GST federalism is usually debated as a slogan. One side says the states of the Indian Union were reduced by the goods and services tax to spending agencies of the Centre, governments that distribute money they no longer raise. The other side answers that the states hold two-thirds of the voting weight in the body that sets the tax, a share no federation anywhere else gives its constituent units over a national levy, and that they consented to the arrangement through their own legislatures. Both slogans contain a fragment of the record, and neither survives contact with it. What the states actually did in 2016 and 2017 was sign a bargain of a specific and unusually explicit kind. They surrendered permanent taxing powers of their own, and in exchange they received protections of defined content: a compensation guarantee written into the constitutional amendment itself, a collective voting weight larger than the Union’s, a set of carve-outs that kept their most dependable revenue lines outside the shared tax, and the retention of several older levies the amendment never touched. A bargain of that kind is tested not when it is signed but when the money runs short, and the Indian bargain was tested within three years of its making, when the compensation fund could not pay what the guarantee promised and the two levels of government had to decide what a federal promise is worth in a revenue collapse. That dispute, narrated precisely, tells a reader more about the federalism of the tax than the amendment’s text does.

The verdict this article reaches, after setting out both cases at full strength, is a divided one, and the division matters. On design, the charge of hollowing out is weaker than its proponents claim: the states were not dispossessed by stealth but bought out at a price they negotiated, and the price included the only statutory revenue guarantee any Indian state has ever held against the Union. On operation, the charge is stronger than the design suggests, because the protections turned out to be of unequal durability. The voting weight cannot be used by any state alone and has almost never been used at all. The carve-outs endure, but they endure precisely because the Union depends on them too. The guarantee, the one protection that operated automatically rather than at anyone’s discretion, was for five years exactly what it promised to be, and then, at the moment of maximum stress, its financing was converted from a payment obligation into a borrowing arrangement whose debt the states carry on their own accounts while the Centre services it from a cess the states no longer control. What states gave was permanent. What they received was a mixture of the permanent and the time-limited, and the time-limited part was the part written in the largest figures.

The bargain written into the constitutional text

The constitutional machinery is best understood as three moves made simultaneously, and the federalism analysis fails whenever any one of them is described without the other two. The first move was the grant of power. Article 246A, inserted by the Constitution (One Hundred and First Amendment) Act, 2016, provides that Parliament and the legislature of every state have power to make laws with respect to the goods and services tax imposed by the Union or by that state. The formulation is without precedent in the Indian Constitution: a taxing power held concurrently and symmetrically by both levels of government over the same base, rather than divided between them by entry and List. The same article reserves to Parliament alone the power to legislate where the supply of goods or services takes place in the course of inter-state trade or commerce, so the integrated tax on cross-border supplies is a Union levy by constitutional command rather than by arrangement. Article 269A then provides that the tax on inter-state supplies shall be levied and collected by the Government of India and apportioned between the Union and the states in the manner Parliament may by law provide on the recommendations of the Council, with the amounts so apportioned forming part of neither government’s consolidated fund by default but reaching each through the apportionment itself. The second move was the surgery on the Seventh Schedule, examined below, which deleted or narrowed the older taxing entries so that the new concurrent power would operate on a cleared field; the division of taxing powers of which that surgery formed part is mapped in the series account of the Seventh Schedule and its three lists. The third move was the creation of the Goods and Services Tax Council under Article 279A, a constitutional body in which the Union finance minister sits as chairperson alongside the finance ministers of every state, charged with recommending to the Union and the states the taxes to be subsumed, the goods and services to be taxed or exempted, the model laws, the principles of levy and apportionment, the thresholds below which suppliers escape registration, the rates including floor rates with bands, and any special provision for the states of the north-east and the hills.

The breadth of the Council’s remit under clause (4) of Article 279A is itself a federalism datum that repays attention, because it defines the territory over which state discretion was pooled. The Council recommends not only rates and exemptions but the model laws on which every state statute is based, the principles that govern the place of supply and therefore the destination of revenue, the threshold below which a supplier need not register, the floor rates within which the tax must move, any special rate to be raised during a natural calamity or disaster, and special provisions for the states of Arunachal Pradesh, Assam, Manipur, Meghalaya, Mizoram, Nagaland, Sikkim, Tripura, Himachal Pradesh and Uttarakhand, together with the State of Jammu and Kashmir as the text then stood. A state legislature enacting its state tax statute is therefore legislating inside an architecture whose every load-bearing decision, the base, the rates, the thresholds, the supply rules, the disaster levy, has been framed in a body where its minister held one equal voice among the states’ bloc. The special calamity provision is the single exception in the scheme that contemplates a state-specific fiscal emergency, and it operates, characteristically, by recommendation of the Council rather than by the state’s own decision: even a flood within one state’s borders becomes, for tax purposes, a question the federation answers together, which the flood cess episode examined later in this article illustrates in practice. No previous Indian institution had been given a remit of this width over the day-to-day content of state taxation, and the width, more than any single decision taken under it, is the measure of the change of 2016.

Two features of that text carry the whole federalism argument, and both are routinely misstated. The first is that the amendment did not leave compensation to a speech or a press release. Section 18 of the amendment Act itself directed that Parliament shall, by law, on the recommendation of the Council, provide for compensation to the states for loss of revenue arising on account of implementation of the goods and services tax for a period of five years. The guarantee was therefore part of the constitutional settlement, not an executive assurance layered on top of it, and any account that treats it as a political promise that happened to be honoured or dishonoured has already misunderstood the instrument. The second feature is the text’s silence about its own enforcement. The Council recommends; the clause on disputes, clause (11) of Article 279A, directs the Council to establish a mechanism to adjudicate disagreements between the Union and one or more states, or between states, arising out of its recommendations or their implementation. For years that mechanism did not exist as a standing body, a fact whose significance becomes plain when the first great disagreement arrived and had to be settled by negotiation under fiscal duress rather than by adjudication under a rule. The amendment built a forum for consensus, a fund for compensation, and a direction to build a court for disputes, and then history tested them in exactly that order.

Why does every GST Council decision need the Union’s agreement?

Because of the weighted threshold in Article 279A. A decision needs three-fourths of the weighted votes of members present and voting; the states together hold only two-thirds, so no decision can pass against the Union’s opposition. The Union’s one-third share therefore operates as a standing veto over rates, exemptions and rules.

That answer contains the article in miniature, and the sections that follow unpack its two halves. The legislative competence question was eventually put beyond argument by the Supreme Court, and it is treated in the discussion of the Mohit Minerals judgment below. The practical question, why a legally available power remains unusable, is the deeper one, and it is the question the voting arithmetic, the credit mechanism and the compensation history each answer from a different side. Before turning to what the states received, the surrender itself has to be itemised honestly, because the debate is corrupted at source whenever the surrender is described vaguely as a loss of autonomy rather than specified as a list of entries deleted, narrowed or fenced in a schedule to the Constitution.

A design the states co-authored

One fact about the settlement is so often omitted from the autonomy debate that its omission distorts everything downstream: the design was not imposed on unwilling states by a Union drafting in private. The amendment altered the Seventh Schedule and the distribution of legislative powers, and it therefore required, under the proviso to Article 368, ratification by the legislatures of not less than one-half of the states before it could be presented for presidential assent. The states were constitutionally incapable of being mere recipients of the change; they were co-authors of it in the strict legal sense, and they ratified with a speed that surprised the amendment’s own managers, the process completing within weeks of Parliament’s passage in August 2016. Nor was the ratification the states’ first involvement in the drafting. For more than a decade before the amendment, the Empowered Committee of State Finance Ministers had functioned as the negotiating forum in which the model’s central questions, the destination principle, the rate structure, the treatment of petroleum and alcohol, the compensation demand, were argued out among the states themselves and between the states and the Union, under chairmen drawn from state governments. The constitutional text of 2016 bears the marks of that long negotiation in every clause this article has examined. The compensation direction in Section 18 was a state demand conceded by the Union. The exclusion of alcohol was a state demand conceded by the Union. The deferral of petroleum was a joint reservation. Even the discarded provisions mark state influence: the bill as introduced in 2014 contained a proposed additional tax of up to one per cent on inter-state supplies of goods, a safeguard for manufacturing states against the destination principle, and that levy was dropped from the enacted amendment because consuming states and the trading community regarded it as a breach of the common market the tax existed to create. The settlement, in short, was the states’ own bargain as much as it was the Union’s reform, which is why the later language of dispossession sits awkwardly on the record. What the states can legitimately say is not that the bargain was forced on them, but that a bargain consented to in the expectation of one revenue future was performed by the Union in the light of another, and that the instruments of performance turned out to be in the Union’s hands. Consent answers the design question. It does not answer the operation question, and the distinction between the two is the analytical spine of everything that follows.

The states had, moreover, rehearsed a shallower version of the same bargain a decade earlier, and the rehearsal illuminates what was genuinely new in 2016. When the states replaced their cascading sales taxes with a value added tax in the middle of the previous decade, they did so through the same Empowered Committee, accepting common floor rates and a common structure in return for a central compensation package that covered their losses on a declining scale for the first three years: full compensation in the first year, three-quarters in the second, half in the third. The precedent mattered twice over. It taught the states that coordination through a committee of their own finance ministers could work and that a compensation promise from the Union, in that instance, was kept and expired without a dispute that anyone remembers. And it taught the Union that consent to tax coordination is purchasable at the price of a guarantee. The 2016 settlement followed the template at a different order of magnitude: not floor rates under separate state statutes but a single statute book; not three years of tapering compensation but five years at a guaranteed growth rate; not a committee of ministers advising their own legislatures but a constitutional council whose recommendations the legislatures would enact. The states that negotiated the second bargain believed, on the evidence of the first, that the guarantee was the soundest part of it. The events of 2020 are unintelligible except against that belief: the shock in the state capitals was not the shock of governments discovering risk, but of governments discovering that a promise kept once, in smaller coin, had taught them to rely on its repetition.

The form of the power the states consented to also deserves a word, because it was chosen against alternatives that would have distributed power differently. The Constitution’s ordinary technique is division: a subject is placed in the Union List, the State List or the Concurrent List, and competence follows the placement. The amendment’s draftsmen declined that technique for the new tax. A value added tax on a national market cannot be divided by List without recreating the border frictions it exists to abolish, so they built instead a single power held twice, the simultaneous power of Article 246A, and a joint institution to coordinate its exercise. The federalism of the tax is therefore unlike the federalism of any other Indian subject. In education or public health, a state that dislikes the Union’s approach can legislate its own within its competence and defend the difference. In the shared tax, a state’s competence is real but its exercise is pre-committed to a common model by the structure of the tax itself, a difference the Supreme Court would eventually make explicit. The states did not merely pool a tax. They adopted a form of shared sovereignty that has no other instance in the Constitution, and the novelty of the form is part of why its stresses, when they came, had no settled doctrine waiting to receive them.

