a closed class of cases, as a matter of policy, and not conceded to have been beyond Parliament’s competence. The distinction was not pedantry. It was the condition on which the withdrawal could be carried by a Parliament being asked to surrender demands worth thousands of crores, and it is the reason the article’s central claim is framed as withdrawal rather than abolition: the legislature that closed the episode took formal care to preserve the power that had opened it.

The conditions are the analytically important part, because they reveal the character of the instrument. Relief was not automatic. A taxpayer seeking nullification and refund had to withdraw, or undertake to withdraw, all pending litigation and arbitration in every forum, domestic and international; to furnish an undertaking that no claim for costs, damages or interest would be filed; and to satisfy the prescribed procedural requirements, for which rules were notified in October 2021. A taxpayer holding a favourable award, as both Cairn and Vodafone were, therefore had to surrender an enforceable international right as the price of recovering money that, in Cairn’s case, had been taken from it by coercive recovery. That is the structure of a settlement, not of a repeal accompanied by an apology, and the article’s central claim depends on the distinction. The state paid no interest on sums it had held for years, under a statute that its own Supreme Court’s reading of the predecessor law would never have authorised it to collect. Cairn gave the undertakings in November 2021, withdrew its enforcement proceedings across the jurisdictions in which it had registered the award, and received a refund of about Rs 7,900 crore. Vodafone likewise brought its proceedings to an end under the statutory framework, its revived demand falling to be nullified on the same conditions, and the claims in the remaining cases of the seventeen were closed out through the same machinery or lapsed with it.

What the 2021 Act did not do is as important as what it did. It did not repeal Explanation 5. It did not touch the taxation of indirect transfers undertaken on or after 28 May 2012, which remains part of the Act and is applied prospectively without serious controversy. It did not enact any general prohibition on retrospective tax legislation, nor any requirement of parliamentary super-majority, nor any compensation principle for taxpayers caught by future retrospective amendments. It did not overrule or even address the constitutional doctrine, described above, under which the 2012 amendment was defensible. And it was itself, in a quiet irony, retrospective in technique: it reached back to nullify demands that were, as a matter of domestic law at the time they were raised, raised under a valid statute. The legal position after 2021 is therefore exactly the legal position before 2012, plus a closed class of settled cases. Parliament may do again what it did in 2012. The only restraints are the political memory of what it cost and the treaty exposure that any repetition would attract, and both of those are matters of reputation and diplomacy rather than of law.

The law as it stands after the round trip

For a transaction undertaken after the 2012 cut-off, the charge on indirect transfers is a working part of the Act, and its administration has been progressively regularised in ways that owe something to the Shome critique. The Finance Act, 2015 supplied the definition that Explanation 5 had lacked. By the insertion of Explanation 6, a share or interest in a foreign company or entity is taken to derive its value substantially from Indian assets only where, on the specified date, the value of the assets located in India exceeds Rs 10 crore and represents at least fifty per cent of the value of all the assets owned by the company or entity. Explanation 7 carved out the small shareholder, excluding a transferor who holds no right of management or control and not more than five per cent of the voting power or share capital of the foreign entity. The same Finance Act introduced section 285A, obliging the Indian concern through which such assets are held to report the transaction, and the computation rules attribute to India only the proportion of the gain that corresponds to the Indian assets. The regime as regularised is recognisably the regime the Shome Committee had been willing to contemplate: defined thresholds, proportionate attribution, protection for portfolio holders, and prospective operation.

The machinery beneath the thresholds was completed by subordinate legislation. Rules 11UB and 11UC of the Income-tax Rules, inserted in June 2016, supply the valuation and attribution code that the 2012 amendment had omitted. Rule 11UB prescribes how the fair market value of assets located in India, and of all the assets of the foreign company or entity, is to be determined on the specified date, drawing on balance sheet values adjusted for assets whose value must be taken at market price. Rule 11UC prescribes the attribution formula: the income from the transfer is multiplied by the ratio of the fair market value of the Indian assets to the fair market value of all assets, so that only the Indian proportion of the gain is charged, and the transferor must file a report in Form 3CT, signed by an accountant, setting out the basis of the apportionment. Under section 285A, the Indian concern whose assets underlie the transfer must furnish the prescribed information in Form 49D to the department, so that the revenue is not dependent on the foreign transferor for knowledge of the transaction. The Central Board of Direct Taxes issued clarifications on the regime’s application to investment funds and other structures. Taken together with the 2015 thresholds, this is the administrative apparatus of a settled indirect transfer tax: defined triggers, proportionate attribution, independent reporting duties and a valuation code. Its existence is the final irony of the chronology. Everything required to tax indirect transfers lawfully and predictably was in place by 2016. It had simply arrived four years after the attempt to tax them backwards, and it operates only forwards.

