The Goods and Services Tax is usually described as one tax, and in economic substance it is: a single destination-based levy on consumption, charged at every stage of a supply chain and credited at every stage after the first, so that the tax sticks only to the value each business adds. But a reader who goes looking for the statute that creates this one tax will not find it, because no such single statute exists. The GST Acts are a family: four Acts of Parliament, a matching Act in every state legislature and in the Union territories with legislatures of their own, and a compensation statute standing behind the whole arrangement, more than thirty enactments that must move in step for the single tax to work at all. This article is the pillar treatment of that family: what each statute levies, how the credit chain that justifies the design actually operates, where the litigation concentrates, and why a taxpayer who received a disputed order in the tax’s early years had, for a very long time, no tribunal to appeal it to.

Three features of the design explain nearly everything that follows. The first is duplication by design: the Centre and each state tax the same transaction at the same time, under separate statutes, because neither level of government would surrender the revenue base to the other. The second is the credit chain: tax paid on inputs is set off against tax charged on outputs, across Centre and state lines and across state borders, which is the mechanism that removes the cascading that disfigured the older system. The third is delegation by design: the Acts are comparatively short, and the operative law, the rates, the thresholds, the exemptions, the return formats and much of the procedure, sits in rules, notifications and circulars issued under them, which is why the practical law of GST is something a practitioner tracks weekly rather than reads once.

What existed before, and why it failed

To see what the GST Acts were built to solve, the starting point is the indirect tax system they replaced, because its defects were structural rather than administrative. Before 1 July 2017, the Centre taxed manufacture through central excise duty, taxed services through a separate service tax, and taxed imports through customs duties with additional components designed to mirror the domestic levies. The states taxed sale of goods through value added tax, taxed entry of goods through entry tax and octroi in many jurisdictions, and levied a scatter of purchase taxes, luxury taxes and entertainment taxes. The two levels of government taxed different events in the life of the same product: the Centre stopped, in principle, at the factory gate, and the state began at the sale. A manufacturer paid excise that a wholesaler could not credit against state VAT. A service provider paid service tax on inputs it could not set off fully against excise. Tax was therefore embedded in prices at every stage, exported between states in the price of goods, and invisible to the final consumer who bore it. The government counted seventeen indirect taxes and twenty-three cesses subsumed into GST, and while the exact count is an official characterisation rather than a legal fact, the fragmentation it describes was real.

The second defect was the border. Because states taxed sales and the Centre could not, goods moving between states attracted central sales tax collected by the origin state, which was not creditable in the destination state and so operated as a tariff inside a country that constitutionally forbids internal trade barriers of exactly that kind. Commercial decisions followed the tax map rather than the market: warehouses were placed state by state to convert inter-state sales into stock transfers that escaped the central sales tax, distribution networks duplicated themselves across borders, and trucks queued at state checkpoints for the verification of forms. The third defect was classification chaos multiplied by jurisdiction: the same product could be an excisable good in the Centre’s tariff, a taxable sale in one state’s schedule at one rate and in another’s at a different rate, and part of a taxable service in a works contract dispute in both. Litigation over whether a transaction was a sale or a service, a work of manufacture or a repair, filled the reports for decades and consumed the appellate machinery the new tax would later inherit.

The constitutional surgery the replacement required can be read off the Seventh Schedule entries it rewrote, and the entries explain why nothing less than an amendment would do. On the Union side, the excise entry was cut down to petroleum products and tobacco, and the services entry in its standalone form gave way to the simultaneous power. On the state side, the entry for taxes on the sale or purchase of goods was confined to the petroleum products and alcoholic liquor carved out of the amendment, the entry tax entry was deleted outright, and the entries for taxes on advertisements and on luxuries and entertainments were deleted or confined to levies by local bodies, so that the states’ general power to tax the sale of goods, the workhorse entry of the old system, survives only in the fields the new tax deliberately left outside itself. A tax on a composite supply, part goods and part services, crossing state lines and consumed in a third jurisdiction, could not be assembled from those entries by interpretation; the entries had to be replaced, and they were, by a power defined over the supply itself.

The political economy of the replacement matters for understanding the statutes’ shape. States were being asked to surrender their most buoyant revenue source, the tax on sales, in exchange for a shared tax whose yield was unproven, and the Centre was being asked to surrender the factory gate. Neither surrender was available for free. The price of the states’ consent was a constitutional guarantee of compensation for revenue loss, and the price of the Centre’s consent was a design in which it collected the inter-state tax and administered the settlement between states. Those two prices are written into the statute family itself: one Act exists only to pay the compensation, and one Act exists only to run the inter-state mechanism. The legislative road that produced this bargain, the seventeen years of negotiation through the Empowered Committee of State Finance Ministers and two failed amendment attempts, belongs to the article on the legislative history of GST in India, and is compressed here to the point that matters for reading the statutes: the Acts are a negotiated federal settlement dressed as a tax code, and several of their strangest features are clauses of that settlement rather than choices a tax designer would make freely.

The constitutional doorway

Ordinarily a tax statute needs no constitutional preamble, but the GST Acts are unintelligible without the amendment that made them possible, because the pre-existing distribution of taxing powers in the Seventh Schedule gave the Centre the power to tax manufacture and services and the states the power to tax sales, and neither could tax the whole of a modern supply. The Constitution (One Hundred and First Amendment) Act, 2016 cut through the division by inserting Article 246A, which gives Parliament and every state legislature simultaneous power to legislate on goods and services tax, reserving inter-state supplies to Parliament alone. Article 269A provides for the levy and collection of the integrated tax on inter-state supplies by the Government of India and for its apportionment between the Union and the states. Article 279A constitutes the Goods and Services Tax Council, the joint body of the Union finance minister and the finance ministers of the states, charged with recommending the taxes to be subsumed, the goods and services to be taxed or exempted, the rates, the thresholds and the model laws themselves. The amendment received assent on 8 September 2016 and was ratified by the required majority of state legislatures in a matter of weeks, a speed that testified to how completely the negotiation had preceded the legislation.

This article states the constitutional basis in outline and links the depth where it belongs: the 101st Amendment and the GST Council owns the amendment’s text, the Council’s composition and weighted voting, and the Supreme Court’s reading of the Council’s powers in the Mohit Minerals litigation, where the Court held that the Council’s recommendations have persuasive rather than binding force and that Article 246A confers a simultaneous power on both levels of government. Readers who need the voting arithmetic, the one-third and two-thirds weights and the three-fourths decision rule, will find them there. What this pillar needs from the constitutional story is only the working consequence: because the power is simultaneous and the market is national, the tax had to be legislated twice over, once by Parliament and once by every state, in texts that mirror each other almost section for section. The multiplicity of the GST Acts is not an accident of drafting. It is the constitutional design operating exactly as written.

The statute family

Four Acts of Parliament stand at the centre of the family. The Central Goods and Services Tax Act, 2017 (Act 12 of 2017) is the workhorse: one hundred and seventy-one sections in twenty-one chapters, carrying the charge on intra-state supplies, the definition of supply, registration, the input tax credit machinery, returns, payment, refunds, assessment, audit, inspection, search, seizure, demands, recovery, offences and penalties, and the entire appellate ladder. The Integrated Goods and Services Tax Act, 2017 (Act 13 of 2017) is a short statute of twenty-five sections that does the cross-border work: it defines inter-state and intra-state supply, contains the place of supply rules that decide which of the two levies applies, levies the integrated tax on inter-state supplies and imports, and provides for apportionment and settlement of funds between the Centre and the states. The Union Territory Goods and Services Tax Act, 2017 (Act 14 of 2017) replicates the state share of the tax for Union territories without legislatures, mirroring the central Act through twenty-six sections. The Goods and Services Tax (Compensation to States) Act, 2017 (Act 15 of 2017), fourteen sections, levies a cess on specified demerit and luxury supplies and routes the proceeds to a compensation fund from which states were paid for revenue loss. All four received the assent of the President on 12 April 2017 and were brought into force on 1 July 2017, the date the tax itself began. Around them stands a State Goods and Services Tax Act in each state and in each Union territory with a legislature, drafted on a common model settled through the Council, reproducing the central Act’s scheme with the state substituted as the taxing authority for the state component, its own officers, its own notifications, and its own amendments that must track the central text whenever the model changes.

The family also includes what the Acts authorise rather than what they contain. Each Act is accompanied by rules, the Central Goods and Services Tax Rules, 2017 being the largest, running to hundreds of rules and forms that prescribe how registration is applied for, how invoices are drawn, how the electronic ledgers are kept and how returns are filed. Under the Acts, the government, acting on the recommendations of the Council, issues notifications that fix rates in schedules, grant and withdraw exemptions, set and reset thresholds, and extend deadlines; the rate schedules that decide what a given product actually bears are notification schedules, not schedules to the Act. Circulars and clarifications from the Central Board of Indirect Taxes and Customs, and from state administrations, interpret the whole mass. The findable artifact of this article, the statute map below, sets the family out in one view, because the single most useful thing a reader can hold in mind about GST is which statute does what and what breaks when one member of the family moves without the others.

Statute	What it levies or governs	Who enacted it	Supplies it reaches	What happens if it falls out of step
Central Goods and Services Tax Act, 2017	The Centre’s share of tax on supplies within a state, plus the common machinery of registration, credit, returns and appeals	Parliament	Intra-state supplies of goods and services	The state mirror statute still levies its share, and the two halves of one tax diverge in procedure and interpretation
Integrated Goods and Services Tax Act, 2017	The combined tax on supplies between states and on imports, the place of supply rules, and settlement between governments	Parliament	Inter-state supplies, imports, and supplies to special economic zones	Cross-border trade loses its single levy, credits strand at state borders, and the destination principle fails
State Goods and Services Tax Acts	The state’s share of tax on supplies within the state, on the common model	Each state legislature, and the legislatures of Union territories with their own assemblies	Intra-state supplies of goods and services	A state amendment out of line with the model breaks uniformity of rates, exemptions and procedure in that state
Union Territory Goods and Services Tax Act, 2017	The territory’s share of tax for Union territories without legislatures	Parliament	Intra-state supplies within those territories	The smallest jurisdictions would have no second levy and no local machinery under the common scheme
Goods and Services Tax (Compensation to States) Act, 2017	A cess on specified supplies, paid into a fund compensating states for revenue loss	Parliament	Notified demerit and luxury supplies, including tobacco products, pan masala, aerated waters and motor vehicles	The federal bargain underpinning the whole family loses its funding mechanism and states bear uncompensated loss
Constitution (One Hundred and First Amendment) Act, 2016	The taxing powers themselves: Articles 246A, 269A and 279A	Parliament with ratification by state legislatures	All supplies, as the source of competence	Without it, neither level of government could levy the tax in its integrated form at all

The last column of the map is the one that explains the administrative character of GST. A rate change for a single product line is not one decision but a coordinated act: the Council recommends, the Centre notifies under the central and integrated Acts, every state notifies under its own Act, and the rate schedules of more than thirty statutes move together. When coordination slips, as it has when state notifications lagged or diverged in wording, the same product bears different treatment across a border the tax was designed to erase. The statutes anticipate this fragility and answer it with the Council, a body whose recommendations are the practical glue; the Supreme Court’s holding in Mohit Minerals that those recommendations are not legally binding makes the glue political rather than legal, a distinction whose fiscal consequences the federalism analysis of GST examines in depth.