What the states surrendered

Before July 2017 a state government commanded a portfolio of indirect taxes that it could raise, lower, redesign or abolish through its own legislature, answerable to its own electorate for the consequences. The largest of them was the tax on the sale of goods within the state, the value added tax that most states had operated since the middle of the previous decade, together with the central sales tax on inter-state sales whose proceeds the origin state collected. Around that core sat the taxes that filled the corners of state budgets: the entry tax on goods brought into a local area, the octroi that some municipal systems still levied, the purchase tax, the luxury tax on hotels and hospitals, the entertainment tax on cinemas and shows, the tax on advertisements, and the duties on lotteries, betting and gambling. The rates on the principal tax were the state’s own choice. A state that wished to court investment could trim its value added tax; a state that needed revenue could raise it; a state that wished to protect a local industry through its tax structure had instruments for doing so. That discretion, exercised well or badly, was the substance of state fiscal autonomy in the indirect tax field, and it is what the amendment took away in return for a share in a common tax whose rates no state controls alone.

The constitutional form of the surrender deserves attention because it shows how complete it was. In the State List, the entry conferring the tax on entry of goods into a local area was omitted altogether, and the entry conferring taxes on advertisements was omitted with it. The entry on entertainment tax was amended so that it survives only where the tax is levied and collected by a Panchayat, a municipality or a local body, converting a state revenue line into a local one. The entry on the sale of goods was rewritten so that it reaches only the products the new tax does not cover: petroleum crude, high speed diesel, motor spirit, natural gas, aviation turbine fuel and alcoholic liquor for human consumption. In the Union List the same operation was performed from the other side: the excise entry was confined to those same petroleum products and to tobacco, and the entries that had carried the taxes on the sale and purchase of newspapers and the tax on services were omitted, the service tax being the Union’s own surrender, the largest single revenue instrument the Centre gave up in the same settlement. What remained to each level of government after this clearing was a common base governed jointly, a small set of reserved products governed separately, and the older non-goods taxes that the amendment never reached at all.

The scale of what was surrendered can be measured in budget terms. The taxes the states subsumed had together supplied well over half of the states’ own tax revenue in the years before the change; the value added tax alone was the largest single head in almost every state budget, the one revenue line a state government could predict, grow and, within limits, steer. What replaced them in the state accounts, the state component of the shared tax together with the state’s apportioned share of the integrated tax, is comparable in size but different in kind: it is revenue the state receives rather than revenue the state raises, and the difference registers in the behaviour of everyone concerned. A tax a government raises has a politics the government owns, since the electorate can reward the services it funds and punish the burden it imposes, and that accountability circuit, crude as it is, is the mechanism by which fiscal federalism disciplines both levels of government. A tax a government merely receives has its politics made elsewhere, in a council chamber and a settlement system, and the state government’s role in its own largest revenue line becomes that of a claimant with a formula. The compensation years disguised the change, because the guarantee made the claim nearly automatic. The years since have revealed it, as state budgets wait on monthly collection figures, settlement statements and Council decisions in a way state budgets never waited on anything when the principal tax was the state’s own.

The surrender had a second dimension that the entry-by-entry account understates, which is the loss of the rate as an instrument of policy. Under the value added tax, a state’s finance minister facing a drought, a pay commission award or an election promise could adjust the state’s principal tax and own the political cost of doing so. Under the shared tax, the rate on almost everything a state’s residents buy is set by a body in which that minister holds one voice among the states’ collective share, and a state that confronts a revenue emergency has no general indirect tax of its own left to move. Its remaining instruments are the reserved products, principally alcohol and petroleum, the taxes on land, buildings, electricity and vehicles that never entered the bargain, the profession tax capped by Article 276 at a figure fixed in nominal terms generations ago, and borrowing within limits the Union supervises under Article 293. The federalism complaint, stated precisely, is not that states lost revenue; in the first five years the guarantee saw to it that most did not. It is that states lost the means of raising revenue by their own decision, which is a different and more durable loss, and one that no compensation formula addressed because compensation replaces money, not discretion.

There is also a geographic asymmetry built into the surrender, and it explains much of the later politics. The tax that replaced the state levies is destination-based: revenue accrues where goods and services are consumed, not where they are produced. Under the older system a manufacturing state collected tax on value created within its borders and exported part of its tax burden to consumers elsewhere. Under the destination principle the producing states of the west and south, and the mining states of the east, watched revenue migrate toward the consuming states of the populous north. The compensation guarantee was the instrument that made this migration acceptable, because it insured every state against loss measured from its own past collections. When the guarantee expired, the migration remained. That is why the demand for its extension, and the alternative demand to rewrite the sharing formula between the Centre and the states, came most insistently from states such as Chhattisgarh, whose finance minister T. S. Singh Deo argued in writing to the Union finance minister that mining and manufacturing states had suffered the most under a consumption-based tax and proposed that if the protected revenue was not continued, the equal division of the tax between the Centre and the states should be recast so that the states’ component carried seventy to eighty per cent of the whole. The argument is structural, not partisan, and it will recur in every future negotiation over the tax.

What the states received

Against that surrender the states received four protections, and the article’s findable artifact sets them side by side because the argument about the tax is really an argument about how each protection has aged. The first and largest was the compensation guarantee. The Goods and Services Tax (Compensation to States) Act, 2017, one of the four central statutes through which the tax was enacted and whose family is described in the series account of the GST Acts of 2017, took the financial year 2015-16 as the base year for every state, projected the state’s revenue from the subsumed taxes forward at a growth rate of fourteen per cent a year, and undertook to pay the state, from a dedicated compensation fund, the difference between that projected revenue and what the state actually collected, for the five-year transition period that ran to the end of June 2022. Fourteen per cent was a generous figure, well above the growth the subsumed taxes had been achieving in aggregate, and it was chosen deliberately: it was the price of consent, and the consent it purchased included that of states which expected to lose from the destination principle from the first day. The fund was to be fed by a cess levied over and above the tax on a specified class of goods, pan masala, tobacco and tobacco products, coal, aerated waters and motor vehicles among them, so that the guarantee was designed to be self-financing out of consumption of the very goods the rate structure taxed most heavily.

One feature of the guarantee’s financing carries a federal consequence that passed almost unnoticed at enactment and has grown in importance since. The cess that feeds the compensation fund is a cess, and under the Constitution’s scheme a cess levied for a specific purpose is not part of the divisible pool of taxes shared with the states under Article 270. The compensation cess was benign in this respect while it flowed back to the states as compensation, since every rupee of it was destined for state treasuries by a statutory route. But the precedent embedded in the design is larger than its occasion: a growing share of the Union’s indirect tax effort conducted through cesses is a share removed from the devolution arithmetic altogether, and state governments have made the general complaint about cesses and surcharges, in Finance Commission proceedings and elsewhere, with increasing sharpness in the years since. When the compensation cess continued after 2022 purely to service debt, the anomaly completed itself: a non-shareable levy, imposed on consumption across every state, collected for a purpose from which no state any longer received compensation, under a statute none could amend alone. The guarantee’s financing mechanism thus illustrates in miniature the operational asymmetry this article keeps encountering. Instruments created jointly acquire, in a crisis, the character of the level of government that administers them.

The second protection was the voting design of the Council. Article 279A provides that every decision shall be taken by a majority of not less than three-fourths of the weighted votes of the members present and voting, that the vote of the Central Government shall carry a weightage of one-third of the total votes cast, and that the votes of all the state governments taken together shall carry two-thirds. The consequences of that arithmetic, which the owner article on the 101st Amendment and the GST Council works through in detail, are threefold. Nothing can be decided against the Union’s opposition, because the states’ two-thirds cannot reach three-fourths unaided; the Union therefore holds a veto it never has to announce. Nothing can be decided by the Union and a handful of compliant states, because the Union’s one-third requires the support of states carrying at least five-eighths of the states’ collective weight to reach the threshold. And the states, acting together with substantial unity, can block anything, because a proposal opposed by states holding more than three-eighths of their collective share cannot pass even with the Union voting for it. What no state possesses is a veto of its own: a single state’s share of the states’ two-thirds, divided equally among more than thirty governments, is a small fraction of the whole, so the protection is collective or it is nothing. The third protection was the set of carve-outs and reservations. Alcoholic liquor for human consumption was placed outside the tax altogether by the constitutional definition itself, leaving the states’ excise duty and sales tax on liquor wholly their own. The five petroleum products were placed inside the tax’s constitutional reach but made taxable only from a date the Council recommends, leaving the Union’s excise and the states’ sales tax on petrol and diesel running in the meantime. The fourth protection was negative but valuable: the amendment cleared only the indirect taxes on goods and services, and left untouched the states’ taxes on land and buildings, on mineral rights, on electricity, on vehicles, on professions within the constitutional cap, and on the carriage of goods and passengers, together with stamp duties and registration fees, so that a state retained a residual fiscal personality outside the shared tax altogether.

The ledger below states the exchange in the form a state finance official would draw it, and its third column is where the argument lives.

What the state surrendered or accepted as constrained	What the state received in exchange	What has happened to that protection since
The state value added tax and the rate discretion that went with it	A share of a single national tax on a destination basis	The discretion has not returned; rates are set collectively and no state has legislated a divergent general rate
Entry tax, octroi and purchase tax, deleted from the State List	Compensation for revenue loss measured against the state’s own past collections	The compensation period ended in June 2022 and the guarantee has not been renewed
Entertainment tax and advertisement tax as state levies	A fourteen per cent assured growth in revenue for five years	The assurance was honoured in cash terms, but in the two pandemic years it was delivered as borrowing the states carry on their accounts
The power to tax services, inter-state sales and production within the state	A two-thirds collective voting weight in the Council	The weight has been exercised by consensus in almost every decision; the single recorded vote, on lottery rates, went against the states that demanded it
Agreement that petroleum products fall within the tax’s constitutional design	Deferral of the petroleum levy until the Council recommends a date	The deferral has proved indefinite; both levels of government continue to tax petroleum outside the shared tax
Acceptance of a common statute book drafted on a common model	Retention of alcohol wholly outside the tax, with state excise and sales tax intact	This protection is intact and has grown more valuable as other revenue lines have narrowed
Acceptance that the Centre collects the integrated tax on inter-state supplies	Apportionment of the integrated tax between the Union and the states by law on the Council’s recommendation	Apportionment disputes and delayed settlements recurred, and one large accumulated balance was cleared only by a one-time transfer from the Consolidated Fund

One entry in the ledger is easily missed because it appears on both sides, and it is the strongest single answer to the dispossession narrative. The surrender was mutual in form: the Union gave up its excise duties on manufacture and its tax on services, historically its most buoyant indirect taxes, and submitted them to the same joint governance as the states’ levies. The consequence is a constitutional novelty that cuts in the states’ favour. Before 2017, a state finance minister had no voice of any kind in the Union’s indirect taxation; the Union budget set excise and service tax rates in Delhi, and the states adjusted their own taxes around the result. Within the Council, the same minister sits as a voting member over the whole of the shared tax, including the central component, and no rate on the Union’s own largest tax can be set or changed over the states’ collective objection. The states traded exclusive control of their own taxes for shared control of everyone’s, including the Centre’s. Whether that trade was wise depends on the value placed on exclusivity against reach, and honest federalists divide on it: the states’ reach into national tax policy has never been greater, and their autonomous command of a tax base has never been smaller, and both halves of that sentence are the settlement working exactly as designed. What the compensation dispute revealed is that the two halves are not held in the same hands when the system is stressed. Reach is exercised jointly and continuously. Autonomy is what a government falls back on when joint exercise fails it, and autonomy is what the settlement retired.