Two lines of authority mark the boundaries of the charge in practice. The first concerns treaties. In Sanofi Pasteur Holding SA v Department of Revenue, decided in 2013, the Andhra Pradesh High Court considered the acquisition of the Shantha Biotechnics business through the purchase of a French company from another French holder. The revenue contended that the gain was taxable in India under the amended Act as an indirect transfer of Indian assets. The Court held that Article 14(5) of the India and France tax treaty confined the taxing right over such gains to France, and rejected the underlying assets theory as a route around the treaty allocation: where a treaty assigns the gain exclusively to the residence state, the domestic charge, however widely drawn after 2012, cannot be applied to defeat it. The department’s appeals in that line were in due course withdrawn before the Supreme Court. The proposition that survives is a structural one, and it disciplines every indirect transfer analysis since: the Act’s deeming provision operates subject to India’s treaties, and a transferor resident in a treaty state whose treaty reserves capital gains to the residence state stands outside the charge unless an anti-avoidance rule is successfully invoked against the arrangement.

The second line concerns that anti-avoidance qualification, and it produced the last act of the long story. Tiger Global’s Mauritius entities had acquired shares in Flipkart’s Singapore holding company between 2011 and 2015 and sold them in 2018, in the course of Walmart’s acquisition of Flipkart, claiming exemption under the India and Mauritius treaty, whose grandfathering clause protected gains on shares acquired before 1 April 2017. The Authority for Advance Rulings rejected their applications in March 2020 as prima facie designed for the avoidance of tax. The Delhi High Court reversed that view in August 2024, holding that a valid tax residency certificate is not to be disregarded on suspicion, that investments originating in Mauritius are not inherently suspect, and that the treaty’s express grandfathering protected the gains in any event. In January 2026 the Supreme Court set the High Court’s judgment aside and restored the Authority’s position, holding that the arrangement was an impermissible avoidance arrangement attracting the general anti-avoidance rule in Chapter X-A, and that gains on transfers effected after 1 April 2017 are taxable in India notwithstanding that the shares were acquired before that date. The reasoning is the counterpoint to Vodafone. The 2012 judgment protected a structure with demonstrated commercial substance from a charge Parliament had not enacted. The Tiger Global judgment withdraws protection from an arrangement found to lack substance, under an anti-avoidance code Parliament has enacted. Between those two poles, substance and statutory warrant, the law of indirect transfers operates.

For a buyer in the market as the law stands, the charge’s sharpest practical edge remains the withholding machinery through which the whole dispute began. A purchaser of shares in a foreign company whose value derives substantially from Indian assets must assume that the consideration is a sum chargeable in India in the seller’s hands, and section 195 obliges the purchaser to deduct tax from it unless a determination is obtained under the provision that permits the payer, or the payee, to apply for an order fixing the proportion of the sum that is chargeable. The discipline this imposes on deal practice is considerable: the Indian element of a global transaction has to be valued before completion, the withholding question priced into the consideration or covered by indemnity and escrow, and the seller’s treaty position established on documents rather than asserted. The Vodafone litigation is the standing demonstration of what the omission costs. The buyer in 2007 took the position, vindicated in the Supreme Court five years later, that nothing was chargeable and nothing was to be withheld. A buyer taking the same position on a comparable purchase contract after 28 May 2012 would not be vindicated at all, because the charge whose absence the Court had declared has since been enacted, with thresholds, valuation rules and reporting duties that leave the withholding question a matter of computation rather than of principle.

The treaty landscape itself was redrawn while the arbitrations ran, in the direction the episode had pointed. In 2016 India signed protocols with Mauritius, Singapore and Cyprus introducing source based taxation of capital gains on shares acquired on or after 1 April 2017, with gains on earlier acquisitions grandfathered and a transitional concession for the first two years. The renegotiations addressed, by agreement and prospectively, the structural feature that had made offshore holding routes so valuable: the residence based allocation of gains on Indian shares. Their significance for this article is comparative. Where the state proceeded by negotiation, it obtained a durable expansion of its taxing rights over future gains, accepted grandfathering for the past, and paid nothing in damages. Where it had proceeded by retrospective legislation, it obtained no durable revenue, conceded the substance in a settlement, and paid in costs and credibility. The two techniques were applied to the same underlying policy problem within the same decade, by governments of both parties, and the contrast between their yields is the closest thing the episode offers to a controlled experiment in legislative method.