The compensation statute at work

The Compensation Act is the shortest of the central four and the one whose terms decided whether the others could exist. Its mechanism is contained in a handful of provisions. Section 3 fixes the base year revenue of each state, the subsumed taxes it collected in the financial year 2015 to 2016, and Section 7 provides for compensation of the gap between that base, grown at a projected rate of fourteen per cent per year, and the revenue the state actually collects from the state tax and its share of the integrated settlement, for a transition period of five years from the commencement of the tax. The cess that funds the promise is levied by Section 8 on intra-state and inter-state supplies of notified goods, tobacco and tobacco products, pan masala, aerated waters and motor vehicles prominent among them, at rates the schedule caps, and is collected into the Goods and Services Tax Compensation Fund under Section 10, from which compensation is released to states in bi-monthly instalments against audited and provisional figures. The five-year period ran to June 2022. The pandemic years broke the arithmetic: collections of the underlying taxes collapsed while the fourteen per cent growth promise ran on, the cess fund fell far short, and the gap was met by back-to-back loans raised by the Centre and passed to the states, Rs 1.1 lakh crore in the financial year 2020 to 2021 and Rs 1.59 lakh crore in 2021 to 2022 on the finance ministry’s published statements, with the cess levy extended to 31 March 2026 for the single purpose of servicing and repaying those loans after the compensation period itself had closed. The extension is the durable lesson of the statute: a transitional fiscal promise, once made the price of a constitutional settlement, acquires a life measured by its debt rather than by its term, and the arguments over extending compensation itself, which the states pressed and the Centre refused, belong to the federalism article’s account of the bargain’s politics.

Inside the central Act: where the weight sits

The central Act’s twenty-one chapters fall into five functional groups, and knowing the groups is worth more than knowing the numbering, because litigation and practice cluster in predictable places. The opening chapters, on definitions, administration and the levy itself, carry the charge, the composition scheme, the time and value provisions and the exemption power; the definitions chapter, Section 2 with its more than a hundred clauses, is applied daily, because questions as basic as what counts as a business, a person, a distinct person, a related person and the place of business of a supplier are answered there and nowhere else. The middle chapters, on registration, invoices, credit, returns and payment, are the compliance engine already described, and carry the system’s administrative weight. The refund chapter and the assessment and audit chapters carry the first level of dispute. The demand and recovery chapters, Sections 73 to 84, are the enforcement core: the two demand tracks and the common provision that succeeds them, interest, the recovery modes of Section 79, the liability rules that reach directors of defaulting companies, partners of firms and legal representatives of the deceased, the bar on proceedings after the limitation periods, and the provisional attachment power. The offences chapter and the appeals chapters complete the Act, with the closing chapters carrying the transitional provisions and the miscellaneous provisions on the burden of proof, the presumption as to documents, restrictions on disclosure of information collected, and the bar on the jurisdiction of civil courts. Two features of the architecture deserve a reader’s notice. First, the Act repeatedly empowers the government, on the Council’s recommendation, to prescribe the operative detail, and the phrase recurs so often that the statute reads in places like an enabling framework for its own rules, which is the textual basis of the notification-driven system described below. Second, the Act was drafted to be mirrored: almost every provision has its twin in the state Acts, so that a judgment on the central text is, in the ordinary run, a judgment on thirty texts at once, which is why a single High Court ruling on a credit condition assumes national significance within days of its delivery and why conflicting rulings between High Courts on the same mirrored words are a standing problem the Supreme Court’s docket inherits.

The levy: supply, charge and the two halves of one tax

Every tax needs a taxable event, and GST’s is the supply. Section 7 of the central Act defines supply expansively to include all forms of supply of goods or services, sale, transfer, barter, exchange, licence, rental, lease and disposal among them, made or agreed to be made for a consideration in the course or furtherance of business. Three extensions give the definition its reach. Schedule I treats specified activities as supplies even without consideration, including the permanent transfer of business assets on which credit was taken, supplies between related persons or between distinct registrations of the same person in the course of business, and imports of services from a related person abroad. Schedule II classifies activities that hover between goods and services, land and building transactions, works contracts, restaurant supply, transfers of business assets and the like, assigning each to one side of the line. Schedule III lists activities that are neither supply of goods nor of services, among them services by an employee to an employer in the course of employment, services by constitutional functionaries, and the sale of land and of completed buildings, which is why a salary attracts no GST and a finished flat, once its completion certificate has issued, falls outside the levy while an under-construction unit falls inside it.

What counts as a supply under the GST Acts?

A supply is any sale, transfer, barter, exchange, licence, rental, lease or disposal of goods or services made for consideration in the course of business, extended by Schedule I to specified transfers without consideration, classified between goods and services by Schedule II, and cut back by Schedule III, which excludes employment services and sales of land and completed buildings.

The charge itself is imposed by Section 9 of the central Act on intra-state supplies, with the state Act imposing its parallel charge in the same words, so that a sale within Maharashtra attracts central tax and Maharashtra tax together, each at half the notified combined rate, collected by the same supplier on the same invoice and reported in the same return. Where the supply crosses a state line, Section 5 of the integrated tax Act charges the full combined rate as integrated tax, collected by the Centre. Imports of goods are treated as inter-state supplies and bear integrated tax at the customs barrier alongside basic customs duty, which alone of the old border charges survives outside GST. Exports and supplies to special economic zone units and developers are zero rated, charged at nil with credit preserved, a point the FAQ section returns to. The design decision embedded in this structure deserves emphasis: the tax follows the destination of consumption, not the origin of production, so revenue accrues to the state where goods and services are consumed. Producing states, which under the old system collected at the factory and the first sale, were the structural losers of the change, and the compensation statute was the instrument that bought their consent to a destination tax.

The rates the charge carries were settled as politics before they were settled as economics. The committee under the Chief Economic Adviser, Arvind Subramanian, had recommended a revenue neutral rate in the region of fifteen to fifteen and a half per cent, with a structure of a standard rate, a lower rate for necessities and a higher rate for demerit goods. What emerged from the Council’s negotiations in 2017 was a four-slab structure of five, twelve, eighteen and twenty-eight per cent, with exemption beneath it and special low rates for gold and rough diamonds beside it, each slab a compromise between the revenue needs of states accustomed to taxing the same goods at value added tax rates of twelve and a half to fifteen per cent plus excise, and the reform argument that multiplicity of rates manufactures classification disputes. The twenty-eight per cent slab drew the compensation cess on top of it for the goods it covered. Within months of the rollout the Council began correcting the structure’s early placements, moving large numbers of goods down from twenty-eight per cent at its Guwahati meeting in November 2017, and the slab question remained the standing item of rate politics until the comprehensive rationalisation of September 2025, taken up in the advance ruling discussion below, collapsed the structure towards the two-rate design the reformers had originally wanted. The Acts themselves hold their peace through all of this: no rate appears in any of them, and a reader of the statutes alone would learn the architecture of the tax and nothing of its weight.

Which levy applies to a given transaction is decided by the place of supply rules, Sections 10 to 13 of the integrated tax Act, and these provisions are among the most consequential and least read in the family. For goods, the place of supply is generally where movement terminates for delivery; for goods that do not move, where they are located at delivery; for goods assembled or installed at site, the place of installation. For services, the general rule takes the location of the recipient, with a ring of specific rules for immovable property services, performance-based services, events, transportation and the like, each overriding the default. A software company in Bengaluru serving a client in Delhi makes an inter-state supply and charges integrated tax; the same company serving a client two streets away charges central and state tax. The invoice is identical in amount, the rate identical, but the head of tax differs, and with it the government that ultimately receives the money and the ledger through which the buyer’s credit flows. Errors in identifying the place of supply do not change the tax paid; they change its character, and they surface years later as demands for the right tax under the right head with refund claims for the wrong one, a category of dispute the statutes created by their own architecture and that the FAQ below takes up directly.

The place of supply rules deserve their own pause, because they are the family’s answer to a question the old system never had to ask: in a tax shared between two governments on every transaction, which government’s share is it? For services, Section 12 of the integrated Act supplies a general rule for domestic supplies, the location of the recipient where the recipient is registered, and the supplier’s location where it is not, and then displaces the general rule wherever a specific rule speaks: services in relation to immovable property take the place where the property stands, so a architect in one state designing a building in another makes a supply located at the building; restaurant, catering, personal grooming and health services take the place of actual performance; training and performance appraisal take the recipient’s location where registered and the place of performance where not; admission to an event takes the place of the event; passenger transportation takes the place where the passenger embarks; and services on board a conveyance take the first scheduled point of departure. For cross-border services, Section 13 applies a parallel scheme: the recipient’s location is the default, with immovable property, performance, event and transport rules of its own, a rule for banking and financial intermediaries that the 2023 omission described later in this article recast, and a special regime for online information and database access or retrieval services supplied from abroad to Indian recipients, under which the foreign supplier registers and pays integrated tax on supplies to unregistered Indian consumers, the mechanism by which streaming and software services sold from overseas entered the Indian net. Each specific rule is a small policy: performance-based rules tax where consumption is visible, the event rule taxes where the audience gathers, the immovable property rule anchors the tax to the one thing that cannot move. The litigation these rules produce is correspondingly granular, and the cost of error is structural rather than financial in the first instance: the same rupee of tax under the wrong head, a refund claim against the government that received it wrongly, a fresh demand from the government entitled to it, and interest arguments between the two, which is why advisers treat the place of supply analysis as the first question in any cross-border or multi-state engagement under these Acts.

The integrated tax is therefore more than a levy; it is a settlement system. The Centre collects integrated tax on inter-state supplies, the exporting state’s businesses claim credit of the central and state tax paid on their inputs against it, the importing state’s buyers claim credit of the integrated tax against their own central and state liabilities, and the funds are apportioned between the Union and the states under Article 269A and the apportionment provisions of the integrated tax Act, with cross-utilisation machinery in the central Act transferring amounts between heads. The chain of transfers runs continuously and invisibly behind every inter-state invoice. When it works, a manufacturer in Tamil Nadu sells to a distributor in Karnataka, the distributor sells to a retailer, and at no point does tax lodge in the price; when it fails, because a credit is blocked or a settlement is delayed, the old cascading reappears inside the new tax. The settlement’s opacity in the early years was a grievance of its own: the Comptroller and Auditor General’s early audit reporting on the tax drew attention to the size of the integrated tax balances awaiting apportionment and to the limits of the information available to the states about the settlement of their own shares, and the Council’s meetings in the period record the states pressing for faster and more transparent settlement as a matter of cash management, since a state whose share of the integrated tax on its imports is settled late has in substance lent its revenue to the system. Ad hoc advances of integrated tax shares to the states were made in the tax’s second year to relieve exactly that pressure, and the episode fixed a principle the family operates by: the integrated tax belongs to the governments it is apportioned among from the moment of collection, and the Centre’s role in it is custodial and mechanical, however invisible the mechanism remains to the taxpayer whose invoice sets it in motion. The integrity of this settlement is the real achievement of the design, and most of what can go wrong with GST in practice is a failure somewhere in the chain the settlement depends on: the credit chain, examined next.