The ledger should also record a receipt of a different kind, accruing to the states as economies rather than as treasuries, because it formed part of what their governments believed they were buying. The old system taxed a state’s residents imperfectly and its producers oppressively at the border: goods bought by a state’s consumers from sellers outside it largely escaped the state’s own tax net, while goods its producers sold outward bore the central sales tax and the friction of check posts, entry levies and divergent state forms. A destination tax inverts the reach. Every act of consumption within a state’s borders feeds that state’s component of the tax whoever the seller is and wherever the seller sits, including the vast commerce conducted through national online platforms that the old law could not follow, and the consuming states of the Union, among them some of its poorest, acquired through the design a tax handle on the consumption of their own populations more complete than any they had administered for themselves. State governments understood this in 2016 and said so: the destination principle was defended in the state capitals that expected to gain by it as fiscal justice for states whose poverty had consisted precisely in producing little and consuming much on incomes remitted from elsewhere. The federalism of the tax therefore redistributes among the states as well as between the states and the Union, and an assessment of the bargain from the standpoint of Bihar or Uttar Pradesh differs in kind, not merely in degree, from the assessment made in Maharashtra or Chhattisgarh. The Union contains both standpoints permanently, and its fiscal constitution must be judged as a settlement among all of them, which is the final reason the argument about the tax cannot be conducted as a simple contest between a grasping centre and a united periphery. The periphery is not united, its interests are not identical, and the design’s endurance owes as much to that fact as to anything the Union has done.

Two things follow from reading the columns together. The protections a state received were of two different species, and the species matters more than the size. The carve-outs and the retained levies are structural: they exist because the constitutional text withholds the subject from the shared tax, and no Council decision and no revenue emergency can remove them without another amendment ratified by the states themselves. The guarantee and the voting weight, by contrast, are operational: the guarantee depended on a fund and on a construction of the statute when the fund ran dry, and the voting weight depends on states acting together, which is a political fact and not a legal one. The history of the tax since 2017 is largely the history of that distinction asserting itself.

A tax administered twice

The federalism of the tax is not exhausted by its legislative design, because a tax is also an administration, and the administration of the shared tax is where most citizens and most state officials actually meet it. The settlement divided the taxpayer base between the two levels of government rather than merging the two administrations into one. Under the arrangement the Council adopted, taxpayers below a turnover threshold of Rs 1.5 crore were divided between the state administrations and the central administration in the proportion of ninety to ten, and larger taxpayers were divided equally, with the object that a given taxpayer would answer to a single administration for both the central and the state components of the tax. Each administration was cross-empowered to act under the other’s statute, so that a state officer could enforce the central tax and a central officer the state tax within their allotted base. The design was defended as cooperative administration, and in one respect it genuinely is: it preserved the state tax services as institutions, with their own officers, intelligence and enforcement cultures, at a moment when a single national administration was a real alternative that the states resisted and defeated. A state that had lost its tax department would have lost something no compensation could replace, the institutional capacity to administer a tax at all, and the survival of those departments inside the new system is a federal achievement the autonomy debate rarely credits.

But dual administration carries its own federal frictions, and they have a consistent direction. A trader operating in several states deals with many state administrations but one design of law, while a trader allotted to the central administration in one state may find a counterpart allotted to the state administration in another applying the same provisions with a different temper. Assessment, audit and investigation can be initiated by whichever administration holds the taxpayer, and the other is expected to hold its hand, an expectation that depends on coordination machinery between governments rather than on any legal compulsion. The deeper point for this article’s argument is where the coordination lives: in the Council and its committees, in the common portal operated by the Goods and Services Tax Network, and in fitment and law committees of officers drawn from both levels. The working constitution of the tax is therefore bureaucratic as well as ministerial, and it runs on consensus at the officer level long before a question reaches a finance minister. States participate in that machinery as equals in form. The Union’s officers, however, serve the government that also drafts the model law, chairs the Council, controls the borrowing decisions and interprets the compensation statute, and the accumulation of those roles in one party to the federation is the administrative face of the asymmetry the compensation dispute exposed in its fiscal face.

Even the ownership of the common portal became a federal question, and its resolution is one of the settlement’s more instructive details. The Goods and Services Tax Network, the company that built and operates the technology through which returns are filed, credits matched and settlements computed, was incorporated with a majority of its shares held by private financial institutions and the two levels of government together holding the remainder, a structure defended at the time on grounds of agility and attacked, by state governments and by parliamentarians, on the ground that the data of every taxpayer in the country should not sit in a majority-private company. In 2018 the Council decided that the governments would acquire the whole of the company, with the Centre and the states together taking the private share so that the platform would be wholly government-owned. The episode is minor beside the compensation dispute, but it follows the same grammar: a component of the joint system designed at the centre of the federation was rebalanced toward joint ownership when the states pressed the federal objection, and the rebalancing, like the back-to-back borrowing two years later, was achieved by negotiation inside the Council rather than by any entitlement a state could enforce. Where the states organise and the Union calculates that concession costs less than resistance, the system bends. Where the sums are large enough to matter to the Union budget, as they were in 2020, it does not, and the difference between the two classes of case is the most reliable predictor of outcomes the institution offers.

How the destination principle moves money is worth setting out plainly, because the geographic politics of the tax follow from the mechanics and not from anyone’s intention. When a manufacturer in one state sells to a dealer in another, the integrated tax is charged on the supply and paid to the Union. The dealer claims credit for that tax against the dealer’s own liability on resale within the destination state, and the settlement process then transfers to the destination state the share of the integrated tax that the credit has consumed, while the state component on the final sale to the consumer accrues where consumption occurs. The origin state, whose factories, power, roads and industrial policy made the production possible, receives nothing from the chain except the taxes on its own residents’ consumption. Under the pre-2017 system the same origin state collected its sales tax on the first sale and the central sales tax on the inter-state movement, which is why industrial states regarded the older dispensation as partly a reward for having built industry. The destination principle is not a defect in the design; it is the design, chosen because a value added tax can only avoid cascading and border distortion if revenue follows consumption. But a principle can be correct as tax theory and redistributive as fiscal fact at the same time, and the redistribution runs from the states that industrialised early to the states whose markets the industry serves. Compensation at fourteen per cent insured the losers against the transition. Nothing insures them against the steady state, and the producing states’ argument, that a share of the settlement’s permanent gains ought to have been priced into the sharing formula rather than into a five-year bridge, is the one argument from the entire negotiation that events have strengthened rather than overtaken.

The integrated tax adds a second administrative dependency, of the states upon the Union’s settlement machinery. Because the tax on inter-state supplies is collected by the Union and then apportioned, a consuming state’s revenue depends on the accuracy and promptness of a settlement process it does not run, on returns filed in other states, and on the matching of credits across jurisdictions through the common portal. In the tax’s early years the apportionment accumulated large unsettled balances, and states complained in the Council and outside it that money collected on consumption within their borders was reaching their treasuries late or in unexplained amounts; the transfer of Rs 33,412 crore of accumulated balances into the compensation fund in 2019-20 was an admission that the plumbing had retained sums on a scale no settling system should. None of this implies bad faith. It implies something more structural and more relevant: in a destination-based tax administered through a common platform, the states’ revenue rights are mediated by systems owned and operated at the centre of the federation, and mediation is itself a form of power, quieter than a veto and exercised continuously rather than on occasions.

The compensation dispute

For the first three years the guarantee performed exactly as drafted, and the amounts measure how much of state finance it was quietly carrying. The Finance Ministry’s own accounting records compensation of about Rs 49,000 crore released for 2017-18, rising to about Rs 83,000 crore for 2018-19 and to about Rs 1.65 lakh crore for 2019-20. Even in that third year, before any pandemic, the cess collected fell well short of the requirement, and the gap was met partly from cess surpluses accumulated in the first two years and partly by transferring Rs 33,412 crore of accumulated integrated tax balances into the compensation fund. The Fifteenth Finance Commission, chaired by N. K. Singh, recorded the structural fact underneath those figures when it observed that the continuing dependence of states on compensation from the central government, twenty-one states out of twenty-nine in 2018-19, was a matter of concern, and that the shortfall of collections against the original forecasts was among the tax’s central implementation problems. A guarantee on which twenty-one of twenty-nine states depend in a normal year is not a transitional cushion; it is a load-bearing wall.

The wall was struck in 2020. The pandemic collapsed consumption, and with it the tax, in the first quarter of the financial year. The Finance Ministry’s arithmetic, placed before the Council at its forty-first meeting on 27 August 2020, estimated the compensation requirement for 2020-21 at about Rs 3 lakh crore, the cess expected to yield about Rs 65,000 crore, and the gap therefore at about Rs 2.35 lakh crore. The decisive move in that presentation was the division of the gap into two parts: about Rs 97,000 crore attributed to the implementation of the tax itself, and about Rs 1.38 lakh crore attributed to the pandemic, which the Union finance minister described at that meeting as an act of God. On the strength of an opinion of the Attorney General for India, K. K. Venugopal, the Union then took the position that the obligation under the compensation statute was an obligation to pay compensation out of the compensation fund, that the fund was fed by the cess and by nothing else, that the Union was not required to make good the shortfall from the Consolidated Fund of India, and that the proper course was for the Council to recommend arrangements under which the states would borrow. The states were offered two options: to borrow the implementation portion of about Rs 97,000 crore through a special window coordinated with the Reserve Bank, with principal and interest to be met from future cess collections and the borrowing not to be treated as the state’s own debt for the norms that govern state borrowing; or to borrow the entire Rs 2.35 lakh crore from the market, with the Union committing to repay the principal from cess proceeds but the interest burden on the pandemic portion falling differently. The Finance Ministry’s statements and the Attorney General’s opinion, on which so much turned, were published as official documents, and a reader who wishes to test the competing constructions against the primary material can open the ministry’s compensation statements in an office file viewer for spreadsheet, document and presentation files and read the two options in the form in which they were actually communicated to state governments.

What did the Attorney General advise in the compensation dispute?

The Attorney General advised that the compensation statute obliged payment out of the compensation fund and not out of the Consolidated Fund of India, that the shortfall caused by the pandemic was not a loss the Union was bound to make good from its own resources, and that the Council could recommend borrowing arrangements to bridge the gap.