The practical consequences distribute themselves by reader. For the practitioner, the episode’s residue is procedural as much as substantive: an indirect transfer analysis in India runs through the Explanation 5 charge, the Explanation 6 thresholds, the treaty allocation, and the Chapter X-A anti-avoidance screen, in that order, and each stage can terminate the inquiry. For the investor, the lesson is that the security of an Indian exit rests on treaty text, on grandfathering clauses, and on the commercial substance of the holding structure, and only residually on assurances about legislative behaviour, since the one general protection that would bind a future Parliament was never enacted. For the taxpayer without a treaty to invoke, the remedy against a retrospective demand remains the domestic one, the long appellate road through the tribunal and High Court hierarchy that hears tax appeals, measured in years. The general position described in this article is legal information drawn from the public record of the dispute; it is not advice on any individual transaction, and anyone facing a demand founded on a retrospective amendment needs advice on the specific facts, because limitation, forum and the terms of any applicable treaty will decide the outcome long before the large principles are reached.

The state’s balance sheet

An episode conducted in numbers should be closed in numbers, with the caution about composition with which this article began. The revenue at stake, as the government presented it in the 2012 debates, was about Rs 40,000 crore across the class of cases the amendment might reach. The revenue actually and finally collected on the transactions in that class was, at the end, nil: the flagship demand of Rs 11,218 crore against Vodafone, about Rs 7,900 crore of tax with the balance interest, was never paid, and the one substantial recovery, the about Rs 7,900 crore taken from Cairn against its demand of Rs 10,247 crore, was refunded under the 2021 settlement without interest. Set against that nil return the costs side of the ledger. India bore the costs of two treaty arbitrations conducted over years in foreign seats, and was directed by the Vodafone tribunal to reimburse a share of the claimant’s costs reported at 4.3 million pounds. It carried, between December 2020 and the settlement, an enforceable exposure under the Cairn award of 1.2 billion dollars with interest and costs, computed by the claimant at about 1.72 billion dollars, secured against Indian state property in at least one foreign capital. It expended, across nine years, the senior administrative and legal attention of two governments on a single amendment, including a Cabinet-approved conciliation, two changes of stated policy and a final Act of Parliament to undo the first. And it paid the unquantified item, the premium that foreign investors attach to Indian tax risk, which no ministry ledger records and which the 2021 Act was expressly introduced to reduce. The balance sheet is the shortest statement of the article’s thesis: retrospective taxation, in the only full test India has run of it, collected nothing and cost a great deal, and the law was changed in the end not because the power was doubted but because the account would not close any other way.

What the episode settled, and what it did not

The namable claim of this article can be stated in its precise form. Retrospective taxation was not abolished in India; it was withdrawn, once, for one class of transaction, before one date, on conditions set by the withdrawing state. The constitutional power to legislate backwards in fiscal matters stands exactly where it stood in 2011. No statute forbids its exercise, no judgment of the Supreme Court has condemned the 2012 amendment as unconstitutional, and the clarificatory technique by which it was effected remains in daily use across the Act. What changed is the price list attached to the power. A Parliament that reverses a decided tax case by retrospective amendment, at least where foreign investors with treaty protection are concerned, can expect arbitral liability measured in billions of dollars, enforcement proceedings against state property in foreign capitals, and a settlement whose terms include refunding what was collected without the interest that a court would have awarded. The deterrent is real, but it is a deterrent of cost and reputation, and it binds only for so long as governments continue to find the cost prohibitive.

The boundaries of the withdrawal generated their own instructive edges, and they confirm the settlement character of the instrument. A transfer undertaken in the interval between the Supreme Court’s judgment on 20 January 2012 and the assent to the Finance Act on 28 May 2012 fell within the relief, because it preceded the date the amendment became law, even though the taxpayer in that interval had the strongest possible notice that a reversal was coming: relief followed the statute’s text, not the equities of reliance. A transfer undertaken on 29 May 2012 fell outside it, however innocent, because the line had to be drawn somewhere and the state drew it at its own enactment. A taxpayer with no investment treaty behind it, whose demand belonged to the seventeen but who had never been able to carry the fight beyond the Indian courts, received exactly the relief the treaty claimants received, no more and no less, which is the one egalitarian feature of the settlement; but it received that relief only because the treaty claimants had made the class untenable, and not because its own arguments had prevailed anywhere. And a taxpayer who had paid a retrospective demand in full and closed its affairs before 2021 recovered its money on the same conditions as the resisters, surrendering any claim to interest for the years of payment, so that compliance and resistance were priced identically at the end. A repeal grounded in principle would have attended to some of these distinctions. A settlement attends to none of them, because its object is not justice between cases but the closure of all of them at a price the state can bear.