Input tax credit under GST: the mechanism that is the whole point

Input tax credit is the entitlement of a registered person to deduct the tax paid on inputs, input services and capital goods from the tax payable on outputs, so that each business remits tax only on its own value addition and the consumer bears the full amount at the end of the chain. Section 16 of the central Act opens with the entitlement in broad terms and then fences it with conditions, and the fencing is where the tax actually lives for most taxpayers. The claimant must be registered; the goods or services must be used or intended to be used in the course or furtherance of business; and Section 16(2) stacks further conditions in a series. The claimant must possess a tax invoice or debit note issued by a registered supplier. The goods or services must actually have been received. The details of the invoice must have been communicated by the supplier in its outward supply return and must appear in the auto-generated statement, Form GSTR-2B, made available to the recipient, a condition inserted with effect from 1 January 2022 after a transitional period in which provisional credit beyond the matched amount was squeezed down in stages from ten per cent to nil. The tax charged must actually have been paid to the government by the supplier, in cash or through credit. And the claimant must have furnished its own return. Credit is taken through the electronic credit ledger, is available for output tax in a prescribed order of set-off between integrated, central and state heads, and cannot be used against interest, penalty or the reverse charge liability itself, which must be discharged in cash.

Each condition has a rationale in enforcement. The invoice condition ties credit to a document trail. The receipt condition ties it to real movement. The statement condition ties the buyer’s credit to the seller’s reporting, so that a supplier who omits a sale breaks the buyer’s chain and the buyer becomes the administration’s unpaid inspector of its own supply line. The payment condition ties credit to the tax having reached the exchequer. Cumulatively, they convert a self-assessed credit system into a matched system in which the weakest counterparty sets the compliance standard for everyone downstream. The scale of the machinery is worth registering: the finance ministry’s published figures record gross GST collections of Rs 20.18 lakh crore for the financial year 2023 to 2024, an average monthly collection of Rs 1.68 lakh crore, every rupee of it netted through credit claims of the kind these conditions govern. The conditions are defensible as anti-evasion architecture in the abstract. Their contested character comes from where they place the loss when a supplier fails: on the buyer, who has already paid the tax once, inside the price, to the seller.

A worked example fixes the mechanics better than the definitions do. Take a manufacturer in Pune selling machinery components to a distributor in the same state for Rs 10 lakh, at a combined rate of eighteen per cent. The invoice charges Rs 90,000 of central tax and Rs 90,000 of Maharashtra tax. The manufacturer had bought steel and sub-assemblies for Rs 6 lakh plus tax, and services for Rs 1 lakh plus tax, accumulating credit of roughly Rs 1.26 lakh across the two heads; it sets that credit against the Rs 1.8 lakh charged on its sale and pays the balance, about Rs 54,000, in cash, which is the tax on its own value addition of Rs 4 lakh of margin and overheads recovery. The distributor sells on to a retailer in Indore, in another state, for Rs 12 lakh. The supply is inter-state, so the invoice charges integrated tax of Rs 2.16 lakh; the distributor uses the Rs 1.8 lakh of credit inherited from its purchase, in the statutory order of set-off, and pays the rest in cash. The retailer sells to a consumer in Madhya Pradesh for Rs 14 lakh, charging central and state tax of Rs 1.26 lakh each; the retailer’s credit is the integrated tax of Rs 2.16 lakh, which the cross-utilisation rules allow against its central and state liabilities in the prescribed order, and the governments then settle among themselves so that Madhya Pradesh, the consumption state, receives the state share and the Centre its own. At the end of the chain the consumer has borne Rs 2.52 lakh of tax on a Rs 14 lakh purchase, each business has remitted only the tax on its margin, and no tax has lodged in any price on the way. Every failure mode described in this article is a break in this sequence: a supplier who does not file, a credit blocked by Section 17(5), a place of supply misidentified so that the wrong head is charged, a buyer whose seller collected the Rs 1.8 lakh and vanished. The arithmetic is elementary; keeping it unbroken across crores of invoices a month is the administrative undertaking.

When must a tax invoice be issued under GST?

A supplier of goods must issue a tax invoice before or at removal of the goods where movement is involved, or before or at delivery where it is not; a supplier of services must issue it within the prescribed period, thirty days in the ordinary case. The invoice is the document on which the recipient’s credit rests.

Two further disciplines govern the ledger’s use rather than its accumulation. The utilisation order in Sections 49A and 49B and the rules under them requires integrated credit to be exhausted before central or state credit is touched, and prescribes the sequence among the heads, a discipline whose amendment in 2019, making the order of set-off more flexible in one respect and more rigid in another, stranded balances in the state and central ledgers of businesses whose trade ran mainly in one direction and generated its own corrective notifications. And a cash discipline in the rules, directed at large monthly filers, caps the proportion of output liability that may be discharged from the credit ledger at ninety-nine per cent for taxpayers above a turnover line, requiring the balance in cash; the measure, aimed at fly-by-night operators whose ledgers were their only asset, was felt by genuine exporters and inverted-duty businesses as a working capital charge and illustrates the system’s habit of answering fraud at the margin with friction applied to the whole flow.

Two further fences complete the picture. Section 16(4) imposes a long-stop date: credit on an invoice or debit note for a financial year must be claimed by 30 November of the following financial year or the date of furnishing the annual return, whichever is earlier, a deadline the Finance Act, 2022 substituted for the earlier September-return formulation and one that closes the books on a year’s credit with finality. Section 17(5) blocks credit outright on a list of items Parliament considered prone to personal consumption or abuse: most motor vehicles and their repair and insurance, food and beverages and outdoor catering, health and life insurance in the ordinary case, goods and services for personal consumption, works contract services for immovable property other than plant and machinery, goods lost, stolen, destroyed or written off, and tax paid in the fraud and detention provisions among others. The blocked list is a favourite examination topic and a steady source of disputes at its edges, over what counts as plant and machinery, over vehicles used for specified transport businesses that are carved back into eligibility, and over employer obligations, where credit is allowed where the employer is statutorily obliged to provide the benefit. The general principle that organises the section is that credit follows business use, and that where an input plausibly serves the proprietor’s private consumption, the Act refuses the credit rather than policing the use.

The supplier-payment condition and the credit litigation

The condition in Section 16(2)(c), that the tax charged on a supply must actually have been paid to the government by the supplier, is the single most contested feature of Indian GST, and the brief for this article flags it as such for good reason. Its operation is stark. A buyer selects a registered supplier, receives the goods, pays the invoice including the tax through banking channels, and claims credit. The supplier pockets the tax and disappears, or files no return, or files and does not pay. The department, unable to recover from the supplier, reverses the buyer’s credit with interest, reasoning that the exchequer never received the money and the credit was therefore never earned. The buyer has paid the tax twice in substance: once to the supplier in the price, and once to the government in the reversal. The statutory text supports the department; Section 155 of the central Act places the burden of proving eligibility to credit on the person claiming it, and the Supreme Court, in State of Karnataka v Ecom Gill Coffee Trading, decided in 2023 under the predecessor value added tax but in reasoning the Court itself connected to the GST burden provision, held that a purchasing dealer must prove the genuineness of the transaction and the actual movement of goods, and that tax invoices and payment records alone do not discharge that burden where the sellers turn out to be fictitious or defaulting.

Against that current runs a line of High Court authority protecting the bona fide buyer. The Madras High Court, in D.Y. Beathel Enterprises v State Tax Officer, decided on 24 February 2021, quashed orders that had fastened the entire liability on purchasing dealers whose sellers, registered in the same assessment circle, had collected tax and never remitted it. The Court held that the recovery machinery ought first to be directed at the defaulting seller, that the buyers had paid the tax component through banking channels against valid invoices, and that fastening liability on them without even examining the sellers offended natural justice; the matters were remitted for fresh enquiry against the sellers. Other High Courts have reasoned in the same direction in cases of retrospective cancellation of a supplier’s registration, holding that a buyer who dealt with a supplier registered at the time of the transaction cannot be made to answer for a cancellation effected backwards. The department’s answer in these cases, and it is not a weak one, is that credit is a statutory concession conditioned by the Act, that the conditions are cumulative, and that a system which paid out credit on invoices whose tax never arrived would invite invoice mills to industrialise the fraud, a species of evasion GST administrations have fought continuously through fake registration drives and blocked credit of identified shell suppliers.

The honest position, and the one this article adopts, is that the litigation is the visible surface of a design choice that has never been politically owned. Parliament conditioned the buyer’s credit on the seller’s payment because the administration lacked, and lacks, the capacity to verify millions of transactions in real time; the condition drafts every buyer into enforcement and allocates the residual fraud loss to the party least able to bear or prevent it. High Courts soften the edge case by case, through the bona fide buyer doctrine, without displacing the condition, and the Supreme Court’s burden-of-proof reasoning pushes the other way on proof of the underlying transaction while leaving the payment condition itself untouched. The scale of the fraud the condition polices is not in dispute, and fairness to the design requires recording it. The administration’s periodic drives against fake registrations and fake invoicing, conducted jointly by the Centre and the states and reported by the finance ministry in cumulative terms, have identified networks of shell registrations existing only to issue invoices, claim and pass credit, and vanish, with detected fraudulent credit running to amounts that are material against the tax’s total yield; the invoice mill is not a hypothetical of the drafting committee but the standing adversary of the credit chain, and the buyers in the decided cases include both the mill’s victims and, in other proceedings, its customers. It is this mixture, in the same evidentiary posture, of the duped and the complicit, that makes a rule of general protection for buyers as hard to administer as the condition is hard to bear, and that has kept the courts working case by case rather than laying down the general rule either side wants.

The questions the courts have not squarely answered, whether Section 16(2)(c) can constitutionally deny credit to a buyer who proves everything within its own control, and whether interest can run against such a buyer for the seller’s default period, remain genuinely open, and challenges raising them have been pursued in several High Courts without a final Supreme Court word on the condition’s validity. A related and quieter body of litigation concerns the matching condition: the phased restriction of credit to what appears in the recipient’s statement, first through Rule 36(4) and then through the statutory GSTR-2B condition, was challenged as arbitrary and as punishing reporting failures indistinguishable, from the buyer’s side, from payment failures, and the transitional credit cases of the rollout years, in which dealers claimed credit carried over from the excise and value added tax regimes into GST, generated their own stream of writ petitions over forms, deadlines and portal failures. The pattern across all of it is the one the series thesis predicts: the operative law is made less by the Act’s text than by the friction between the text and the systems built under it.

Registration and the thresholds

Registration is the gate into the system: below the thresholds, a supplier is outside the Act entirely, charges no tax, claims no credit and files no returns; above them, the full machinery applies. Section 22 of the central Act fixes the basic lines by aggregate turnover in a financial year. A supplier of goods must register where turnover exceeds Rs 40 lakh, a threshold doubled from the original Rs 20 lakh with effect from 1 April 2019. A supplier of services must register above Rs 20 lakh. In the special category states named in the Article 279A framework and the registration provisions, the lines sit lower: Rs 20 lakh for goods and Rs 10 lakh for services, and the composition ceiling and other concessions follow the same geography of disadvantage. Turnover here is aggregate turnover across all of a person’s registrations on the same permanent account number, counting taxable, exempt and export supplies together, which is why a modest business with branches or verticals in several states crosses lines faster than its owners expect, and why registration is taken state by state, producing a separate registration, a separate identification number and a separate return set for each state of operation. The Goods and Services Tax Identification Number itself encodes the design: fifteen characters, opening with the two-digit state code, followed by the ten-character permanent account number of the person, an entity code for multiple registrations within a state, and a check character.