The reaction of the states that were governed by parties in opposition at the Centre was immediate and was framed in constitutional language rather than fiscal language. Kerala’s finance minister, Thomas Isaac, described the Union’s reading as unconstitutional and said publicly that a state denied its compensation would have to consider its legal remedies, and finance ministers of several states spoke openly of testing the position in the Supreme Court. Punjab’s finance minister, Manpreet Singh Badal, recalled that the Council’s chairman had recorded at an early meeting that compensation was a constitutional commitment of the central government to be met in full, and said that cooperative federalism had been converted into coercive federalism. Chhattisgarh, Jharkhand, West Bengal, Maharashtra and Delhi took positions in the same sense, though not all in the same words, and for some weeks the dispute stood as a genuine constitutional standoff: the Union asserting that the fund was the measure of the obligation, the states asserting that the obligation was the measure of the fund. Behind the legal argument sat a fiscal fact neither side disputed, which is that a state told to borrow for its compensation would do so at the higher yields state development loans command, while the Union borrows more cheaply than any state, so that the allocation of the borrowing was itself a transfer of cost.

The settlement that emerged over the following months preserved the Union’s construction and abandoned its consequence. The Union itself borrowed Rs 1.1 lakh crore in 2020-21 and passed the proceeds to the states as back-to-back loans, and in 2021-22 it borrowed a further Rs 1.59 lakh crore on the same pattern, front-loaded into the early part of the year so that states had the resources when they needed them; the Finance Ministry later stated that Rs 7.35 lakh crore had been released to states in all under the compensation arrangements, a figure that includes these loans. Principal and interest on the borrowing were to be serviced from the compensation cess, and to make that possible the levy of the cess was extended beyond the compensation period itself. The accounting consequence for the states is the point on which the whole episode turns. The money arrived, and it arrived in full against the protected revenue for both years, so the guarantee was honoured in cash. But it arrived characterised as loans to the states, recorded on state accounts, in place of compensation grants; the debt that finances the guarantee sits in state balance sheets while the revenue stream that retires it is a cess levied under a central statute whose proceeds the states do not control. The Comptroller and Auditor General added a separate grievance to the record by reporting that in the first two years of the tax, cess collections of Rs 47,272 crore had been retained in the Consolidated Fund instead of being credited in full to the compensation fund the statute created, so that the fund from which the Union later said payment alone could come had itself been under-credited in the years when it was in surplus.

The October 2020 meeting of the Council exposed a further feature of the institution that the federalism analysis must record, because it concerns the integrity of consensus itself. When the Council decided that the cess would be extended beyond June 2022 to service whatever borrowing was undertaken, ministers from several opposition-ruled states complained publicly that the decision had been announced as the Council’s collective view although they had recorded their disagreement with the borrowing options and had asked for the Union to borrow instead, and that the official account of the meeting did not carry their dissent. The complaint matters beyond its occasion. In a body that decides by consensus, the record of what was agreed is itself a constitutional instrument, and a practice under which the chair’s summary becomes the decision, with disagreement absorbed into a general statement of unanimity, relocates power from the members to the secretariat that drafts the summary. The states that complained did not lack for a remedy in strict theory; the voting rule existed, and a member could have demanded its use, as Kerala’s minister had demanded it on lotteries the previous year. That they complained about the minutes rather than forcing a division tells its own story about the institution’s culture: a vote demanded and lost on compensation would have settled the question against the states with a finality that continued disagreement in the press did not, and the dissenting ministers preferred a grievance they could keep alive to a division they would certainly lose. Consensus, in such a body, protects the weaker parties precisely by leaving outcomes formally undecided, and the Union’s willingness to proceed on the chair’s summary rather than by vote was itself a judgment that an announced consensus, even a contested one, served the settlement better than a lawful victory.

The dispute did not end with the first year’s settlement, and its second act confirms the reading the first suggested. Through the autumn of 2020 the opposing finance ministers continued to press for the Union to borrow for the whole gap rather than for the implementation portion alone, and the Council’s meetings in that season recorded a federation arguing with itself in public to a degree without precedent in its fiscal history. When the second pandemic year produced the same shortfall in a milder form, the argument was not re-fought: the Council settled the 2021-22 arrangement on the pattern of the first, with the Union borrowing Rs 1.59 lakh crore and passing it through as back-to-back loans, this time decided early enough in the year for the money to be front-loaded. Precedent had done the work that adjudication had not been available to do. The construction the Union had asserted in August 2020, that the fund measures the obligation and that borrowing is the bridge when the fund falls short, had hardened by repetition into the operating constitution of the guarantee, accepted by states that continued to deny its correctness in principle because the alternative to acceptance was a revenue collapse in their own treasuries. This is how federal settlements actually change: rarely by amendment, sometimes by judgment, and most often by a practice adopted under pressure and then repeated until repetition gives it the appearance of design. A reader of the amendment’s text in isolation would find no trace of the back-to-back loan, the two-part shortfall or the extended cess. All three form part of what the settlement means, and all three were written by the stronger party at the moment of stress, subject to the weaker parties’ ability to negotiate at the margin, which they exercised with some success on quantum and timing and none on principle.

The two cases at full strength

The compensation dispute is the rare fiscal controversy in which both legal readings are honestly available, and the article’s obligation under its charter is to give each the force its proponents give it before saying what the episode does and does not settle.

The Union’s case rests on the statute’s architecture. The Compensation Act does not say that the Union shall pay compensation from general revenues; it creates a fund, credits the cess to the fund, and directs payment of compensation out of the fund. Section 18 of the amendment Act obliged Parliament to provide for compensation by law, and Parliament did so, choosing a dedicated levy as the source. On that design, the guarantee was always a guarantee that the mechanism would pay at the protected rate for five years, funded by the cess; it was not an unlimited charge on the Consolidated Fund. The Union adds three supporting points. The protected rate of fourteen per cent was itself generous beyond anything the subsumed taxes had delivered in aggregate, so the settlement cannot be described as the states having been promised merely their due and then denied it. The pandemic collapse was not a loss arising on account of implementation of the tax within any fair reading of the statutory phrase, since the same revenue would have collapsed under the old taxes too, and indeed the states’ reserved taxes on alcohol and petroleum fell in the same months without anyone suggesting the Union owed compensation for those. And when the fund proved insufficient, the Union did not leave the states to the market: it borrowed on its own credit, at rates no state could have obtained, passed the whole protected amount through, and arranged for the debt to be serviced from the cess rather than from state budgets. A state finance secretary who had to find the money in August 2020, the Union says, was made whole; what he lost was a form of words about the instrument through which wholeness arrived.

The states’ case rests on the bargain’s reliance interest. The guarantee in the amendment Act speaks of compensation for loss of revenue arising on account of implementation for a period of five years, without qualifying the loss by cause, and the Act of 2017 repeated the protected-revenue formula without an act of God exception; the distinction between implementation loss and pandemic loss appears in a finance ministry presentation in 2020, not in any text the states ratified. The states surrendered their taxing powers permanently and irreversibly, and they did so in reliance on a five-year bridge at a stated rate; a construction that makes the bridge conditional on the sufficiency of a cess whose rates the states do not control converts a guarantee into an aspiration at exactly the moment a guarantee is needed. The states add that the cess itself was part of the shared design and its inadequacy was a design failure shared by the Council, not a state failure; that the Attorney General’s construction made the Union’s obligation smaller the worse the crisis became, an inversion of insurance; and that converting compensation into loans recorded on state accounts worsened state debt metrics and borrowing headroom for a liability that was never the states’ in substance, whatever the servicing arrangement. Kerala’s Thomas Isaac put the reliance point in its sharpest form: the states had given up fiscal sovereignty in exchange for the guarantee, and a Union that could reinterpret the guarantee when it became expensive had in substance reserved a power to revise the constitutional settlement unilaterally.

What is settled by the episode can be stated shortly. It is settled that the protected revenue for the five years was in fact delivered to the states in full, by a combination of cess, transfers and Union borrowing. It is settled that no court decided the construction question, because the settlement made litigation unnecessary for the states that had threatened it, and the question therefore remains legally open: a future compensation statute drafted in these terms would present the same ambiguity. It is settled that the cess outlived the guarantee, being levied from 1 July 2022 to 31 March 2026 under the Goods and Services Tax (Period of Levy and Collection of Cess) Rules, 2022 for the single purpose of servicing the borrowing, so that consumers of the taxed goods continued to pay a compensation cess for years after compensation itself had ended. What turns on construction is the narrow legal point of whether the fund was the ceiling of the obligation or merely its first source. What no construction resolves is the political question underneath: whether a federation in which one level of government drafts the insurance, interprets the insurance and decides whether the insurance pays can describe the other level’s fiscal position as autonomous in any sense that matters. That question the article records rather than answers, because it is a question about the kind of federation India intends to be, and it is contested on values as much as on text.

The guarantee ends, and the cess outlives its purpose

The second test of the bargain came on schedule, in the middle of 2022, and its result is the clearest available evidence about where power in the design actually sits. Well before the compensation period expired, states began pressing for its extension, arguing that two of the five protected years had been consumed by a pandemic and that the revenue structure the guarantee was meant to bridge toward had not stabilised. The demand was not confined to a fringe: at the Council’s forty-seventh meeting in Chandigarh on 28 and 29 June 2022, some sixteen states spoke on compensation, and about a dozen sought an extension. Kerala’s finance minister K. N. Balagopal asked for five more years and said that almost all states were seeking an extension; Tamil Nadu’s finance minister P. Thiaga Rajan asked for at least two years; Chhattisgarh’s chief minister Bhupesh Baghel sought continuation for five years and quantified his state’s expected loss if it was refused; Punjab’s Manpreet Singh Badal had earlier sought three years; Delhi’s Manish Sisodia supported the demand in similar terms. A smaller number of states said that governments had to learn to stand on their own revenues and not depend on compensation indefinitely, a position the Union finance minister recorded with approval when she reported the meeting. No extension was granted. The Union’s position, stated by Nirmala Sitharaman after the forty-fifth meeting in Lucknow in September 2021 and maintained thereafter, was that the regime of paying compensation for revenue shortfall would end in June 2022 as the constitutional provision contemplated, and the regime ended. A dozen states, including states whose support the Union needed on other questions, asked through the very body designed to embody their collective weight, and the answer was determined by the one member whose concurrence the arithmetic makes indispensable.

What happens to the compensation cess after the compensation period?

The cess continues, but its beneficiary changed. Under rules notified in June 2022 the levy runs from 1 July 2022 to 31 March 2026, and its proceeds no longer compensate any state for any shortfall; they service the principal and interest of the borrowing raised in the two pandemic years.

The Council has since constituted a group of ministers to examine the cess’s future beyond March 2026, including whether a levy of the kind should continue in some altered form after the loans are retired, a question on which the states’ interest is direct, since a continuing cess whose proceeds entered the divisible pool or were shared by formula would be worth materially more to them than a cess retained at the centre of the system.