Measured against its own objectives, the 2012 amendment failed completely. It was enacted to collect tax on the transfer of Indian businesses through offshore holding companies, and on the transaction that provoked it, it collected nothing that survived. Its largest single recovery, from Cairn, was refunded. Its demands produced two arbitral awards against the state and no precedent in the state’s favour, the one constitutional question it raised never having been decided by the Supreme Court at all. Set against that record, the Supreme Court’s judgment of 20 January 2012 looks less like the obstacle the government of the day took it for and more like the cheapest available exit from a charge the Act did not contain: accept the reading, legislate prospectively with defined thresholds, as the Finance Act 2015 in due course did for the continuing regime, and collect the tax on the next transaction rather than litigating the last one for fourteen years. The road not taken in 2012 was not a concession to tax avoidance. It was the ordinary discipline of a legislature that changes the law forwards and leaves decided cases decided.

One further consequence deserves recording, because it is the form in which the episode survives in practice. The phrase “Vodafone tax” entered the vocabulary of Indian public finance as a standing warning, invoked in budget commentary, in parliamentary debate and in the advice given to every foreign investor structuring an entry into India, and it continues to be invoked whenever a retrospective amendment of any significance is proposed in any fiscal statute. That survival is not a legal restraint, and this article has been careful to insist that no legal restraint emerged from the episode. It is a reputational one, operating through the memory of officials who administered the dispute and of ministers who paid its bills. Reputational restraints decay with personnel, and a legislature ten or twenty years removed from the Cairn refund may find the warning easier to discount than the governments that received the arbitral awards did. The record assembled here is the corrective to that decay. It does not argue that retrospective taxation is impossible, unconstitutional or even always unwise. It establishes, from the state’s own documents and the tribunals’ awards, what the one full exercise of the power cost and what it collected, and it leaves the arithmetic where it belongs, in front of any future Parliament tempted to repeat the experiment.

There remains the series thesis with which this cluster of articles is concerned, and the Vodafone record states it as cleanly as any episode in the series. A judgment, reversed by statute. A statute, condemned by arbitral awards the statute could not reach. Awards, overtaken by a second statute reversing the reversal. None of that sequence is visible in the current text of section 9, where Explanation 5 sits in the neutral language of clarification, “deemed to be and shall always be deemed to have been situated in India”, giving no sign that the words were the most expensive sentence in the history of the Indian Income-tax Act. The text is the smallest part of the law on this subject. The record is the law’s meaning, and the record is a round trip: nine years, two governments, two arbitrations and one refund, to arrive where a single judgment had already placed both parties in January 2012.

Frequently Asked Questions

Q: What did the Supreme Court decide in the Vodafone case?

The Supreme Court held on 20 January 2012 that the sale by Hutchison of CGP, a Cayman Islands holding company, to Vodafone’s Netherlands company was not taxable in India. The capital asset transferred was a foreign share, situated outside India, and section 9(1)(i) of the Income-tax Act, which deems income to arise in India on the transfer of a capital asset situate in India, did not extend to the transaction. The Court held that the provision was not a look through provision, that the holding structure had commercial substance and was not a colourable device, and that the transaction was a bona fide structured foreign direct investment outside India’s territorial tax jurisdiction. Vodafone therefore had no duty to withhold tax, the demand founded on sections 195 and 201 fell, and the department was directed to return the Rs 2,500 crore deposit with interest at four per cent.

Q: What was the 2012 retrospective amendment to the tax law?