Section 24 overrides the thresholds for categories Parliament wanted inside the net regardless of size: persons making inter-state taxable supplies of goods, casual taxable persons, persons liable to pay under reverse charge, electronic commerce operators and those who supply through them within the provision’s scope, non-resident taxable persons, input service distributors, and persons required to deduct or collect tax at source under Sections 51 and 52. The inter-state override was softened for services and for notified handicraft goods, but for goods it remains the provision that pulls a very small seller into registration the moment a parcel crosses a state line, a deliberate choice to keep the settlement chain unbroken at borders. Voluntary registration is available below every threshold, and its economics explain much small-business behaviour under the tax: a supplier selling to registered businesses registers voluntarily because its customers demand invoices carrying credit, accepting the full compliance load as the price of market access, which is the credit chain operating as an enrolment device. The procedure of registration carries the system’s identity architecture. Application is made on the common portal against the applicant’s permanent account number, with the place of business and the promoters’ particulars declared; authentication of applicants by Aadhaar was layered into the process by amendment, and physical verification of premises follows in the cases the rules flag, so that the shell registrations the credit condition battles are fought at this gate as well. Registration takes effect from the date liability arose where application is made within thirty days of that date, and from the date of grant where it is later, a distinction that decides the validity of invoices and the credit of customers in the intervening weeks. Amendments to the registration, of address, constitution or places of business, follow the graded route of Section 28, core fields requiring the officer’s approval and the rest taking effect on intimation; the casual taxable person and the non-resident taxable person register in advance of their activity and deposit the tax on their estimated liability upfront, the system’s way of taking security from persons with no continuing presence to proceed against. Registration can be cancelled by the officer for contraventions including non-filing and fraudulently obtained registration, and by the taxpayer on closure or falling below threshold, and cancellation carries its own litigation, particularly where effected retrospectively, since retrospective cancellation reaches back to destroy the credit its customers already claimed in good faith.

The composition scheme: simplicity at the price of the chain

For small suppliers the Acts offer a deliberate exit from the credit chain. Section 10 of the central Act lets a registered person with aggregate turnover up to Rs 1.5 crore in the preceding financial year, Rs 75 lakh in the named special category states, pay tax at a flat rate on turnover instead of the standard rates on value added: one per cent for manufacturers and traders, five per cent for restaurants not serving alcohol, with a parallel scheme in Section 10(2A) for service providers and mixed suppliers up to Rs 50 lakh at six per cent. The composition dealer files a quarterly statement of self-assessed tax and an annual return, issues a bill of supply rather than a tax invoice, must state on that bill that it is a composition taxable person not eligible to collect tax on supplies, and is spared the invoice-level matching machinery that governs the regular register. The concessions are purchased with three prohibitions that define the scheme’s place in the system: the composition dealer cannot collect tax from its customers, cannot claim input tax credit on its own purchases, and cannot make inter-state outward supplies. Its tax is a cost embedded in its prices, exactly the cascading the main design exists to remove, tolerated at small scale because the administrative cost of full registration for a neighbourhood trader exceeds the cascade it would prevent.

The exclusions around the scheme are as instructive as its terms. Manufacturers of notified goods, ice cream, pan masala and tobacco products among them, are barred from composition altogether, as are suppliers through electronic commerce operators who collect tax at source, and casual and non-resident taxable persons. The restaurant rate of five per cent without credit made composition effectively mandatory in character for much of the small restaurant trade, where input credits are thin and consumer prices sensitive. What the scheme means in practice, and what the FAQ below answers directly, is that composition is a decision about position in the chain: a business selling to final consumers, with few creditable inputs, gains from simplicity; a business selling to other businesses loses, because its customers receive no credit from its invoices and will price that loss into what they pay it. The scheme’s take-up and periodic revision, and the complaints that the ceiling leaves a gap between the composition trader and the fully registered competitor just above the line, recur in Council deliberations precisely because the boundary is doing economic work: every threshold in these Acts sorts businesses into regimes, and every sorting creates a cliff edge that behaviour then clusters against.

Returns, ledgers and the compliance architecture

The Acts are administered through a common digital infrastructure run by the Goods and Services Tax Network, a not-for-profit company incorporated in 2013 to build and operate the common portal, and subsequently brought into full government ownership shared between the Centre and the states, an institutional journey that mirrors the tax’s federal character. Registration, returns, payments, refunds and the electronic ledgers all live on that portal, and three ledgers structure every registered person’s position: the electronic liability register, recording what is owed; the electronic cash ledger, recording tax, interest and fee paid in cash; and the electronic credit ledger, recording input tax credit available for set-off. The monthly cycle runs in two movements. In the outward return, Form GSTR-1, the supplier reports invoice-level details of its supplies; those details flow into the recipients’ auto-drafted statement, Form GSTR-2B, which fixes the credit a recipient may take for the period. In the summary return, Form GSTR-3B, the taxpayer declares its output tax, claims the credit the statement supports, and pays the difference in cash. An annual return, Form GSTR-9, consolidates the year for taxpayers above the notified turnover line, with a self-certified reconciliation statement above a higher line. What began in 2017 as a three-return matching design, in which buyer and seller returns were to be reconciled invoice by invoice before credit settled, was simplified within the first years into this statement-based system after the original matching returns proved unworkable at scale; the statute’s credit conditions were then rewritten, as described above, to rest on the auto-drafted statement rather than on the abandoned reconciliation.

One characterisation question from the early years shows how much turns on the forms’ legal status. Because the abandoned matching design had treated the summary return as a provisional statement, taxpayers argued that credit deadlines and interest computations keyed to the furnishing of a return could not run from a form that was not, in the statutory sense, the return; the Gujarat High Court, in the AAP and Company litigation, held that Form GSTR-3B is a return under Section 39 for the relevant purposes, reading the transitional dispensation by which the summary form was pressed into service as making it the operative return of the system. The holding mattered because the whole compliance calendar, the credit long-stop, the interest start date, the limitation for demands, keys off returns, and the case is a reminder that under these Acts the forms prescribed by rule are not stationery but load-bearing law, amended, substituted and litigated as such.

The compliance burden this architecture imposes is the second great criticism of the tax, and it is a criticism about arithmetic as much as principle. A business registered in ten states holds ten registrations, files the monthly pair of returns in each, maintains state-wise ledgers, and reconciles its purchase register against a statement generated from its suppliers’ filings on a timetable it does not control. The Acts set the frame: Section 37 for outward supplies, Section 38 for the inward statement, Section 39 for the returns themselves, Section 59 for self-assessment as the governing principle, with scrutiny, audit under Section 65 and special audit under Section 66 behind it, and inspection, search and seizure under Section 67 at the coercive edge. Late filing attracts a late fee under Section 47 and interest under Section 50, eighteen per cent per annum on delayed payment of tax in the ordinary case, with interest on wrongly availed and utilised credit addressed after litigation and amendment on the question whether interest can run where credit was availed but never used. The assessment the evidence supports is double-sided. For a business already inside the formal economy, the system replaced a dozen incompatible state and central filings with one calendar and one portal, and the credit chain made its tax cost visible and recoverable. For a small business entering formality, the same system presented monthly deadlines, accounting software costs, and liability for the filing behaviour of every supplier, a burden qualitatively different from the annual value added tax return it replaced. Both accounts are true, and the distance between them tracks firm size almost exactly.

How is a GST demand raised when tax has been short paid?

A demand begins with a show cause notice and an adjudication order. Where no fraud or wilful misstatement is involved, Section 73 governs, with an order ordinarily within three years of the annual return’s due date; where fraud is alleged, Section 74 applies with five years. Section 74A, inserted in 2024, governs later years.

One refinement of the demand machinery changed its economics materially and deserves its place here. Interest on delayed payment was charged in the tax’s first years on the gross tax liability, including the portion the taxpayer could have discharged from credit, so that a return filed a month late on a liability fully covered by credit attracted interest as though nothing had been available to pay with. Taxpayers resisted on the text of Section 50, which charges interest on tax not paid, arguing that tax dischargeable from the ledger was not unpaid in the relevant sense; the amendment that restricted interest to the net cash liability, the portion actually payable in cash after credit is exhausted, was enacted and then, by the Finance Act, 2022, given retrospective effect from the commencement of the tax, extinguishing a large population of gross-liability interest demands and confirming that in this system the largest financial consequences often turn on a proviso rather than on a charging section.

The demand provisions repay attention because they define the practical life of a GST dispute more than the appellate provisions do. Interest follows the tax as a matter of course; penalty scales with the culpability finding, ten per cent of the tax in the non-fraud route and up to the full tax in the fraud route, with statutory discounts for early payment at the notice and order stages that make settlement arithmetic a standing feature of GST practice. The classification of a case as fraud or non-fraud therefore decides the limitation period, the penalty exposure and the settlement terms at once, which is why the invocation of the fraud route against what taxpayers characterise as interpretive disagreements, over classification, place of supply or credit eligibility, is itself heavily contested in the writ courts. Recovery follows under Section 79 through the usual modes, garnishee of bank accounts and debtors among them, and the provisional attachment power in Section 83, exercisable during the pendency of proceedings to protect revenue, has generated a sharp line of cases insisting that so drastic a power be used sparingly and on recorded reasons, an insistence the Supreme Court voiced in the attachment litigation arising from the Himachal Pradesh administration. None of this machinery is unique to GST in kind; what is unique is its volume, driven by the data-matching notices the portal makes possible, in which a difference between returns, between the outward statement and the summary return, or between the tax return and the income tax record, generates a system-prompted enquiry that the taxpayer must answer with documents.

Scrutiny, assessment and audit: the road to a demand

Before a demand comes the assessment architecture that produces it, and the Acts build it in layers of increasing intrusiveness. The foundation is self-assessment under Section 59: the registered person assesses its own liability in its returns, and the administration’s first instrument is not an assessment but a scrutiny under Section 61, in which discrepancies noticed in a return, increasingly discrepancies thrown up by data matching between the outward statement, the summary return, the e-way bill record and the income tax filings, are communicated to the taxpayer, whose explanation, if accepted, closes the matter, and if not, feeds the coercive layers. Those layers are audit by the department under Section 65, conducted at the taxpayer’s premises or the officer’s on fifteen working days’ notice and ordinarily to be completed within three months, extendable by six; special audit under Section 66, directed where the Commissioner considers that value has not been correctly declared or credit not correctly availed, and conducted by a nominated chartered or cost accountant at the department’s instance; summary assessment under Section 64 in urgent cases; and the two best judgment provisions, Section 62 for the person who simply does not file, whose assessment is made to the officer’s best judgment and stands withdrawn if the return is filed within the statutory window, and Section 63 for the unregistered person who ought to have registered. The pattern across the layers is deliberate: the Acts prefer the return to the assessment, the explanation to the audit, and the audit to the demand, and reserve the fraud track for the cases the administration is prepared to characterise and prove as such. In practice the scrutiny stage has become the system’s busiest room, because automated discrepancy notices issue at a scale no officer corps could generate unaided, and a taxpayer’s GST life is largely spent answering them; the professional complaint that the automated notice answers the statute’s preference for explanation with volume rather than judgment is the administrative form of the compliance critique already recorded. A connected practice has attracted the writ courts’ attention at the same layer: payment made by a taxpayer during investigation or audit, tendered through the voluntary payment form the rules prescribe, and later sought to be characterised by the payer as extracted under pressure and by the department as an admission closing the matter. The courts have treated the voluntariness of such payments as a question of fact, have directed refunds with interest where payment was taken without any liability in contemplation, and have equally refused to unwind payments that the record showed to be deliberate settlements; the episode belongs in this pillar because it marks the point where the self-assessment system’s informality meets the coercive edge, and the boundary the courts drew is procedural fairness rather than any concession on liability.