That sequence deserves to be read alongside the voting design rather than separately from it. The states’ two-thirds weight is real, and it has shaped outcomes where the Union was indifferent or divided, in rate decisions, in the phasing of compliance burdens and in the protection of individual states’ interests on particular commodities. What the compensation extension episode shows is the weight’s limit: on a question where the Union has a settled fiscal position of its own, the states’ collective preference, even when it approaches unanimity among those who speak, does not produce a decision, because the threshold cannot be crossed without the holder of the one-third. The Council is therefore best understood not as a legislature of the federation in which majority preferences prevail, but as a body whose decisions mark the boundary of what the Union is prepared to accept, negotiated outward from that boundary by genuine state pressure. That is a considerable institutional achievement by the standards of Indian fiscal history, in which tax changes of this magnitude were previously imposed rather than negotiated. It is also less than the word federalism is often asked to mean in the debate about the tax, and the distance between the achievement and the word is the space this article exists to measure.

The behaviour of state budgets after June 2022 supplies its own evidence about where autonomy resides. Deprived of the bridge, states reached for the handles the settlement had left them: in the budgets that followed the guarantee’s expiry, state after state raised what it still controlled, adjusting liquor duties, petroleum taxes, stamp duty schedules and electricity charges, the reserved and residual levies whose rates remain a state legislature’s own decision. The pattern is instructive in both directions. It shows that the fiscal personality of the states was diminished rather than extinguished, since a government that can raise its liquor excise by notification retains a real, if narrow and politically expensive, revenue discretion. It also shows how narrow the discretion has become, because the handles that remain are disproportionately the handles a prudent government would least wish to pull: taxes on alcohol reward the state in proportion to the consumption it elsewhere spends to discourage, and taxes on petroleum fall on the fuel of every supply chain in the state. A fiscal system that leaves its constituent units dependent for marginal revenue on liquor, fuel and property transactions, while the broad consumption base is governed jointly and the proceeds of the joint tax are shared by formula, has made a choice about where state initiative is permitted to operate, and it was not the states that made it alone.

Since the guarantee ended, the fiscal record has run in a direction that complicates both slogans. Collections from the tax strengthened markedly as the economy recovered and as compliance machinery, electronic invoicing and invoice matching tightened, and monthly collections settled at levels far above those of the tax’s first years; states that had feared a revenue cliff in July 2022 did not fall off one. Against that, the structural complaints survived the recovery intact. States still cannot vary the rate on the tax that supplies the largest single share of their own revenues. The equal division of the tax between the Centre and the states, under which half of every rupee of the shared tax accrues to the Union before any devolution of its own share begins, remains the target of the manufacturing states’ argument that the sharing formula should be recast in the states’ favour. The Fifteenth Finance Commission’s concern about dependence has been answered by the expiry of compensation rather than by the restoration of autonomy: the states are less dependent on transfers under the guarantee and no more possessed of taxing discretion than they were while it ran. Dependence and discretion, the record suggests, are different variables, and a reform can improve the first while leaving the second exactly where it found it.

Equal votes and unequal states

A further feature of the voting design deserves separate treatment because it divides the states from one another and not only the states from the Union. Within the states’ collective two-thirds, each state counts alike: the vote of a state of two hundred million people and the vote of a state of well under a million carry the same weight in the division of the bloc. The design is not an oversight. It follows the logic of a compact among governments, in which the parties are the states as constitutional units rather than their populations, and it protects the small states of the north-east and the hills, whose special position Article 279A itself recognises by directing the Council to recommend special provisions for them, from being permanently outvoted by the large states of the plains on questions, such as the taxation of commodities they produce or the exemptions their economies need, where their interests diverge from the majority’s. In that respect the Council reproduces the federal principle visible in the composition of second chambers in many federations, where the units are represented as units.

The cost of the principle is borne by the large states, and they have begun to name it. Tamil Nadu’s representatives have argued in the Council’s debates and outside them that a state’s weight in decisions about a tax should bear some relation to the state’s contribution to that tax and to the national economy, a claim that acquires force from the destination principle itself: a consuming state of great population generates a large share of the revenue and bears a large share of the political consequences of rate decisions, yet commands no more of the bloc than the smallest member. The parallel complaint runs in the opposite direction from the producing states, whose grievance is that the destination principle already transfers their revenue to the consuming states and the voting design then gives the beneficiaries of the transfer an equal say in perpetuating it. The two complaints cannot both be remedied, and that is precisely the point: the equal-vote rule is defensible as federal principle and contestable as fiscal fairness, and any move toward weighted voting by population or contribution would reconstruct the Council as something closer to a second Lok Sabha, destroying the small states’ protection and, with it, their reason to remain in a body where they could be routinely overridden. The states’ bloc is therefore equal by necessity of its own survival, and its equality is one more reason the bloc so rarely acts as one: it contains, permanently, the makings of at least three different coalitions, producing states against consuming states, large states against small, and states governed by the Union’s ruling party against states governed by its opponents, with the third division increasingly overlaying the first two. The Union, facing a bloc divided along lines it can often influence through party channels, negotiates from a strength the bare arithmetic of one-third and two-thirds understates.

Legally free and practically bound

Into this structure the Supreme Court introduced, in May 2022, the proposition on which the whole legal theory of the tax rests. Union of India v. Mohit Minerals Pvt. Ltd. was a case about a levy on ocean freight in import transactions, which the Court struck down; but the bench, in a judgment delivered by D. Y. Chandrachud J., addressed the constitutional architecture directly and held two things of permanent consequence. The recommendations of the Council are recommendations: they have persuasive value as the product of a collaborative dialogue between the Union and the states, but they are not binding on Parliament or on the state legislatures, and to treat them as binding would disrupt the fiscal federalism the amendment was designed to serve. And Article 246A confers on Parliament and on every state legislature a simultaneous power to legislate on the tax, a power neither level holds on sufferance from the other. The Court added that the Constitution does not envisage a repugnancy provision operating in this field, which is a significant omission to notice: the doctrine by which a Union law prevails over a state law in the concurrent field, the subject of Article 254 and the repugnancy rule, has no assigned role between two legislatures taxing the same base under a shared article, and the Court read that absence as deliberate, a design in which harmony is to be achieved by negotiation rather than by hierarchy. The discussion that follows is general information about a constitutional design, not advice on any live dispute, and a reader with a dispute of their own needs counsel on its facts; but as a statement of the design, Mohit Minerals is the text from which every argument starts.

Its federalism consequence is the analytical core of this article, and it is paradoxical. On the day the judgment was delivered, several state governments welcomed it as a vindication of state rights, and at the Council meeting the following month some states cited it in support of freer hands. Legally, the welcome was justified: after Mohit Minerals no one can argue that a state legislature which departs from a Council recommendation acts unconstitutionally merely by departing. Practically, the freedom is unusable in any general form, for reasons that have nothing to do with law. The tax works because a supplier charges it, a buyer claims credit for it against the buyer’s own liability, and the credit survives the movement of goods across state lines through the integrated tax and the shared return and payment infrastructure administered on a common portal. A state that set its own rate materially below the common rate would invite its tax base to be invoiced through structures it could not police; a state that set its rate above the common rate would tax its own traders out of competitiveness and break the credit chain for every buyer elsewhere who dealt with them, since the credit available downstream is the tax charged upstream and a divergent charge propagates error through thousands of returns the state does not administer. The administrative interdependence is total: a single registration architecture, a single return logic, rate notifications that must match across thirty-one statutes for the machinery to function. Divergence is therefore legally open and commercially self-defeating, which is why, years after the judgment confirmed the freedom, no state has exercised it on any matter of generality. The one experiment in state-specific levy, Kerala’s calamity cess of one per cent on intra-state supplies imposed for two years from August 2019 to fund flood reconstruction, proves the boundary precisely: it was possible because the Council itself recommended and accommodated it, it operated inside the common machinery rather than against it, and it expired on schedule.

The official response from the Union underlined how little the judgment disturbed the working arrangement. Union officials said publicly that the Court had stated the position that had always obtained, that the Council had always proceeded by recommendation and consensus, and that nothing in the tax’s operation would change; state governments that welcomed the decision discovered within weeks that this assessment was, as a prediction, correct. The June 2022 meeting at which the compensation extension was refused was conducted in the judgment’s immediate aftermath, with the states’ legal freedom freshly confirmed by the highest court, and the freedom made no difference to the outcome, because the question at issue was not within any state’s legislative gift: no state legislature could vote itself an extension of a central compensation statute. The episode is the cleanest demonstration available that the constraints on state action under the tax are located in the fiscal and administrative structure, and that constitutional litigation, however successful, does not reach them. A court can declare what the law is. It cannot give a state a tax base, a borrowing rate or a seat at a settlement table on terms other than the terms the structure supplies.

The comparison the brief for this article demands is with the dispute that never had a forum. A state convinced that the Union has broken the settlement has, after Mohit Minerals, a confirmed legislative freedom it cannot use and, under clause (11) of Article 279A, a promised adjudication mechanism that was not established. Its real remedies are the remedies of politics: argument in the Council, where the threshold gives the Union the last word; public pressure through its finance minister; and an original suit in the Supreme Court under Article 131, the route several states contemplated in 2020 and none ultimately took to judgment. A federation that gives its units a right without a remedy and a forum without a mechanism has not thereby acted in bad faith; the Council was designed as a negotiating chamber, and negotiating chambers rely on the parties needing each other. But the design choice has a consequence its authors may or may not have intended: every serious disagreement within the tax is settled by relative fiscal strength, and in any contest of fiscal strength between the Union and a state, or between the Union and the states severally rather than jointly, the outcome is known in advance. The guarantee mattered so much to the states, and its construction mattered so much in 2020, because it was the single element of the settlement that did not depend on relative strength. It operated by arithmetic. When the arithmetic produced a figure the fund could not pay, the settlement reverted to strength, and strength decided.

Petroleum, alcohol and the taxes that never entered the room

The carve-outs reward a closer look, because the distinction between the two of them is the most frequently garbled point in the entire debate, and the brief history of how the distinction came to be drawn is told in the series account of the legislative history of the GST settlement, to which the reader is referred for the negotiation itself. Alcoholic liquor for human consumption is excluded from the tax by the constitutional definition of the goods and services tax in Article 366, which means it is not that the tax has not been extended to liquor; the tax, as a constitutional category, does not reach it, and only another constitutional amendment could bring it within reach. The states’ excise duty on liquor and their sales tax on it remain wholly state levies, set by state legislatures, and they are for many states the single largest own-tax revenue line that survives the settlement, which is why no state government of any party proposes to surrender them. Petroleum crude, high speed diesel, motor spirit, natural gas and aviation turbine fuel stand in the opposite position. They are within the tax’s constitutional design; Article 279A directs the Council to recommend the date from which the tax shall be levied on them; and until that recommendation is made, the Union continues to levy its excise duty and the states their sales tax on the products, outside the credit chain, at rates each sets for itself.

Do the states still control any major tax of their own?

Yes, and the list is not residual trivia. Alcohol carries state excise and sales tax at rates each legislature fixes. Petroleum products carry state sales tax at rates that differ visibly across state borders. Stamp duties, registration fees, taxes on land and buildings, electricity duty, vehicle taxes and the profession tax all remain state subjects untouched by the settlement.