The Finance Act, 2012 amended the Income-tax Act with retrospective effect from 1 April 1962, the date the Act itself commenced. It inserted Explanations 4 and 5 into section 9(1)(i): Explanation 4 declared that the word “through” in the clause had always borne an extended meaning, and Explanation 5 declared that a share or interest in a company or entity incorporated outside India is deemed to be and has always been deemed to be situated in India if it derives, directly or indirectly, its value substantially from assets located in India. Companion amendments widened the definitions of capital asset and transfer in section 2 and the withholding obligation in section 195. A validation clause, section 119 of the Finance Act, 2012, further provided that notices and demands founded on the indirect transfer theory would be treated as valid notwithstanding the Supreme Court’s judgment, so that the amendment revived the Vodafone demand itself rather than merely changing the law for other cases.

Q: Why was retrospective taxation so controversial?

The controversy was not, at its core, about whether India may tax indirect transfers, a power many states exercise and which India has exercised prospectively since 2012 without comparable dispute. It was about the temporal direction and the target. The amendment followed within months of a Supreme Court judgment that had decided the very question against the revenue after full argument, and it was drafted as a clarification that deemed the law to have borne the opposite meaning since 1962, accompanied by a validation clause reviving the annulled demand. Taxpayers who had completed transactions on the law as declared by the highest court found liabilities, interest and penalties constructed backwards over closed deals, including a failure to withhold tax that the Court had held was never due. Critics, including the government’s own expert committee, treated the device as a reversal of a decided case under a clarificatory label. The government’s answer was that the substance of the transactions was the transfer of Indian businesses, that Parliament is constitutionally competent to legislate retrospectively in fiscal matters, and that the tax base of a developing economy should not turn on the geography of incorporation.

Q: What happened in the Vodafone and Cairn arbitrations?

Both companies invoked India’s bilateral investment treaties after the domestic dispute became incapable of resolution in their favour. Vodafone commenced arbitration in 2014 under the treaty with the Netherlands, and on 25 September 2020 a tribunal administered by the Permanent Court of Arbitration held unanimously that imposing the tax, with interest and penalties, through the retrospective amendment breached the fair and equitable treatment guarantee, directing India to cease enforcement. Cairn invoked the treaty with the United Kingdom in 2015 over a demand of Rs 10,247 crore raised on its 2006 internal reorganisation, from which the department had already recovered about Rs 7,900 crore by selling its shares and seizing dividends and refunds. On 21 December 2020 the tribunal seated at The Hague held unanimously that India had breached fair and equitable treatment and awarded damages of 1.2 billion US dollars plus interest and costs. India challenged both awards, and Cairn began enforcement proceedings against Indian state assets abroad, including the freezing of government properties in Paris, before the 2021 settlement overtook the contests.

Q: Did India withdraw the retrospective tax?

India withdrew it for a defined class only. The Taxation Laws (Amendment) Act, 2021, which received assent on 13 August 2021, provided that no demand shall be raised on the basis of the 2012 retrospective amendment for any indirect transfer of Indian assets undertaken before 28 May 2012, the date the Finance Act, 2012 received assent, and that demands already raised for such transfers would be nullified on conditions. The withdrawal was therefore real but bounded: it closed the amendment’s operation for transactions before the cut-off, while the charge on indirect transfers on or after that date remains in the Act and continues to be applied. The Act did not repeal Explanation 5, did not enact any general prohibition on retrospective tax legislation, and was presented by the government as a measure to restore investor confidence and end damaging litigation rather than as an acceptance that the 2012 amendment had been unlawful or beyond Parliament’s competence.

Q: What did the Shome Committee recommend?

The expert committee chaired by Parthasarathi Shome, constituted in 2012 to examine the retrospective amendments relating to indirect transfers, released its draft report in October 2012. It concluded that retrospective application of a tax law should occur only in exceptional, rarest of rare cases: to correct apparent mistakes or anomalies, to give effect to provisions that are genuinely clarificatory, or to protect the tax base from highly abusive schemes lacking economic substance, and not to expand the tax base. Applying that test, it held that the indirect transfer provisions of the Finance Act, 2012 were not clarificatory and tended to widen the base, and recommended that they be applied prospectively once the operative definitions were clarified. It further recommended that, if retrospectivity were retained, no person should be treated as an assessee in default under section 201 in respect of such a transfer, and that neither interest nor penalty should attach to the resulting demands. The recommendations were not acted on for nine years, although the settlement enacted in 2021 adopted their substance in almost every particular.

Q: Is retrospective taxation permitted in India?