The e-way bill and the movement of goods

A destination-based credit tax still has to police the physical movement of the goods it notionally follows, and the e-way bill is that police mechanism in digital form. Under Rule 138 of the central rules, a registered person causing movement of goods of consignment value exceeding Rs 50,000, in a vehicle or conveyance, must generate an electronic way bill on the common portal before the movement begins, declaring the supplier, the recipient, the goods, their value and the transporter; the bill carries a validity measured in days by distance, and the person in charge of the conveyance must carry the invoice and the bill’s particulars for verification en route. The system replaced the forest of state entry forms, check posts and transit passes of the value added tax era with a single national document, and its data feed gives the administration a real-time map of goods movement that no Indian tax had possessed. Exemptions cover specified goods, specified movements such as transport to a weighbridge within a short radius, and movements by non-motorised conveyance, and states administer parallel intra-state thresholds under their own rules, so the operative detail is again found in rules and notifications rather than in the Act, which confines itself to the enabling machinery of inspection and interception in Sections 67 and 68.

Where goods move without a valid bill, or in breach of one, Sections 129 and 130 supply the consequences, and their severity is the part of the system transporters actually experience. Section 129 permits detention or seizure of the goods and the conveyance, released on payment of a penalty calibrated to the tax payable where the owner comes forward, and a higher measure linked to the value of the goods where the owner does not; Section 130 permits confiscation of goods and conveyance where the officer finds intent to evade, with a fine in lieu of confiscation and the tax and penalty in addition. The courts have read a mens rea discipline into the confiscation power and have repeatedly released goods where a documentation lapse, an expired bill caught in transit delay, a wrong vehicle number, stood unaccompanied by any intention to evade, while sustaining action where the bill was a facade for unaccounted goods. The FAQ below sets out the consequences in order. The e-way bill’s second life is as data. Because the bill is generated before movement and carries the invoice’s particulars, its aggregate record is a near-real-time census of inter-state and intra-state commerce in goods, and the administration matches it against returns in both directions: movement declared without a corresponding supply in the outward return, and supplies declared without movement where the goods should have moved. The matching feeds the scrutiny layer described earlier and the registration verification drives, since a registration that generates bills without filing returns, or files returns without the movement its trade implies, presents the pattern the system is tuned to find. The transporter’s obligations complete the design: the person in charge of the conveyance carries the documents, the transporter may be required to update the vehicle particulars where goods are transhipped, and the bill’s validity runs with distance so that a delayed consignment must be extended before it expires rather than explained after it is intercepted. The structural point for this pillar is that the e-way bill is the clearest example of the series thesis in the GST family: a two-line enabling provision in the Act, a rule running to pages, a portal, a penalty jurisprudence, and an entire compliance profession between them.

What happens to tax that a supplier collects but never deposits?

Tax collected from a recipient but not paid over must be paid to the government forthwith under Section 76 of the central Act, whether or not the supply was taxable, and is recoverable with interest and penalty. Retention of collected tax is among the gravest defaults in the scheme, and the factual core of most fake-invoice prosecutions.

Anti-profiteering: the provision that outran its method

Section 171 of the central Act states a simple duty: any reduction in the rate of tax on a supply, and the benefit of input tax credit, must be passed on to the recipient by a commensurate reduction in price. The duty was the political corollary of the reform’s promise that removing cascading would lower prices, and it made the GST Acts unusual among tax statutes in reaching past the taxpayer to the price charged to a stranger. Parliament provided for an authority to examine whether the benefit had been passed on, and the government constituted the National Anti-Profiteering Authority in late 2017, supported by a screening committee structure and the Directorate General of Anti-Profiteering as its investigating arm. The provision’s difficulty was present from the first case: commensurate reduction is a formula the Act does not supply. Where a rate falls, is the comparison product by product or across the supplier’s whole range; against which base price, where input costs, discounts and pack sizes move simultaneously; over what period must a credit benefit in a long supply chain express itself in price? The rules prescribed a procedure for complaints, investigation and orders, including orders to reduce prices, refund the profiteered amount with interest to recipients or deposit it in the consumer welfare fund, and penalty in specified cases, but no methodology for computing the amount, and the Authority built its method case by case, generally at the level of the individual product and the individual supplier’s margins.

The complaint pathway shows the mechanism in operation. An interested person, a recipient, a consumer or an organisation, applied to the standing and screening committees constituted under the rules; an application that disclosed a prima facie case went to the Directorate General of Anti-Profiteering, which investigated the supplier’s pricing and credit position across its product range and reported; the Authority then heard the supplier and ordered, where profiteering was found, a reduction in prices, refund of the amount profiteered to the recipients where they could be identified with interest at eighteen per cent, deposit of the amount in the consumer welfare fund where they could not, and in the graver cases penalty and cancellation of registration as the rules provided. The decided cases ran where the reform’s incidence was most visible: builders, whose pre-GST and post-GST effective rates on under-construction housing diverged sharply and whose buyers organised to complain in numbers; restaurants, on the November 2017 rate reduction for the sector; and manufacturers of consumer goods, where the computation ground down to pack sizes, grammage changes and trade margins. The builder cases generated the methodological heartland of the dispute, since the profiteered amount on a multi-year project depends on choices about the project’s total area, the instalment schedule and the stage at which credit benefits are reckoned, and no two of the parties’ computations ever agreed.

The litigation that followed attacked the provision itself: that Section 171 delegates essential policy without guidance, that it confers uncanalised power to fix prices, that it punishes without a measurable standard. The Delhi High Court answered the challenge in January 2024 in the Reckitt Benckiser batch of cases, upholding the constitutional validity of Section 171 and of the rules under challenge, characterising the provision as a consumer welfare measure in a tax reform whose benefits were the public’s, not a price control statute, and rejecting the argument that a single prescribed methodology was a condition of validity, while leaving individual determinations of profiteered amounts to be judged on their facts. The institutional history ran in parallel: the Authority’s tenure was extended repeatedly, its functions were transferred to the Competition Commission of India with effect from 1 December 2022, and anti-profiteering work was subsequently assigned to the Principal Bench of the Goods and Services Tax Appellate Tribunal, with a notified cut-off after which no new applications for examination would be entertained from 1 April 2025, closing the mechanism’s intake while its accumulated cases worked through. The fair assessment is that Section 171 was easy to state, popular to announce and genuinely hard to administer: it policed a real risk, that tax cuts would be absorbed as margin, with an instrument whose precision its own administrators had to invent, and its winding down marks the system’s judgment that the discipline of competition and of notified rate politics would have to do the work instead.

The statutes in the courts

Beyond the credit litigation, three lines of authority mark the courts’ treatment of the family’s foundations. The first concerns the character of the Council. In Union of India v Mohit Minerals, decided in May 2022, the Supreme Court struck down the levy of integrated tax on the ocean freight component of imports made on a cost, insurance and freight basis, holding that the importer, already paying integrated tax on the composite value of goods including freight, could not be taxed again on the freight as a deemed service; the wider observations in the judgment, that the recommendations of the Council are the product of a collaborative dialogue and have persuasive value, that Article 246A treats the Union and the states as equal units in the simultaneous power, and that the Indian federal structure contemplated by the amendment is not a hierarchy in this field, are the passages for which the case is cited, and they are developed in the 101st Amendment article to which this pillar has already linked. The second line concerns the sanctity of the return as filed. In Bharti Airtel v Union of India, the Supreme Court set aside a High Court direction that had permitted a large taxpayer to rectify its summary returns for the rollout quarter of 2017, in which an input credit of some magnitude had been under-reported through the mismatch between the then return design and the taxpayer’s accounting, holding that the Act’s scheme of self-assessment, adjustment in subsequent returns and departmental scrutiny is the route the statute provides, and that the rectification the taxpayer sought lay outside it; the case fixed the principle that the return, once filed, speaks through the statutory correction mechanisms and not through general directions. The third line concerns the place of supply of intermediary services, where Section 13(8)(b) of the integrated Act deems the place of supply to be the location of the supplier, so that an Indian intermediary serving a foreign client was denied export status and zero rating available to comparable service exporters; the Gujarat High Court held the provision unconstitutional in the Material Recycling Association line of cases as discriminating against Indian intermediaries, the position remained contested and administratively worked around, and the provision was omitted by the Finance Act, 2023 with effect from 1 October 2023, restoring the destination rule to that class of services. Each line teaches the same lesson from a different side: the Acts’ framework survived every challenge, and in each case what the courts corrected was a condition, a mechanism or a deeming rule at the framework’s edge, which is where this statute family keeps its litigation permanently.

Advance rulings: certainty at the first door

Between the assessment machinery and the appellate ladder, the Acts place a device for advance certainty: the Authority for Advance Ruling in each state, Sections 95 to 106 of the central Act, to which an applicant may put specified questions on classification, applicability of notifications, determination of time and value of supply, admissibility of credit, liability to pay tax and liability to register, and receive a ruling binding on the applicant and its jurisdictional officers, with an appeal to the Appellate Authority for Advance Ruling in the state. The device was meant to cheapen certainty, and its history is a caution inside the larger story: because each state’s authority rules independently, the same product drew conflicting rulings across states, on the classification of foods, on solar and works contracts, on intermediary services, and the escape route for conflict, a national appellate authority for advance rulings, was legislated late and designated to the Tribunal’s Principal Bench only when the Tribunal finally stood up. The classification disputes that fill the public argument about GST, the different rates borne by apparently similar preparations of food, by a roti and a paratha in the famous ruling on that distinction, by a hand sanitiser treated as a medicament or a disinfectant, are substantially a consequence of the rate structure rather than of administrative caprice: where a schedule taxes at five, twelve, eighteen and, after the rate rationalisation of September 2025 collapsed the middle slabs, principally at five and eighteen per cent with a forty per cent demerit rate above them, a classification line is a rate line, and every rate line through a continuum of similar products will produce pairs of near-identical goods on opposite sides of it. The remedy most often demanded, fewer slabs, is a live policy question the Council has acted on once and will be pressed on again, and this article treats it as such rather than as a settled conclusion in either direction.

What does the Authority for Advance Ruling decide under GST?

It answers an applicant’s questions on classification of goods or services, the applicability of exemption notifications, the time and value of a supply, admissibility of input tax credit, whether a supply is taxable, and whether the applicant must register. Its ruling binds the applicant and the applicant’s jurisdictional officers, and an appeal lies to the Appellate Authority for Advance Ruling.

The GST appellate ladder, and the missing tribunal

The appellate structure of the central Act is conventional on paper. A person aggrieved by an order of an adjudicating authority appeals under Section 107 to the Appellate Authority, within three months of communication, after paying the admitted dues and a pre-deposit of ten per cent of the remaining tax in dispute, subject to a statutory maximum; the filing stops recovery of the balance. The Commissioner may invoke revision under Section 108 against an order erroneous and prejudicial to revenue. From the Appellate Authority or the Revisional Authority, a second appeal lies under Section 112 to the Goods and Services Tax Appellate Tribunal, again within three months, with a further pre-deposit of ten per cent of the remaining disputed tax so that twenty per cent stands deposited in all, and with the recovery of the balance statutorily stayed on payment. From the Tribunal, Section 117 takes an appeal to the High Court on a substantial question of law, and Section 118 to the Supreme Court. The general scheme of taxpayer appeals and of the tribunal system into which this ladder fits is the territory of the article on taxpayer appeals and tribunals; what belongs to this pillar is the fact that for most of the tax’s life, the second rung of its own ladder did not exist.