Why, then, has the petroleum recommendation never been made, when economists across the spectrum describe the exclusion of the economy’s principal fuel from a value added tax as indefensible: it cascades, because the tax embedded in diesel raises the cost of every freight movement in the country and cannot be credited by the businesses that bear it; it fragments the market the tax was built to unify; and it leaves the largest single consumption base outside the destination principle. The answer is not ignorance and not inertia. It is that petroleum taxation is, for both levels of government, the most useful revenue instrument each still possesses, precisely because it sits outside the settlement this article has been describing. The Union’s excise on petrol and diesel can be raised by notification in a budget and lowered when inflation demands it, without reference to any Council. A state’s sales tax on the same products is its one remaining lever of consequence over its own revenue, the lever a finance minister reaches for when the shared tax disappoints. Bringing petroleum into the shared tax would subject both governments’ most flexible instrument to the three-fourths threshold and the collective veto structure, would require a compensation conversation about the certain revenue consequences, and would extinguish the ability of each side to blame the other for pump prices while collecting from them. The carve-out survives, in other words, for the same reason the rest of the settlement strained: the parties to the bargain value discretion over efficiency wherever the two conflict, and petroleum is where the value of discretion is highest. A reader who understands why petroleum stays out understands the settlement better than a reader who can recite the rate schedule.

The Union’s own conduct supplies the confirming evidence, because the Centre has used its petroleum excise exactly as the theory predicts a government uses an instrument of discretion: as a macroeconomic lever answerable to no one. When international prices drove domestic inflation in late 2021 and again in the spring of 2022, the Union cut its excise duty on petrol and diesel by substantial amounts in single announcements, absorbing a revenue cost it calculated for itself, and urged the states to cut their sales taxes in parallel, an exhortation several states publicly resisted on the ground that their petroleum tax was the revenue line standing between their budgets and the expiry of compensation in the same season. The exchange captures the whole petroleum question in one scene. The Union treated its petroleum duty as an instrument of national price policy because that is what an uncontrolled tax is for. The states refused the parallel sacrifice because their petroleum tax had become, by 2022, an instrument of fiscal survival. Neither position is available to either government inside the shared tax, where a rate change on a comparable base would require the Council, the threshold and the negotiation. Petroleum will enter the tax when, and only when, both levels of government are offered protection for the discretion they would surrender that is more credible than the protection whose history this article has traced, and the drafting of that protection, on the evidence of the compensation dispute, is a harder problem than the economics of the inclusion, about which there is scarcely any disagreement left to record.

How the Council actually decides

Between the constitutional text and the recorded consensus stands a machinery of preparation that determines what the Council is ever asked to decide, and a federalism analysis that ignores it mistakes the institution. Recommendations reach the Council through committees of officers: a fitment committee examines rate proposals and their revenue consequences, a law committee examines drafting, and groups of ministers are constituted for the questions too political for officers, of which the long-running work on rationalising the rate structure is the standing example. By the time a proposal reaches the ministers, its revenue arithmetic has been modelled, its losers identified and, in the ordinary course, its text negotiated to the point where opposition would cost more than acquiescence. This is the practical reason the voting rule sleeps. Decisions emerge from a pipeline designed to manufacture agreement, and the pipeline is staffed by the two levels of government together, with the Union’s revenue department providing the secretariat, the modelling capacity and the chair. States with strong tax administrations of their own can contest the pipeline’s assumptions; states whose administrations were thinned by the loss of their own taxes increasingly cannot, and must argue from the Union’s numbers. The asymmetry is subtle and cumulative, and it is the form in which the surrender of 2017 continues to operate long after the surrender itself has ceased to be news.

The consensus the pipeline produces is nevertheless real negotiation, and the record contains decisions that bear the states’ fingerprints in ways no account of Union domination explains. Exemptions dear to particular states have survived repeated rationalisation drives. The compliance calendar has been bent, repeatedly, to the administrative capacity of the smaller states. Rate increases on items of mass consumption have been deferred for years against the advice of reform committees because state ministers would not carry them home. A council that merely ratified the Union’s preferences would not produce that pattern; it would produce a tidier rate structure and a shorter list of exemptions than the tax possesses. The accurate description is neither domination nor parity but something for which the vocabulary of federalism is short of terms: a joint government of the tax, in which the Union holds the chair, the secretariat, the veto and the purse of last resort, and the states hold the numbers, local knowledge and the power to make implementation succeed or fail, and in which outcomes are the product of those holdings traded against each other continuously. Joint government explains both why the states’ complaints are sincere and why their secession threats are empty. A partner in a joint enterprise who is ill-used can complain, obstruct and bargain harder next time. What the partner cannot do is leave, because the enterprise is the market itself.

The mechanisms that were never used

Two features of the design have been described in passing, and they deserve to stand together because they measure the same distance between the paper constitution of the tax and its working constitution. The first is the voting rule. In the entire history of the Council, decisions have been taken by consensus in every instance but one: at the thirty-eighth meeting on 18 December 2019, on the insistence of Kerala’s finance minister Thomas Isaac, the Council voted on whether lottery should bear a single rate, and the proposal for a uniform rate of twenty-eight per cent carried, with twenty-one members in favour, seven against and three abstaining. The episode is remembered because it is unique, and its uniqueness is the datum. A body equipped with an elaborate weighted voting rule has behaved, in substance, as a conference that acts when agreement exists and defers when it does not, which means that the operative rule of the institution is not three-fourths of weighted votes but the prior question of whether the Union and a sufficient body of states can be brought to the same text. Voting exists in the design as a reserve power whose function is to discipline negotiation: the Union’s certain veto shapes what is proposed, the states’ possible blocking coalition shapes what the Union proposes, and the vote itself, like many reserve powers, works by never being taken.

A technical feature of the voting rule sharpens the point and is almost never reported. Decisions are taken by three-fourths of the weighted votes of the members present and voting, and the quorum is one-half of the total membership. The threshold therefore floats with attendance and with the decision to vote rather than abstain: a proposal can pass on a thin attendance with fewer governments concurring than the headline arithmetic suggests, and an abstention, of which the lottery division recorded three, lowers the weight a proposal must gather without adding to its opponents. In a body that decides by consensus this machinery is dormant, but its existence disciplines the consensus itself, since every delegation knows the division that would follow a breakdown and can price its concessions accordingly. The states’ protection, once again, proves on inspection to be a protection of the states present, organised and voting together, which is a description of a political achievement to be renewed at every meeting and not of a legal entitlement that renews itself.

The second unused mechanism is the dispute provision. Clause (11) of Article 279A directs the Council to establish a mechanism to adjudicate disputes between the Union and states, or between states, arising out of the Council’s recommendations or their implementation. For years after the tax began, no such standing mechanism was operationalised; proposals to constitute one surfaced periodically in the Council’s deliberations and in commentary, and the demand for a permanent mechanism became a standard item in the states’ reform list, pressed by the same finance ministers who had fought the compensation battle. The consequence of its absence was visible in 2020: the largest fiscal dispute between the Union and the states since the Constitution’s adoption was resolved without any adjudicative stage, by an exchange of options and acceptances, because there was no body competent to be asked the question the dispute actually posed, which was a question of statutory construction fit for a tribunal or a court. Whether a standing mechanism would have changed the outcome is unknowable; the Union’s fiscal position would have weighed with any adjudicator as a matter of remedy. What can be said is that the federation wrote itself a court and then conducted its greatest tax dispute without one, and that the omission is the clearest single piece of unfinished business the amendment left behind.

The states’ own hesitation before the judicial route deserves a more sympathetic analysis than it usually receives, because it was not simple timidity. An original suit under Article 131, or a challenge to the compensation construction by any other route, risked the one outcome worse for the states than the settlement they obtained: an authoritative judgment that the Attorney General’s construction was correct. A negotiated grievance can be re-argued in every Council meeting and every budget season, and the states did re-argue it, converting their legal complaint into the political capital that produced the Union’s borrowing, the front-loading of the second year’s loans and the extension of the cess to service them. A lost judgment would have closed the question, validated the fund-ceiling construction for all future guarantees, and left the states with a precedent where they previously had an argument. Litigation in a federation is a weapon whose discharge settles the law for the loser as much as for the winner, and parties who expect to need the same ambiguity again rationally preserve it. The consequence, however, is structural and worth naming: the construction of the compensation guarantee remains untested in any court, available to be asserted by a future Union in a future crisis in precisely the form in which it was asserted in 2020, because the states judged, correctly by their own lights, that an untested ambiguity was worth more than a tested defeat. Ambiguity preserved is not autonomy secured. It is autonomy postponed, and the distinction is the same one that runs through every institution this article has examined.

How other federations divide a consumption tax

The Indian design is sometimes described as though pooling a consumption tax necessarily produces the Indian distribution of power. The comparative record says otherwise, and a brief survey disciplines the argument by showing which features of the Indian settlement were choices rather than necessities. Canada operates a federal goods and services tax alongside provincial taxes in arrangements that vary by province: some provinces have harmonised their sales tax with the federal tax under agreements that preserve a provincial rate component, while Quebec administers its own value added tax through its own revenue authority in parallel with the federal administration, the standing proof that a dual administration within one economic space can leave a constituent unit with genuine rate and administrative autonomy where the political will to sustain the duplication exists. Australia took the opposite course on administration and the stricter course on consent: its goods and services tax is levied by the Commonwealth alone and the whole of its net revenue is distributed to the states, but the intergovernmental agreement on which the system rests provides that the rate and the base cannot be changed without the unanimous agreement of the states and territories, a rule that converts every state, however small, into a veto-holder over the tax’s fundamental parameters. Germany divides the proceeds of its value added tax between the federation and the Länder by federal legislation that itself requires the consent of the Bundesrat, the chamber in which the state governments sit, so that the sharing formula is entrenched behind the states’ collective legislative concurrence.

Set against those three, the Indian design’s distinctive choices come into focus. India chose joint decision-making in a council over divided administration on the Quebec model, and joint decision-making by qualified majority over the Australian unanimity rule, and a sharing and compensation structure enacted by ordinary law over a formula entrenched behind a states’ chamber on the German model. Each choice was defensible, and the combination was probably the only one available to a reform that had to be carried through a Parliament in which the Union government of the day lacked a majority in the upper House and needed state ratification besides. But the combination has a cumulative character that its parts do not announce separately: at every point where a constituent unit of another federation holds a legal trump, a rate it controls, a unanimity that protects it, a chamber whose consent entrenches the formula, the Indian state holds instead a negotiating position. Negotiating positions are worth a great deal while the negotiation continues, and the negotiation over the shared tax never formally ends, since rates and rules require the Council’s continuing life. They are worth less at the moments that matter most, when the negotiation is about the negotiators’ own entitlements, because then the party that can wait, borrow more cheaply and administer the machinery sets the terms on which agreement is reached. The comparison does not convict the Indian design of failure; the Canadian arrangement is the deposit of a century of provincial power that India never had, and the Australian unanimity rule has frozen that country’s tax base against reforms its own treasuries desire. It does locate the Indian settlement accurately on the spectrum of federal fiscal designs: nearer to the centralised end than its two-thirds voting weight suggests, and held there not by the arithmetic of the Council but by the absence, everywhere in the design, of a single protection a state can operate alone.