Yes, as a matter of constitutional doctrine, within limits. Article 20(1) of the Constitution, which forbids retrospective penal liability, does not apply to a fiscal charge, and Article 265 is satisfied where the tax is imposed by a duly enacted law. The Supreme Court accepted in Rai Ramkrishna v State of Bihar in 1963 that a legislature may impose a tax with retrospective effect, subject to other constitutional limitations, and the courts have long recognised validating legislation by which a legislature retrospectively alters the legal basis of a judgment it has lost, rather than purporting to annul the judgment directly. What remains contestable in any given case is the character of the particular amendment: whether a provision described as clarificatory genuinely declares the earlier law or in substance imposes a new burden for past years, a question the courts answer by examining the provision and not merely its label. The harder constraints on retrospective taxation in practice have proved to be treaty obligations toward foreign investors and the reputational cost to the investment climate, rather than any constitutional prohibition.

Q: What is an indirect transfer of shares?

An indirect transfer is the transfer of shares or an interest in a company or entity incorporated outside India, where the value of those shares derives substantially from assets located in India. The Indian assets themselves are not sold, and no Indian company’s shares change hands on an Indian register; what changes is the ownership of the foreign holding vehicle above them. The 2007 Vodafone transaction is the standard illustration: the share transferred was in a Cayman Islands company, but that company controlled an Indian telecommunications business. Explanation 5 to section 9(1)(i), inserted in 2012, deems such a foreign share to be situated in India where it derives, directly or indirectly, its value substantially from Indian assets, so that the gain is treated as arising in India. Since the Finance Act, 2015, the statute has defined the threshold: the Indian assets must exceed Rs 10 crore in value on the specified date and represent at least fifty per cent of the value of all the assets of the foreign entity, and only the proportion of the gain corresponding to the Indian assets is charged.

Q: How much tax did the Income Tax Department demand from Vodafone, and what did the demand consist of?

The quantified demand served in October 2010 was Rs 11,218 crore, and its composition matters more than its total. About Rs 7,900 crore of that figure was tax, being the capital gains tax said to be payable on Hutchison’s gain which Vodafone should have withheld under section 195, and the remainder was interest for the period since the 2007 transaction. The demand did not stop growing. In March 2011 the department issued a notice proposing a penalty of up to one hundred per cent of the tax for the failure to deduct, which would have taken the exposure beyond Rs 20,000 crore, although the Supreme Court stayed the penalty proceedings. Reported aggregates at the arbitration stage were higher still, ranging from about Rs 22,100 crore to about Rs 27,900 crore, because each report captured a different stage of accruing interest and threatened penalty. After the Supreme Court’s judgment none of it was payable, since the underlying sum was held not to be chargeable at all; the figures revived only because the 2012 amendment and its validation clause reconstructed the demand the Court had annulled.

Q: What did the Bombay High Court decide in the Vodafone case?

The Bombay High Court decided against Vodafone at both stages at which it was seised of the dispute. In December 2008 it dismissed Vodafone’s challenge to the show cause notice, holding that the department was entitled to inquire into the transaction and that the jurisdictional objection did not bar the proceedings at the threshold. After the department’s order of May 2010 asserting jurisdiction, Vodafone returned to the Court, which on 8 September 2010 upheld the department in substance. The Court held that the transaction could not be characterised as a bare sale of one foreign share, that the bundle of rights transferred included in substance interests in the Indian operating company, and that the department had jurisdiction to tax the gain to the extent attributable to assets in India, leaving apportionment and quantification to the assessment stage. That judgment, which produced the Rs 11,218 crore demand, was the judgment the Supreme Court set aside in January 2012, holding that section 9(1)(i) did not authorise the look through approach on which the High Court’s reasoning rested.

Q: Why was Vodafone treated as an assessee in default?

Because the revenue’s claim was constructed through the withholding machinery rather than through an assessment on the seller. Section 195 of the Income-tax Act obliges a person paying a non-resident any sum chargeable under the Act to deduct tax at source, and section 201 provides that a payer who fails to deduct is deemed to be an assessee in default in respect of the tax, liable to pay it with interest and exposed to penalty. Hutchison, the seller, had received the price and exited; Vodafone, the buyer, had paid it and remained present in India with a large operating business. The department’s September 2007 show cause notice therefore alleged that the consideration was chargeable in India, that Vodafone had been bound to withhold from it, and that Vodafone stood in default for the whole tax. The theory made the collectible party answerable for another person’s income, which is why the litigation turned on chargeability itself: a payment that is not chargeable under the Act attracts no withholding duty, and the Supreme Court’s holding that the gain lay outside section 9(1)(i) destroyed the default at its root.