The Tribunal was constituted by Sections 109 and 110 of the central Act as enacted: a National Bench, regional and state benches, each bench of one judicial member and two technical members, one drawn from the Centre and one from the state, with qualifications for appointment set out in Section 110, including eligibility of members of the Indian Legal Service of specified seniority as judicial members. In Revenue Bar Association v Union of India, decided on 20 September 2019, the Madras High Court struck down Sections 109(3) and 109(9), which prescribed that bench composition, and Section 110(1)(b)(iii), which made the legal service route to judicial membership available, holding that the number of expert members on a bench cannot exceed the number of judicial members, that a tribunal deciding questions of statutory interpretation and the legality of administrative action requires a properly trained judicial mind not in a decorative minority, and that the composition as enacted offended the Supreme Court’s tribunalisation jurisprudence in the Madras Bar Association line of cases. The Court rejected, it should be recorded, the argument that excluding advocates from judicial membership was itself unconstitutional, while recommending that Parliament consider making advocates eligible. The effect of the judgment was that the Tribunal could not lawfully be populated under the enacted text, and the text was not promptly rewritten.

What followed was the general tribunal reform saga running through the tax statute. The conditions of service of tribunal members across statutes had been legislated in the Finance Act, 2017 and the rules under it, struck down in material part by the Supreme Court in the Rojer Mathew judgment; a Tribunals Reforms Ordinance in 2021 and then the Tribunals Reforms Act, 2021 re-enacted a framework of tenure and appointment which the Supreme Court in its turn struck down in part, in the Madras Bar Association judgments of 2021, for re-enacting what had already been held impermissible on tenure and minimum age. The tax statute was then amended again: the Finance Act, 2023 rewrote Sections 109 and 110 with effect from 1 August 2023, reorganising the Tribunal into a Principal Bench and state benches and raising the judicial strength of each bench to two judicial members alongside two technical members, so that technical members no longer outnumber judicial members, and the Central Goods and Services Tax (Amendment) Act, 2023 aligned the qualifications and conditions further with the tribunal reforms framework. The Goods and Services Tax Appellate Tribunal (Procedure) Rules followed in 2025. A President, Justice Sanjaya Kumar Mishra, a retired Chief Justice of the Jharkhand High Court, had been appointed in May 2024, and the Tribunal was formally launched on 24 September 2025, with a Principal Bench at New Delhi, thirty-one state benches at forty-five locations, an electronic filing portal, and staggered windows notified for the enormous accumulated stock of appeals against first appellate orders, the filing window for the legacy stock running into 2026, with hearings commencing bench by bench thereafter. The Tribunal as finally constituted is worth describing in its working form, because its design carries the settlement of the 2019 judgment. The Principal Bench sits at New Delhi and the state benches are distributed so that every state and group of Union territories has a bench within reach; a bench ordinarily sits with two judicial members and two technical members, one of the Centre and one of the state, in parity the Madras High Court’s reasoning required, and the President may constitute smaller benches, including single-member benches, for classes of smaller appeals, a case-management necessity where the bulk of the accumulated stock involves modest demands. The procedure rules of 2025 put the Tribunal’s work on an electronic footing from its first day: appeals are filed on the Tribunal’s own portal, with electronic records, virtual hearings and cause lists, a deliberate break from the paper culture of the departmental appellate authorities below it. The Principal Bench carries the additional national functions the system had parked for want of a home, including the anti-profiteering jurisdiction described earlier and the national appellate function for advance rulings, so that the institution whose absence distorted the system for eight years arrives loaded with the three hardest unification tasks in the tax: appellate consistency between states, classification consistency between advance ruling authorities, and the residue of the price-passing experiment.

Between the launch of the tax on 1 July 2017 and the launch of its tribunal in September 2025, eight years passed. The pre-deposit design of the ladder also shaped behaviour in the interval in ways worth recording. Twenty per cent of a disputed demand, deposited across the two appellate stages in addition to the admitted liability, is a working capital charge that falls with particular weight where the demand itself arises from a credit reversal: the taxpayer has, in substance, paid the disputed tax once to its supplier, faces reversal with interest, and must deposit a fifth of the sum for the privilege of contesting the reversal. The department, for its part, litigates under monetary discipline: the Board has fixed monetary limits below which departmental appeals are not to be filed before the Tribunal, the High Courts and the Supreme Court, a policy the courts hold the revenue to with some strictness, so that the appellate population is skewed towards the larger demands and the smaller taxpayer’s dispute ends, for practical purposes, at the first appellate authority whose order the writ courts may or may not disturb. High Courts, receiving the writ traffic the absent tribunal displaced, developed a local jurisprudence of restraint and intervention: many declined to entertain challenges where an appellate remedy would in principle exist, many others entertained them precisely because it did not, and the same order could be unreviewable in one state and examinable in another for no reason beyond the accident of which High Court’s roster the assessee’s registration fell under. A national tax administered through state registrations reproduced, at the appellate stage, the forum inequality the tax was designed to abolish at the levy stage, and the restoration of a functioning second appeal is best understood as the repair of that inequality rather than as a mere addition of capacity.

The interval was not lawless, but it was expensive, and the statutes and orders that managed it deserve recording because they show a system improvising around an absent institution. The government’s Central Goods and Services Tax (Ninth Removal of Difficulties) Order, 2019, issued on 3 December 2019, provided that the three-month period for appealing to the Tribunal would run from the later of the date of communication of the order and the date on which the President or the State President of the Tribunal entered office after its constitution, so that the appeal clock could not expire against a forum that did not exist. The Finance (No. 2) Act, 2024 wrote a version of the same protection into Section 112 itself, running the period from the later of communication and a date the government notifies on the Council’s recommendation for filing appeals before the Tribunal. Taxpayers who could not wait went where Indian litigants always go when a statutory forum fails: to the High Courts under Article 226, at writ-petition cost, with results that varied by jurisdiction and with the courts repeatedly recording their disquiet at deciding, as the first appellate forum in substance, questions a tribunal existed on paper to decide. Official statements at the Tribunal’s launch put the accumulated stock of appeals in the region of five lakh cases awaiting the second appellate stage. The delay’s cause, it should be said plainly, was not administrative sloth alone: the same design question, how much judicial presence a tribunal that displaces High Court jurisdiction must carry, delayed the company law tribunals for a generation, and the profile of the NCLT and NCLAT traces the identical contest there, which is why this article links the two stories rather than treating the GST delay as a local accident. A general description of a tax dispute in this period should also carry the caution this series applies to all procedural writing: the account here is of the statutory structure and its history, and it is not advice on any individual matter, for which limitation periods and forum requirements are unforgiving and specific advice is necessary.

Time of supply, value of supply and the tax base

Three sets of provisions decide the questions that arise on every single invoice: when the tax falls due, on what amount, and at what rate. The time of supply provisions, Sections 12 to 14 of the central Act, fix the point at which liability crystallises, for goods by reference to the invoice date and the last date on which the invoice ought to have been issued, for services by reference to invoice and payment dates in a prescribed sequence, with special rules for advances, for supplies under reverse charge, and for vouchers and interest or late fees added after the event. The provisions matter beyond bookkeeping because they allocate a supply between tax periods and, at every rate change, between the old rate and the new one, and because interest runs from the period the time of supply assigns. The value of supply, Section 15, adopts the transaction value as the base: the price actually paid or payable where the parties are unrelated and price is the sole consideration, with statutory additions for amounts the supplier is liable to pay but the buyer bore, incidental expenses, interest for delayed consideration, and subsidies linked to price, and a set of valuation rules in the rules for the cases the definition cannot reach, supplies between related persons and distinct registrations, supplies through agents, and cost-based fallbacks. The related-person rules matter because the distinct-persons fiction, under which each registration of the same legal person is treated as a separate person for GST, makes even an internal stock transfer or a head office service to a branch a supply to be valued, a consequence Schedule I creates and the valuation rules must then price, generally by the open market value or, where the recipient is entitled to full credit, by the invoice value the supplier declares, a concession that confines most intra-group valuation disputes to cases where the recipient’s credit is restricted.

Rate, the third variable, is not in the Act at all, as the notifications section below explains, but the classification on which rate depends is a statutory exercise: the goods schedules adopt the customs tariff’s harmonised system nomenclature, and the services are classified under a scheme of service accounting codes, so that a rate dispute is usually a tariff interpretation dispute conducted with the tools of customs law. The September 2025 rationalisation, which collapsed the twelve and twenty-eight per cent slabs and left five and eighteen per cent as the standard rates with a forty per cent rate reserved for demerit goods, changed the stakes of classification without changing its method, and transitional questions of stock, credit accumulation and anti-profiteering on the changeover were handled, in character, by notification and clarification rather than by amendment of the Acts.

Payment, set-off and refunds

Credit is useful only in the order the law permits it to be spent, and the payment provisions of the central Act, Sections 49 to 53, prescribe that order with care, because each head of tax belongs to a different government. Integrated tax credit is spent first against integrated tax, then against central tax and then against state or territory tax in the notified order; central tax credit cannot be used against state tax nor state credit against central tax, a hard wall between the two halves of the intra-state levy that preserves each government’s revenue while the settlement machinery transfers the corresponding funds between them. Amounts are credited and debited through the electronic ledgers described earlier, and transfers of a person’s business, amalgamations and demergers carry the ledger balances with them under Section 18 and the transfer provisions, so that credit survives the reorganisation of the person who earned it. Interest, penalty and fee cannot be paid from the credit ledger at all.

Refunds occupy Chapter XI of the central Act, centred on Section 54, and they are the mirror image of the credit system: where credit accumulates faster than output tax absorbs it, on exports and on purchases taxed higher than the goods sold under the inverted duty structure, the surplus is refundable in money, subject to the doctrine of unjust enrichment that runs through all Indian indirect tax refunds, under which a refund is paid to the claimant only where the tax burden was not passed on, and otherwise credited to the consumer welfare fund, with export refunds carved out of the doctrine because the exporter passes the burden to no domestic consumer. Zero rating under Section 16 of the integrated Act gives the exporter two routes: supply under bond or letter of undertaking without paying integrated tax and claim refund of unutilised credit, or pay the integrated tax and claim refund of the tax paid. The inverted duty refund generated a significant holding when the Supreme Court, in Union of India v VKC Footsteps India, in 2021, upheld the validity of the rule that restricts the refund formula to inputs that are goods, excluding input services from the computation, against the argument that the Act’s promise of refund of unutilised credit admitted no such distinction; the case is the standing illustration that the operative content of a GST entitlement is as often found in a rule and a formula as in the section that announces it. Delayed refunds carry interest under Section 56, and the sixty-day clock in that section is one of the few statutory timelines in the family drafted against the administration rather than against the taxpayer. In practice the refund chapter works through the portal against the ledgers: a zero-rated exporter receives a provisional refund of ninety per cent of the claimed amount on a prima facie verification, the balance following scrutiny, and since the disbursement of refunds was consolidated in a single authority in the taxpayer’s own administration, the applicant no longer pursues the central and state components separately through two treasuries, a simplification that mattered more to exporters’ working capital than several headline rate cuts. The deficiency memo, by which an incomplete application is returned to be filed afresh, restarts the limitation clock and is the refund system’s characteristic small injustice, attended by a line of orders insisting that a curable defect pointed out late cannot be used to defeat a claim that was timely when first made.