The reform agenda the dispute produced

Disputes of this magnitude usually leave an institutional residue, and the compensation dispute has left a reform agenda that functions as a map of the settlement’s acknowledged weak points. Its first item is the dispute mechanism itself: the demand that the Council finally establish the adjudicative mechanism clause (11) of Article 279A contemplates, pressed by state finance ministers and by commentators across the argument, on the ground that a federation which cannot adjudicate its tax disputes will settle them by strength and call the settlement consensus. Its second item is the redesign of protection for any future transition: where a guarantee is again required to purchase state consent, whether for the inclusion of petroleum or for a restructuring of rates, the states’ experience argues for a guarantee financed from general revenues or from a fund with a statutory charge, expressed to be payable regardless of the state of any cess, since a guarantee routed through a dedicated levy has been shown to be a guarantee of the levy’s yield and not of the beneficiary’s revenue. Its third item is the sharing formula, where Chhattisgarh’s proposal to recast the equal division in the states’ favour stands as the opening position of the producing states for the next negotiation. Its fourth is the overlap between the Council and the Finance Commission, the two constitutional bodies through which Indian fiscal federalism is conducted: the Commission, appointed under Article 280, must take the tax system the Council produces as the given frame within which devolution is calculated, while the Council sets rates and structures with consequences for the vertical balance that the Commission exists to correct, and neither body is charged with the whole picture. The Fifteenth Finance Commission’s observation about compensation dependence was, in substance, one constitutional body remarking on the consequences of another’s design, and the absence of any forum in which the whole of fiscal federal relations is reviewed at once is a gap the reform debate has identified without filling.

What unifies the agenda is that every item on it seeks to convert a negotiating position into a legal one: a mechanism that must decide, a guarantee that must pay, a formula that cannot be revised by the party it constrains, a review that someone is obliged to conduct. The states learned from the dispute not that the Union acted in bad faith, the record does not sustain that charge in its simple form, but that the value of a federal protection is a function of its independence from the discretion of the party against whom it protects. That lesson, rather than any particular grievance about 2020, is the dispute’s durable contribution to Indian constitutional practice, and it will shape the drafting of every subsequent fiscal bargain, including the bargain over petroleum, whenever it comes.

Verdict on GST federalism: a bargain kept in cash and revised in character

The sharpest form of the charge, that the states have been reduced to spending agencies, fails on the evidence of the states’ own behaviour, and it is worth saying plainly why before the qualifications are entered. A spending agency does not negotiate the instrument by which it is funded, yet the states negotiated this one for a decade and ratified it through their legislatures. A spending agency does not sit in the body that sets its funder’s taxes, yet state ministers vote on the central component of the largest tax in the country. A spending agency does not retain, outside the arrangement, taxes on alcohol, petroleum timing, land, buildings, electricity, vehicles and professions, or raise those taxes when its circumstances demand, as state after state did when the guarantee expired. And a spending agency is not courted, as the states are courted in every Council season, by a Union that needs their votes, their administrative machinery and their political ownership of a tax whose failure would be laid at the Union’s own door. The states of the Indian Union are constrained partners in a joint fiscal enterprise whose terms of trade have shifted against them in operation. Constrained partnership is a serious condition, and this article has documented its mechanisms in detail. It is not agency, and the vocabulary of agency should be retired from the debate, because it prevents the accurate complaint from being heard: not that the states were made instruments, but that the instruments they were promised in exchange for partnership proved, under load, to be weaker than the partnership deed suggested.

The claim that the tax hollowed out state fiscal autonomy can be given the answer it deserves, which is that it is true of one layer of autonomy, false of another, and unanswerable as a slogan. It is false as a claim about 2016: the states were not stripped but bought, and the price, a guaranteed fourteen per cent growth written into the amendment and the statute, a two-thirds collective weight in the deciding body, and the retention of alcohol, petroleum timing and the older non-goods taxes, was the most favourable set of terms any group of Indian states has extracted in a fiscal negotiation with the Union. It is true as a claim about discretion: the states permanently lost the power to vary the principal indirect tax on their own economies, and nothing they received restored an equivalent power, so that a state government’s fiscal personality expresses itself at the margins of a shared tax, in the reserved products and in expenditure choices, rather than in the design of its own revenue. It is most illuminating as a claim about guarantees. The compensation episode established, for any state contemplating a future federal bargain, the going rate of a Union guarantee: it will be honoured in amount, it will be construed by the guarantor when the amount becomes difficult, its financing may be converted into borrowing on the beneficiary’s account, and its extension will be refused at expiry however many of the beneficiaries ask. None of that makes the original bargain a mistake; the states that signed it had no better offer, and the unified market it purchased is real. But it does mean that the constitutional text of 2016 is an incomplete guide to the federalism of the tax, and that the complete guide is the text plus the dispute. The Council and the guarantee were the two institutions the amendment created. The record shows the Council to be a negotiating chamber in which the Union’s concurrence is the operative rule, and the guarantee to be a promise whose cash was good and whose construction was contested. That is the state of Indian fiscal federalism under the tax, and it is knowable only from the dispute, not from the amendment.

The series to which this article belongs argues that a statute is the record around its text, and the fiscal constitution of the shared tax is the argument’s strongest exhibit. The amendment’s text created two institutions, a council and a guarantee, and a reader who stopped at the text would describe a federation that had solved the problem of joint taxation by joint decision and insured the solution at a generous rate. The record around the text shows the council behaving as a negotiating chamber whose operative rule is the Union’s concurrence, the guarantee performing flawlessly in fair weather and being reconstructed under load by the party that drafted it, the adjudication mechanism unbuilt, the exit impassable, and the states adapting with intelligence inside a structure whose fixed points they no longer set. Neither the text nor the record alone is the truth of the arrangement. The truth is their superposition, and it is the superposition, not the amendment and not the dispute in isolation, that a legislator amending the system, a finance official negotiating within it, or an aspirant examined on it must learn to hold in view at once.

For the three readers this article most concerns, the payload is distinct. The state finance official inherits the dispute as precedent: any future protection offered in a federal negotiation should be drafted against the 2020 construction, financed independently of a dedicated levy where the protection is meant to be absolute, and accompanied by the adjudication mechanism the amendment promised, because a protection without a forum is a protection at the guarantor’s pleasure. The researcher inherits a distinction the literature on the tax has been slow to formalise, between the design bargain and its operation: the design distributes legal competence generously and symmetrically, the operation distributes effective power according to fiscal strength, administrative control and the capacity to wait, and an analysis that reads only the competence will call the system federal while an analysis that reads only the operation will call it centralised, each mistaking its layer for the whole. The aspirant, for whom this dimension of the tax is heavily examined, should hold the three anchors that survive every framing: the simultaneous power of Article 246A, the weighted threshold of Article 279A under which the Union’s concurrence is indispensable and the states’ weight is collective or nothing, and the compensation sequence from guarantee through dispute to borrowing, which is the examined case study because it is the real one. Arguments about the tax that cannot be attached to one of those anchors are rhetoric. Arguments that can, are the federalism of India as it is practised, and the dispute over compensation is where that practice is documented most completely for anyone willing to read it.

Frequently Asked Questions

Q: Did states lose taxing power under GST?

Yes, in the specific sense that matters most: they lost the power to decide. The states surrendered their value added tax, entry tax, purchase tax, luxury tax, entertainment tax and advertisement tax, and with them the ability of any state legislature to set the rate on the principal tax paid by its own residents. In exchange they received a share of a common tax governed jointly, a two-thirds collective weight in the Council, and a five-year compensation guarantee. The power to tax goods and services was not extinguished; Article 246A gives it to state legislatures as fully as to Parliament. What was extinguished was unilateral discretion over the base that had supplied well over half of the states’ own tax revenue.

Q: What was the GST compensation guarantee and what happened to it?

The guarantee was written into Section 18 of the 101st Amendment itself and enacted through the Goods and Services Tax (Compensation to States) Act, 2017. Each state’s revenue from the subsumed taxes in 2015-16 was projected forward at fourteen per cent growth a year, and the Union undertook to pay the difference between that protected revenue and the state’s actual collections for five years, from a fund fed by a cess on specified goods such as tobacco, pan masala, coal, aerated waters and motor vehicles. It performed in full for three years. In the two pandemic years the fund fell far short, the Union declined to pay from general revenues on the Attorney General’s advice, and the protected amounts were delivered instead through Union borrowing passed to states as back-to-back loans, serviced by extending the cess to March 2026. The guarantee expired in June 2022 and was not renewed, despite demands from about a dozen states.

Q: Can a state levy its own tax on goods after GST?

On goods inside the shared tax, only within the common design: a state legislature retains the legal competence under Article 246A, confirmed by the Supreme Court in Mohit Minerals, but a divergent state levy would break the input tax credit chain and has never been attempted on any general matter. On goods the settlement reserved, the state’s power is complete and routinely exercised. Alcoholic liquor for human consumption sits outside the constitutional definition of the tax, so states levy their excise duty and sales tax on it freely. Petroleum crude, diesel, petrol, natural gas and aviation turbine fuel are inside the design but not yet taxed under it, so states levy their sales tax on them at rates they choose, which differ sharply across state borders.

Q: Why did states borrow to cover the GST compensation shortfall?

Strictly, the states did not borrow at first asking; they were offered borrowing, refused its terms, and in the end received Union borrowing instead. The 2020 arithmetic was a compensation requirement of about Rs 3 lakh crore against expected cess of about Rs 65,000 crore. The Union, advised that the compensation statute obliged payment only from the compensation fund, offered states two borrowing options rather than payment from the Consolidated Fund. After months of dispute, the Union borrowed Rs 1.1 lakh crore in 2020-21 and Rs 1.59 lakh crore in 2021-22 itself and passed the proceeds to states as back-to-back loans against their protected revenue. The loans appear on state accounts, while principal and interest are serviced from the compensation cess, whose levy was extended beyond the compensation period for that purpose.

Q: Is the GST Council a threat to fiscal federalism?

The answer depends on which federalism is meant. As a matter of legal form, the Council protects states better than any predecessor institution: its recommendations bind no legislature, the states hold two-thirds of the weighted vote, and the Union cannot carry a decision without substantial state support. As a matter of operation, the Council concentrates the initiative in the Union, which chairs it, houses its secretariat, supplies its revenue modelling and must concur in every outcome, while the states’ weight is usable only collectively. The compensation dispute showed the operational truth prevailing over the formal one. The Council is best described neither as a threat nor as a safeguard but as a negotiating chamber whose rules systematically advantage the party with the deeper pocket and the longer patience.

Q: What happens if a state disagrees with a GST Council decision?