Q: What did companies have to give up to obtain relief under the Taxation Laws (Amendment) Act 2021?

Relief under the 2021 Act was conditional, and the conditions were the price of the settlement. A taxpayer with a demand founded on the retrospective amendment had to withdraw, or furnish an undertaking to withdraw, every pending proceeding challenging that demand, in Indian courts, in foreign courts and in treaty arbitration, including enforcement proceedings founded on a favourable award. It had to furnish a further undertaking that no claim for costs, damages or interest would be filed in any forum, and to complete the prescribed procedural formalities under the rules notified in October 2021. Only on fulfilment of those conditions was the demand nullified and any amount paid or collected refunded. The surrender was substantial in the two cases that mattered most: Cairn and Vodafone each held a unanimous arbitral award against India, an enforceable international right, and each gave up the award, the accrued interest and any claim to damages as the consideration for recovering the principal that had been demanded or collected. The structure was that of a negotiated peace, not of a restitution ordered by any court.

Q: Why is 28 May 2012 the cut-off date in the 2021 Act?

Because 28 May 2012 is the date on which the Finance Act, 2012 received the assent of the President and the retrospective amendment became law. The Taxation Laws (Amendment) Act, 2021 drew its boundary at the moment of enactment rather than at the date of the Supreme Court’s judgment or the date from which the amendment purported to operate. Transfers undertaken before the amendment existed as law, but which the amendment had reached backwards, were taken out of the charge and their demands nullified on conditions. Transfers undertaken on or after the day the amended law stood enacted were left within it, on the reasoning that a taxpayer transacting after 28 May 2012 did so with the amended statute in force and cannot claim to have relied on the earlier position. The choice of date is the settlement’s logic in miniature: it concedes nothing about the validity of retrospective legislation as such, restores the position only for those who acted before the law was changed, and preserves the indirect transfer charge, with its thresholds and reporting machinery, for every transaction that followed.

Q: How did the Cairn Energy dispute differ from the Vodafone dispute?

The two disputes arose under the same amendment but from opposite kinds of transaction. Vodafone’s case concerned an arm’s length sale between two unrelated groups, in which an Indian business changed hands through a foreign holding company, and the demand was pursued against the buyer as a withholding defaulter for tax it had never paid. Cairn’s case concerned an internal reorganisation in 2006, in which a parent company moved the shares of its Indian holding company down its own chain ahead of a stock market listing, with no outside buyer and no value leaving the group. Cairn’s demand of Rs 10,247 crore, raised in 2014, was also different in consequence: the department actually collected about Rs 7,900 crore from Cairn by selling its residual shareholding, appropriating dividends and withholding refunds, so Cairn arbitrated for the return of money already taken, and its award of 1.2 billion dollars plus interest was restitutionary in character. Vodafone, by contrast, never paid the tax; its award directed India to stop demanding it. The distinction explains why Cairn’s enforcement campaign against Indian assets abroad, and its eventual refund, dominate the closing phase of the episode.

Q: What happened to the other demands raised under the retrospective amendment?

Vodafone and Cairn were the largest of a class, not the whole of it. The statement of objects and reasons accompanying the Taxation Laws (Amendment) Act, 2021 recorded that demands had been raised in seventeen cases on the basis of the retrospective amendment, and that arbitration under India’s investment treaties had been invoked in four of them, with tribunals ruling against the government in the two that reached an award. The remaining demands covered other offshore restructurings and share transfers with underlying Indian assets, in amounts well below the two headline cases, and several were entangled in domestic litigation before the High Courts, two of them under stay. When the 2021 Act took effect, the demands in the covered class, being demands for transfers undertaken before 28 May 2012, were nullified on the same conditions as the headline cases: withdrawal of proceedings and undertakings against further claims, with amounts collected refunded without interest. Cases that did not accept the statutory terms retained whatever rights the unamended law gave them, but the class as a practical matter was closed out through the settlement machinery rather than through further adjudication.

Q: What threshold decides whether a foreign share derives its value substantially from Indian assets?

The threshold is supplied by Explanation 6 to section 9(1)(i), inserted by the Finance Act, 2015, because Explanation 5 as enacted in 2012 used the word “substantially” without defining it. Under Explanation 6, a share or interest in a company or entity incorporated outside India is taken to derive its value substantially from assets located in India only if, on the specified date, the value of the assets located in India exceeds Rs 10 crore and represents at least fifty per cent of the value of all the assets owned by that company or entity. Both limbs must be satisfied; a foreign holding company with a small Indian asset, or with a large Indian asset that is a minority of its total value, falls outside the deeming provision. The specified date is tied to the accounting period of the foreign entity preceding the transfer, asset values are taken at fair market value without deduction of liabilities, and where the threshold is met the Act charges only the proportion of the gain reasonably attributable to the Indian assets, rather than the whole gain on the foreign share.