Who administers the Acts

The statutes are administered twice over, by design. Sections 3 to 6 of the central Act create the central officer hierarchy, the Board at its head, and then provide for cross-empowerment: officers appointed under the state Acts are authorised as proper officers under the central Act, and central officers under the state Acts, so that one officer can ordinarily complete one proceeding across all heads. The taxpayer base is divided administratively between the Centre and the states, by a formula settled in the Council under which the preponderance of smaller taxpayers is administered by the states and the balance shared, with the promise that a taxpayer answers to one administration only for audit and enforcement, a single-interface assurance that holds in the ordinary case and frays where intelligence-based enforcement is concerned, since both administrations receive the same data and either may act on it. State Acts replicate the structure with the state commissioner at the head, and the result is that the same statutory text is worked by thirty-odd administrations whose circulars, trade notices and audit selections diverge in emphasis while the law they administer is uniform in terms. The Goods and Services Tax Council’s secretariat and the Network complete an institutional triangle, legislature, administration and platform, that no earlier Indian tax possessed, and the friction among the three, a Council recommendation awaiting notification, a notification awaiting a portal change, a portal facility awaiting a rule, accounts for a recognisable share of the compliance failures taxpayers attribute to the law itself.

Exemptions and the boundary of the levy

The levy is bounded on three sides, and the boundaries are litigated as hard as the charge itself. The constitutional boundary keeps alcohol for human consumption outside the GST Acts altogether, by the definition of the tax in Article 366(12A), and keeps petroleum crude, high speed diesel, motor spirit, natural gas and aviation turbine fuel outside until the Council recommends a date for their inclusion, which it has not done, so the two largest revenue sources of the old system remain in the old world of excise and value added tax, a carve-out whose fiscal federalism dimensions the federalism article owns and which this pillar records only as a boundary of the statute family. The statutory boundary is drawn by Section 11, under which the government, on the Council’s recommendation, exempts goods and services, unconditionally or subject to conditions, by notification; exempt treatment differs sharply from zero rating, because the exempt supplier charges no tax and also loses credit on its inputs in proportion under Section 17(2), embedding tax in its cost, whereas the zero-rated supplier charges nil and keeps credit. The third boundary is the nil rate and the negative list of Schedule III already described. The exemption schedules are the most politically active documents in the system: exemptions for food items, health services, educational services and the like are announced, conditioned, withdrawn and re-announced through them, and the distinction they create between an unregistered supplier, an exempt supplier and a zero-rated supplier, three states that all present as charging no tax on the invoice, is among the most persistent sources of error in the system’s daily operation.

One model, thirty administrations: the state Acts in practice

The state Acts are the least read members of the family and the ones under which most taxpayers are actually assessed, because the state administration handles the preponderance of the register. Each state Act reproduces the central Act with the substitutions the model requires: state tax for central tax, the state commissioner for the central authorities, the state consolidated fund as the destination of collections, and the state’s own appellate authority and its bench of the Tribunal. The common model law was settled through the Council before enactment, and the state legislatures passed their Acts in special sessions in the weeks before 1 July 2017, an enactment effort without precedent in its simultaneity. Two consequences flow from the mirror design and are worth a reader’s attention. The first is that amendment is a relay: when Parliament amends the central Act through a Finance Act, each state must amend its own Act to match, and the matching amendments trail the central change by months in some states, leaving windows in which the central text and a state text differ on a live point, a threshold, an interest provision, a return requirement, and the taxpayer registered in both jurisdictions lives under both versions. The second is that administration is genuinely plural within a uniform text: states issue their own circulars on the same provisions, empanel their own audit selections, run their own enforcement drives and staff their own advance ruling authorities, and the same transaction structure can attract a ruling in one state that its neighbour’s authority declines to follow, a divergence the national appellate mechanism for advance rulings was designed, belatedly, to discipline. Jammu and Kashmir’s position illustrates the family’s federal plumbing: the central Act initially excluded it, a separate extension Act of 2017 (Act 26 of 2017) carried the central levy there once the state’s own constitutional steps were completed, and the state’s reorganisation in 2019 moved its units between the state and Union territory columns of the family. The Acts are uniform; the system they create is uniform law administered as thirty related systems, and much of what practitioners call GST practice is the management of that plurality.

Special channels: distributors, job work, electronic commerce and deduction at source

Four mechanisms handle the points where the ordinary invoice chain does not fit the commerce it must tax. The input service distributor, Sections 2(61) and 20 of the central Act, is an office of a supplier that receives invoices for common input services, head office legal, audit, software and brand services, and distributes the credit to the registrations that consumed them by a prescribed formula, because the registration that receives the invoice is rarely the registration that uses the service; the compulsory routing of such distribution through the distributor mechanism, tightened by amendment, and the parallel question whether a head office must cross-charge branches for internally generated services, produced a settled clarification only after conflicting advance rulings and a Board circular accepting that where the branch takes full credit, the invoice value declared is to be accepted for internally generated services. Job work, Sections 19 and 143, lets a principal send inputs and capital goods to a job worker without paying tax, against the discipline of return within one year for inputs and three years for capital goods, failing which the dispatch is deemed a supply on the day it was made; the mechanism preserves the pre-GST job-work economy of the textile, engineering and jewellery trades inside the credit chain. Electronic commerce is taxed through Section 52, under which the operator collects tax at source at a small notified percentage of the net value of supplies made through it by other suppliers, and through the registration overrides that pull operators and their sellers into the net. Section 51 requires government departments, agencies and notified persons to deduct tax at source on payments to suppliers above a threshold value. Each mechanism is a patch on the chain at a point where value moves without a conventional sale, and each generates its own compliance calendar and its own refund tail, since tax collected or deducted at source enters the supplier’s cash ledger and is absorbed against liability or refunded.

Offences, penalties and the coercive edge

Chapter XIX of the central Act arranges the consequences of default in ascending order: monetary penalties in Section 122 for twenty-one specified offences, from supplying without an invoice to issuing an invoice without supply; a general penalty in Section 125 up to Rs 25,000; confiscation and penalty for goods in transit in Sections 129 and 130 as already described; prosecution in Section 132 for the gravest offences, including supplying without invoice with intent to evade, issuing invoices without supply, and collecting tax and failing to deposit it beyond three months, with imprisonment graded by the amount involved, rising to five years where the tax evaded exceeds Rs 5 crore, and a bar on prosecuting without the sanction of the Commissioner, with compounding of offences available under Section 138 on payment. The power of arrest in Section 69, exercisable by the Commissioner on reasons to believe that a specified grave offence has been committed, has been the most contested of the coercive provisions; the Supreme Court’s jurisprudence on arrest in tax statutes, requiring that the power be exercised on material and not as a routine lever of recovery, and the 2025 judgment in the challenge to the arrest and search provisions of the GST and customs laws, which upheld the provisions while reading in procedural safeguards drawn from criminal procedure, frame the limits within which the power operates. The design question underneath the litigation is the same one the credit condition poses in civil form: how much coercive pressure a self-assessed mass tax may place on the compliant to reach the evasive, and the courts’ consistent answer has been that the powers stand but their routine use against interpretive disputes does not. The penalty code has itself been amended in the direction of the fake invoice economy: a provision inserted in 2020 makes any person who retains the benefit of a transaction covered by the invoice offences, the beneficiary behind the paper supplier, liable to a penalty equal to the tax sought to be evaded, reaching through the shell to the person the shell served. Compounding under Section 138 is unavailable for the gravest repeat conduct, and the prosecution thresholds have been revised by amendment to reserve the criminal process for the larger evasions, with the monetary floor for prosecution raised for most offences. The trajectory across the family’s first decade is a steady migration of the offences chapter away from the general taxpayer and towards the invoice trade, which is an implicit admission, in legislative form, of where the system’s real exposure lay all along: not in the arithmetic of honest returns but in the documentary fiction the credit chain can be made to honour.

Can a taxpayer be arrested under the GST Acts?

Yes. Section 69 of the central Act empowers the Commissioner, where there are reasons to believe that a person has committed specified grave offences, including issuing invoices without supply or collecting tax without depositing it, to authorise arrest. The power is graded by the amount involved, requires recorded reasons, and is exercised within the safeguards of criminal procedure.

The transition: entering the new system

The Acts did not start from a blank register. Chapter XX of the central Act, Sections 139 to 142, carried the old world into the new: existing registrants under excise, service tax and state value added tax were migrated to provisional registration and then to final registration on furnishing the prescribed particulars; the closing balances of credit under the old laws were carried into the electronic credit ledger through the transitional forms, subject to conditions that the credit was admissible under both laws and was carried in the last returns; opening stock of dealers who had been outside the old credit chains received deemed credit at prescribed percentages; and works contracts, instalments and continuous supplies straddling 1 July 2017 were apportioned by the point of taxation rules the chapter supplied. The transitional credit, known by the form that claimed it, TRAN-1, generated a disproportionate share of the new tax’s first litigation, over the form’s deadline, over portal failures that defeated timely filing, and over whether a credit earned under the old law was a vested right the new law’s procedure could extinguish; the Supreme Court’s eventual intervention directed the administration to open a window for taxpayers to file or revise the transitional forms, and the episode stands in this pillar as the first demonstration of the pattern the credit litigation later confirmed, that in this system procedural failure and substantive entitlement collide constantly, and the courts are asked to price the collision.

The amendment arc: a statute rewritten annually

No account of this family is complete without the rhythm of its amendment, because the central Act has been amended in almost every year of its life, principally through the annual Finance Act and occasionally through a dedicated amendment Act, and the state Acts have followed in the relay described earlier. The pattern of the amendments teaches the system’s learning curve. The amendment Act of 2018 rationalised the cross-utilisation of credit, recast the registration provisions for electronic commerce and casual persons, and adjusted the composition scheme’s exclusions. The Finance Acts of the pandemic period adjusted deadlines, interest and the late fee structure, and wrote the restriction of interest to the net cash liability. The Finance Act, 2021 wrote the GSTR-2B condition into Section 16(2), converting the matching experiment from a rule into a statute, and tightened the detention and confiscation provisions for goods in transit. The Finance Act, 2022 moved the credit long-stop to the thirtieth of November and gave the net cash interest position retrospective effect. The Finance Act, 2023 and the dedicated amendment Act of that year rebuilt the Tribunal provisions and omitted the intermediary place of supply rule. The Finance (No. 2) Act, 2024 consolidated the demand provisions into the common Section 74A for the years from 2024 to 2025, wrote the notified-date protection into the Tribunal appeal period, reduced the maximum pre-deposit at the appellate stages, and made distribution of common input service credit through an input service distributor compulsory with effect from 1 April 2025, closing the cross-charge controversy by channel. Two conclusions follow for the reader of this pillar. The Acts are not a settled code being glossed; they are a working machine being continuously adjusted by its operators, and the direction of adjustment is consistent: conditions migrate from rules into the statute once they survive challenge, tribunal and appellate design is rebuilt under judicial pressure, and every simplification for the compliant arrives paired with a fresh instrument against the evasive. A summary of GST that does not date its propositions is therefore unreliable in a way a summary of a stable statute is not, which is why this article has stated its mechanisms with their amendment history attached at the points where the history changes the meaning.