Formally, very little need happen. Council decisions are recommendations, and after the Supreme Court’s judgment in Mohit Minerals a state legislature is legally free to legislate differently, because Article 246A gives it simultaneous power. Politically, the disagreeing state’s options are to argue in the Council, where it needs the support of states holding a large share of the bloc to block anything; to press its case publicly through its finance minister; or to litigate, including by an original suit in the Supreme Court under Article 131. Practically, unilateral divergence on a rate or a rule breaks the credit chain for its own traders and disrupts the shared administrative machinery, so disagreement is almost always absorbed into negotiation. The one recorded vote, on lottery rates in December 2019, shows both that dissent can be formalised and how rarely any state chooses to formalise it.

Q: Which taxes remain outside GST?

Alcoholic liquor for human consumption is outside altogether by the Constitution’s own definition, carrying state excise duty and state sales tax. Five petroleum products, crude oil, high speed diesel, motor spirit, natural gas and aviation turbine fuel, are inside the constitutional design but await a Council-recommended date, and meanwhile carry Union excise duty and state sales tax. Beyond those, the settlement never touched the older state levies on subjects outside goods and services taxation: taxes on land and buildings, on mineral rights, on electricity, on vehicles, on professions subject to the constitutional ceiling in Article 276, and on the carriage of goods and passengers, together with stamp duties and registration fees. Customs duty also remains a Union levy outside the shared tax, apart from the integrated tax charged on imports.

Q: Has GST helped or hurt state revenues?

Both, in sequence, and the division matters more than the net figure. During the five compensation years, state revenues were protected at a fourteen per cent growth path that most states would not have achieved from their own subsumed taxes, so the tax, with its guarantee, plainly helped. After June 2022 the cushion ended, and states became dependent on the buoyancy of a tax whose rates they do not control; collections strengthened considerably with recovery and tighter compliance, so the feared revenue cliff did not arrive, but the structural position had changed underneath the good numbers. Producing states continue to lose relative ground to consuming states under the destination principle. The honest summary is that the tax helped state revenues while the guarantee ran and left state revenue autonomy permanently narrower when it stopped.

Q: Why did the GST compensation fund run out of money?

Because the guarantee was indexed to a promise the cess was never sized to keep in a downturn. Protected revenue grew at a guaranteed fourteen per cent a year for every state regardless of economic conditions, while the cess was levied on a narrow class of goods, tobacco, pan masala, coal, aerated waters and automobiles, whose consumption could not expand at anything like that rate. Even in 2019-20, before the pandemic, compensation of about Rs 1.65 lakh crore had to be met from cess collections of roughly Rs 95,500 crore by drawing on earlier surpluses and transferring accumulated integrated tax balances. In 2020-21 the requirement rose to about Rs 3 lakh crore while cess receipts collapsed with consumption to an expected Rs 65,000 crore. The Comptroller and Auditor General also reported that cess collections of Rs 47,272 crore in the first two years had been retained in the Consolidated Fund rather than credited fully to the fund.

Q: What changed for state finances when the compensation period ended in June 2022?

The protected revenue floor disappeared. From July 2022 each state has lived on its actual collections from the shared tax: its own state component and its apportioned share of the integrated tax, supplemented by whatever the Union’s devolution and grants provide. Budgets that had been planned against a guaranteed fourteen per cent growth path had to be replanned against collection figures that arrive monthly and vary with the economy. States responded by reaching for the levies they still control, raising or restructuring liquor duties, petroleum taxes, stamp duties and electricity charges in the budgets that followed. The compensation cess itself did not end with compensation; it continued, and continues to March 2026, but its proceeds service the pandemic borrowing rather than any state’s shortfall, so states lost the protection while the levy raised in its name remained.

Q: Why do manufacturing states say a destination-based tax works against them?

Because revenue follows the consumer, not the producer. Under the pre-2017 system an industrial state taxed production within its borders and collected the central sales tax on goods leaving the state, so its factories generated revenue for its treasury wherever the goods were finally sold. Under the destination principle the tax on the same goods accrues to the state where the buyer consumes them, and the producing state receives nothing from the chain beyond tax on its own residents’ consumption. States such as Maharashtra, Gujarat, Tamil Nadu and Chhattisgarh, which industrialised or mined early, regard themselves as permanently transferring revenue to populous consuming states. Chhattisgarh’s finance minister T. S. Singh Deo argued the point formally, proposing that if protection ended, the equal division of the tax between Centre and states should be recast with seventy to eighty per cent going to the states.

Q: If a state needs more revenue quickly, which taxes can it still raise on its own?

The levers the settlement left are real but narrow. A state legislature can raise its excise duty and sales tax on alcoholic liquor, its sales tax on petrol and diesel and the other deferred petroleum products, its stamp duty and registration fees on property transactions, its tax on land and buildings, its electricity duty and its vehicle taxes, all without any other government’s consent. What it cannot do is touch the rate on the shared tax that supplies the largest part of its own revenues, or levy a new tax on goods and services generally. The remaining levers are therefore concentrated on alcohol, fuel and property, which is why state budgets after the compensation period ended leaned visibly harder on exactly those heads, and why the fiscal incentives built into the settlement reward states for consumption they might otherwise discourage.

Q: Did the compensation borrowing count against the states’ own debt limits?

In substance the borrowing was arranged so that its servicing did not fall on state budgets, since principal and interest are paid from the compensation cess, whose levy was extended to March 2026 for the purpose. The Union’s original first option had also promised that borrowing through the special window would not be treated as a state’s debt for the norms governing state borrowing. In form, however, the Rs 1.1 lakh crore of 2020-21 and the Rs 1.59 lakh crore of 2021-22 were passed to states as back-to-back loans and are recorded in state accounts, which is why state governments objected to the characterisation even while accepting the money: a liability entered in a state’s books affects how that state’s finances are read by lenders, rating analysts and finance commissions, whatever the servicing arrangement behind it. The episode is a compact example of the difference between the cash of a federal promise and its accounting.

Q: Why do large and small states carry equal weight inside the states’ two-thirds share?

Because the Council is a compact among governments, not a parliament of citizens. Article 279A divides the vote between the Union as one party and the state governments as the other, and within the states’ bloc each government counts alike, so a state of two hundred million people and a state of under a million divide the two-thirds equally between them. The design protects small states, particularly the north-eastern and hill states for which the article separately directs special provisions, from being overridden on matters where their interests diverge from the large states. Large states have begun to contest the fairness of the arrangement, arguing that weight should reflect contribution to the tax and the size of the economy, but any move to population or revenue weighting would dissolve the protection that makes the bloc acceptable to its smaller members, which is why the equality has survived every reform discussion.

Q: Why has only one GST Council decision ever been taken by a vote?

Because the institution was built, and has behaved, as a consensus body. Decisions are prepared through committees of officers and groups of ministers precisely so that a text capable of general agreement reaches the ministers, and the chair’s practice has been to announce the sense of the meeting rather than divide it. The single exception proves the rule’s strength: at the thirty-eighth meeting on 18 December 2019, Kerala’s finance minister Thomas Isaac insisted on a vote on whether lotteries should bear a single rate, and the uniform rate of twenty-eight per cent carried with twenty-one members in favour, seven against and three abstaining. Members have generally preferred a deferred decision or an announced consensus to a recorded defeat, since a vote lost on the record settles a question that continued negotiation leaves alive, and the Union, whose concurrence the arithmetic requires in any event, has no need to force divisions it can win by patience.

Q: Could a state withdraw from GST altogether?

As a matter of strict law, the path is blocked at the constitutional level: the shared tax rests on the 101st Amendment, and a state cannot repeal the amendment for itself or reclaim the deleted Seventh Schedule entries by its own legislation. It could, after Mohit Minerals, legislate divergently within Article 246A, but that is non-cooperation from inside, not withdrawal, and its costs fall on the state’s own traders through the broken credit chain. As a matter of practice, withdrawal is not a serious possibility at all, because the state’s businesses would lose seamless credit on inter-state trade, its consumers’ purchases elsewhere would still bear the tax, and its administration would have to rebuild a tax apparatus it dismantled in 2017. The honest answer is the one the compensation dispute taught: exit from the settlement is legally obstructed and commercially self-defeating, so a dissatisfied state’s real instruments are voice, negotiation and litigation, never departure.

Q: What role does the Finance Commission have when the Centre and the states dispute GST revenues?

None that is adjudicative, and that absence is part of the design problem. The Finance Commission, appointed under Article 280 every five years, recommends how the divisible pool of Union taxes should be shared and what grants should flow; it does not decide disputes, and it must take the tax structure the Council produces as given when it calculates what states need and receive. Its role in the GST story has been observational and, at moments, pointed: the Fifteenth Finance Commission recorded that the continuing dependence of states on compensation, twenty-one of twenty-nine states in 2018-19, was a matter of concern, and identified the shortfall of collections against forecast among the tax’s central problems. The federation thus has one constitutional body that designs the tax and another that measures its fiscal consequences, with no forum charged with reconciling the two views.

Q: Why did states demand that compensation continue after June 2022?

On three grounds, each stated publicly by the ministers who pressed the demand. First, two of the five protected years had been consumed by a pandemic, so the transition the guarantee was meant to finance had not taken place on the terms assumed. Second, collections had not stabilised at the protected path for many states, and finance ministers such as Kerala’s K. N. Balagopal, who sought five more years, and Tamil Nadu’s P. Thiaga Rajan, who sought at least two, argued that an abrupt end would force cuts in welfare and capital spending. Third, producing states argued that the destination principle imposed a permanent redistribution for which five years was an arbitrary price. About a dozen states raised the demand at the Council’s June 2022 meeting in Chandigarh. The Union declined, holding that the constitutional provision fixed five years and the regime would end on schedule, and it ended.

Q: How does the equal division of GST revenue between the Centre and the states affect the bargain?

Every rupee of the shared tax collected within a state is divided equally between the Union as the central component and the state as the state component, with the integrated tax on inter-state supplies apportioned by the same logic. The states’ complaint, voiced most formally by Chhattisgarh’s T. S. Singh Deo, is that the equal split understates the states’ share of public expenditure: states together carry the larger part of general government spending, on health, education, police and welfare, yet receive half of the principal consumption tax before the Union’s own half is partly redistributed through Finance Commission devolution. The Union’s answer is that devolution of forty-one per cent of its divisible pool under the Fifteenth Finance Commission’s award already rebalances the total, and that the central component finances subjects the Union alone carries. The dispute is really about which figure, the tax split or the post-devolution total, is the right measure of the federation’s fairness.

Q: Why was the compensation dispute settled without using the dispute mechanism in Article 279A?

Because the mechanism did not exist to be used. Clause (11) of Article 279A directs the Council to establish a mechanism to adjudicate disputes between the Union and states, or between states, arising out of its recommendations or their implementation. For years after the tax began, no such standing mechanism was operationalised, so when the largest fiscal dispute in the tax’s history arrived in 2020 there was no adjudicative body to which the construction of the compensation statute could be referred. The states could have litigated in the Supreme Court, and several finance ministers spoke of doing so, but a judgment risked settling the construction against them permanently. The dispute was therefore settled the only way the institution allowed: by negotiation in the Council and between finance ministries, in which the Union’s construction prevailed and the states’ pressure improved the quantum and the instrument. The demand to operationalise the mechanism has remained on the states’ reform agenda ever since.