Q: Did the Sanofi Pasteur judgment protect an indirect transfer made under a tax treaty?

In the circumstances of that case, yes. Sanofi Pasteur, a French company, had acquired the Shantha Biotechnics business by purchasing a French holding company from another French holder, and the revenue contended that the gain was taxable in India under the Act as retrospectively amended, because the French company’s value derived from the Indian operating company beneath it. The Andhra Pradesh High Court, in its 2013 judgment in Sanofi Pasteur Holding SA v Department of Revenue, held that Article 14(5) of the India and France tax treaty confined the right to tax such gains to France, and rejected the argument that the underlying Indian assets allowed India to tax notwithstanding the treaty. The department’s appeals in that line were in due course withdrawn before the Supreme Court. The durable proposition is structural: the indirect transfer charge in Explanation 5 operates subject to India’s treaty allocations, so a transferor protected by a treaty that reserves capital gains to the residence state is outside the charge unless the revenue can displace the treaty protection by establishing an abusive arrangement under the anti-avoidance rules.

Q: What is fair and equitable treatment, the standard India was held to have breached?

Fair and equitable treatment is a standard of protection written into bilateral investment treaties, by which the host state undertakes to treat a foreign investor’s investment fairly and equitably according to international law. Its content is not exhaustively defined in the treaties, and tribunals have built it from elements that include stability and predictability of the legal framework, respect for the investor’s legitimate expectations formed on the law as it stood, transparency, and freedom from arbitrary or discriminatory conduct. It does not prohibit a state from changing its tax law, and it does not prohibit retrospective legislation in the abstract. In the Vodafone award the tribunal held that imposing the tax with interest and penalties through a retrospective amendment, over a transaction the host state’s own Supreme Court had held to be outside the statute, breached the standard, and directed India to cease enforcement. In the Cairn award the tribunal likewise found a breach in the application of the retrospective amendment to Cairn’s investment and awarded damages. Both awards thus turned on the treatment of specific investors under specific facts, and neither announced a general rule against retrospective taxation.

Q: Were the refunds under the 2021 Act paid with interest?

No. The Taxation Laws (Amendment) Act, 2021 provided expressly that amounts paid or collected in the covered cases were to be refunded without any interest, and the taxpayers’ undertakings included a waiver of any claim to interest, costs or damages. The term is worth pausing on, because it cuts in one direction only. Cairn had been kept out of about Rs 7,900 crore for years, money the department had taken by selling its shares and appropriating its dividends and refunds, and it recovered the principal alone. The contrast with the domestic phase of the Vodafone dispute is sharp: when the Supreme Court allowed Vodafone’s appeal in 2012, it directed the return of the Rs 2,500 crore deposit with interest at four per cent, so that a taxpayer who won in court was compensated for the time value of money the state had held. A taxpayer who settled under the 2021 Act was not. The absence of interest was part of the price the state exacted for nullifying the demands and ending the arbitrations, and it is one reason the 2021 measure is accurately described as a settlement rather than as restitution.

Q: What did the Supreme Court decide in the Tiger Global case?

The Tiger Global litigation tested the indirect transfer charge against treaty protection and the anti-avoidance code. Tiger Global’s Mauritius entities sold shares in Flipkart’s Singapore holding company in 2018, having acquired them between 2011 and 2015, and claimed exemption under the India and Mauritius treaty, whose grandfathering clause protected gains on shares acquired before 1 April 2017. The Authority for Advance Rulings rejected their applications in 2020 as prima facie designed for the avoidance of tax. The Delhi High Court reversed that view in August 2024, holding that a valid tax residency certificate should not be disregarded on suspicion and that the grandfathering protection applied. In January 2026 the Supreme Court set the High Court’s judgment aside, allowed the revenue’s appeals and restored the Authority’s position, holding that the arrangement was an impermissible avoidance arrangement attracting the general anti-avoidance rule in Chapter X-A of the Income-tax Act, and that gains on transfers effected after 1 April 2017 are taxable in India even though the shares themselves were acquired before that date.