Why the practical law lives in notifications

Step back from the machinery and the family’s oddest feature comes into focus: the Acts, all thirty-odd of them together, do not tell a trader what tax a given supply bears. The rate schedules are notifications. The exemptions are notifications. The thresholds for registration were changed by notification before being absorbed in amendment. The composition rates, the late fee waivers of the pandemic years, the extension of every filing deadline that has ever been extended, the very dates on which provisions commenced, all arrived by notification, most of them on the recommendation of the Council, whose meetings function as the tax’s real legislative sessions. The annual Finance Act amends the central Act, and the states follow, but between Finance Acts the law moves by delegated instrument at a tempo no legislature could sustain: the Council has met more than fifty times, and the notification output of the system runs to many thousands of instruments across the Centre and the states. For the reader this series addresses, the consequence is a discipline rather than a complaint. Reading the GST Acts is necessary and radically insufficient. The statute supplies the architecture and the liabilities in principle; the liability in fact is assembled each quarter from the Act, the rules, the rate and exemption notifications in force for the period, and the clarificatory circulars that tell officers how the administration reads them, and any account of GST that stops at the statute book, as most summaries do, describes a skeleton and calls it the animal.

This is also where the teacher’s and the aspirant’s payloads sit, and the companion habit this series recommends for examination subjects applies with full force to tax: the Acts reward being turned into one’s own provision maps and rate tables, revised as notifications land. Readers preparing teaching material from this family of statutes will find that laying the credit chain and the statute map out as slides, and reading the deck in the browser through the ReportMedic PowerPoint viewer, is a practical way to keep a thirty-statute structure in view at once while the details shift underneath it.

The honest assessment

The reform narrative and the compliance critique are the two standing accounts of this statute family, and each holds part of the truth at full strength. The reform narrative, stated by the governments that enacted it and by the finance ministry’s reporting since, points to a genuine structural achievement: a fragmented system of cascading central and state levies, internal tariffs and border checkpoints was replaced by a single destination-based tax with a continuous credit chain; the tax base widened as the chain pulled suppliers into formality; collections, on the finance ministry’s published figures, reached Rs 20.18 lakh crore in the financial year 2023 to 2024 and rose further the following year, a scale no combination of the predecessor taxes approached; and interstate trade was freed of the checkpoint and the origin-state levy that had distorted the geography of Indian distribution. The corporate response was visible within the first years: distribution networks designed around state tax borders were redrawn around consumption, small state warehouses were consolidated into regional hubs, and the logistics industry reorganised itself for a country that had become, for the movement of goods, a single territory in substance. On the structural questions, whether a credit chain beats a cascade and whether a common market beats seventeen tax jurisdictions, the narrative is simply right.

The compliance critique, stated most forcefully by the small-business associations and the practitioners who serve them, and acknowledged in pieces by every committee that has reviewed the system, is also right about its own subjects. The design decision to condition the buyer’s credit on the seller’s payment transferred enforcement risk from the state to the taxpayer, and that transfer, not any rate, is the single most resented feature of the tax among honest businesses. The return and reconciliation burden fell on firms that had never dealt with a national tax administration, in monthly instalments, under penalty. The appellate tribunal whose absence forced a generation of disputes into writ courts was a failure of the state’s side of the bargain: taxpayers were held to monthly deadlines by a system that took eight years to build the forum its own statute promised. And the anti-profiteering experiment showed the limits of legislating an outcome, lower prices, without legislating a method. The verdict this article reaches is that the GST Acts are a sound architecture imperfectly honoured by their own administrators: the statutes’ central mechanisms, the dual levy, the integrated settlement and the credit chain, have worked as designed and will not be unbuilt, while the system’s lived defects, credit denial for counterparty default, procedural multiplication across thirty registrations, and the tribunal gap, are defects of the conditions and institutions around the text rather than of the text’s conception. Reform pressure therefore runs, correctly, toward easing the credit conditions for the demonstrably bona fide, collapsing the slab structure further, and making the tribunal system ordinary, rather than toward replacing the family itself.

Four open questions stand where the family’s next amendments will be fought. The inclusion of the petroleum products, constitutionally scheduled to enter the tax on the Council’s recommendation, would complete the credit chain for the economy’s largest input and remains blocked by the revenue dependence of both levels of government on the old levies. Electricity, taxed by states under a separate duty outside the Acts, breaks the chain for every business that consumes power, which is every business. Real estate straddles the boundary in the system’s most awkward posture: under-construction property bears the tax while land and completed buildings fall outside it, stamp duty continues on the whole transaction value in parallel, and the industry’s credit restrictions on residential projects are a studied compromise that satisfies neither the affordability argument nor the chain’s logic. And the inverted duty structure, in which inputs bear more than outputs in sectors such as textiles, footwear and fertilisers, converts the refund machinery from a safety valve into a permanent feature for those industries, periodically corrected by rate changes that move the inversion rather than abolish it. None of these is a drafting accident; each is a place where the federal bargain or the revenue imperative overrode the design, and each will be settled the same way the original bargain was, in the Council and then in thirty notifications.

What remains for the reader is the cluster around this pillar. The constitutional settlement underneath the Acts, the Council and its voting, is treated in the 101st Amendment article linked at the constitutional section above; the passage story belongs to the legislative history article and the federal bargain to the federalism article, both linked at their points of need in the text. The appeals system as a whole, across income tax and GST together, and the general pathology of Indian tax tribunals, belong to the taxpayer appeals article. This pillar’s claim can be left in one sentence: India chose to build one tax out of more than thirty statutes that must move in step, and everything distinctive about GST in practice, its notification-driven law, its credit litigation, its missing tribunal and its federal friction, is that choice working itself out.

Frequently Asked Questions

Q: Which statutes make up India’s GST framework?

Four Acts of Parliament and a state Act in every state make up the framework. The Central Goods and Services Tax Act, 2017 carries the Centre’s share of the tax on supplies within a state and most of the common machinery. The Integrated Goods and Services Tax Act, 2017 covers supplies between states, imports, and the place of supply rules. The Union Territory Goods and Services Tax Act, 2017 covers Union territories without legislatures. The Goods and Services Tax (Compensation to States) Act, 2017 levies the cess that funded compensation to states. Alongside them, each state legislature and each Union territory with a legislature enacted its own State Goods and Services Tax Act on a common model, so the family runs to more than thirty statutes, all founded on the Constitution (One Hundred and First Amendment) Act, 2016.

Q: What is the difference between CGST, SGST and IGST?

The three are the heads under which the same tax is collected, and which head applies depends on where the supply begins and ends. CGST is the Central Goods and Services Tax, the Centre’s share on a supply made within one state. SGST is the State Goods and Services Tax, the state’s share on that same supply, charged alongside CGST at half the combined rate each. IGST is the Integrated Goods and Services Tax, charged at the full combined rate on supplies between states and on imports, collected by the Centre and then apportioned between the Centre and the destination state. A supply within Maharashtra bears CGST plus Maharashtra SGST; the identical supply from Maharashtra to Karnataka bears IGST. The amount is the same; the government that receives it differs.

Q: How does input tax credit work under GST?

A registered business charges GST on its sales, called output tax, and pays GST on its purchases, called input tax. Input tax credit is the right to set the input tax against the output tax, so the business remits only the difference, which equals the tax on the value it added. Credit is claimed through the electronic credit ledger on the common portal and may be used against integrated, central and state tax in a prescribed order. The entitlement in Section 16 of the central Act is fenced by conditions: the buyer must hold a tax invoice, must have received the goods or services, the invoice must appear in the buyer’s auto-drafted statement, the supplier must actually have paid the tax to the government, and the buyer must have filed its return. Credit that accumulates beyond output tax, as with exporters, is refundable in the cases the Act specifies.

Q: Who must register under GST?

Registration turns on aggregate turnover and, for some categories, on the nature of the activity alone. A supplier of goods must register where aggregate turnover in a financial year exceeds Rs 40 lakh, and a supplier of services where it exceeds Rs 20 lakh; in the named special category states the lines are Rs 20 lakh for goods and Rs 10 lakh for services. Turnover is counted across all registrations held on the same permanent account number. Section 24 of the central Act requires registration regardless of turnover for specified persons, including those making inter-state taxable supplies of goods, casual taxable persons, persons liable under reverse charge, electronic commerce operators within the provision’s scope, non-resident taxable persons and input service distributors. A person below every threshold may register voluntarily, and many do, because registered customers prefer suppliers whose invoices carry credit.

Q: Who can opt for the composition scheme under GST?

A registered person whose aggregate turnover in the preceding financial year did not exceed Rs 1.5 crore may opt for the composition scheme under Section 10 of the central Act; the ceiling is Rs 75 lakh in the named special category states. Manufacturers, traders and restaurants not serving alcohol are the core eligible classes, paying tax at flat rates on turnover, and a separate scheme covers service providers and mixed suppliers up to Rs 50 lakh. Eligibility is lost by specified conduct and status: manufacturers of notified goods such as ice cream, pan masala and tobacco products are excluded, as are casual and non-resident taxable persons and suppliers making inter-state outward supplies or supplying through electronic commerce operators who collect tax at source. The option is exercised for the financial year and applies to all registrations of the person together.

Q: What is the anti-profiteering provision in GST?

Section 171 of the central Act requires that the benefit of a reduction in the rate of tax on a supply, and the benefit of input tax credit, be passed on to the recipient by a commensurate reduction in price. The provision was enforced first by the National Anti-Profiteering Authority, constituted in 2017 with an investigating arm, then by the Competition Commission of India from 1 December 2022, and subsequently by the Principal Bench of the Goods and Services Tax Appellate Tribunal, with a notified cut-off closing new applications from 1 April 2025. The Authority could order price reductions, refunds of the profiteered amount with interest, deposit in the consumer welfare fund where recipients could not be identified, and penalty in specified cases. The Delhi High Court upheld the provision’s constitutional validity in the Reckitt Benckiser batch in January 2024, while leaving individual computations to be judged on their facts.

Q: Why did the GST Appellate Tribunal take so long to start working?

The Tribunal’s constitution provisions were struck down before a single bench could sit. In Revenue Bar Association v Union of India, decided on 20 September 2019, the Madras High Court struck down the provisions of the central Act prescribing a bench of one judicial member and two technical members, holding that expert members cannot outnumber judicial members on a tribunal deciding questions of law. The general law on tribunal service conditions then went through its own cycle: a reforms ordinance and the Tribunals Reforms Act, 2021 were struck down in part by the Supreme Court, and the tax statute had to be amended again, by the Finance Act, 2023 and a dedicated amendment Act, to raise the judicial strength of each bench and align qualifications with the Court’s rulings. A President was appointed in May 2024, procedure rules followed in 2025, and the Tribunal was formally launched on 24 September 2025, more than eight years after the tax began.

Q: What is an e-way bill and when is it required?

An e-way bill is an electronic document generated on the common portal before goods begin to move, declaring the supplier, the recipient, the goods, their value and the transporter. Under Rule 138 of the central rules, a registered person causing movement of goods in a vehicle or conveyance, where the consignment value exceeds Rs 50,000, must generate the bill before the movement starts; the person in charge of the conveyance carries the invoice and the bill’s particulars for verification on the road. The bill’s validity runs in days measured by the distance to be covered, and it must be extended if the journey overruns. Specified goods and specified movements are exempt, and states prescribe parallel requirements for movement within their borders. The system replaced the entry forms and check posts of the value added tax era with a single national document.

Q: What is the place of supply under GST and why does it matter?

The place of supply is the statutory test that decides which government taxes a transaction and under which head. Sections 10 to 13 of the Integrated Goods and Services Tax Act supply the rules. For goods, the place is generally where movement ends for delivery, where goods sit if they do not move, and where they are installed if assembly at site is involved. If the place of supply and the supplier are in the same state, the supply is intra-state and bears central plus state tax; if in different states, it is inter-state and bears integrated tax. For services, the default is the recipient’s location, displaced by specific rules for property-related, performance-based, event and transport services. The answer changes nothing about the amount of tax and everything about its destination, which is why an error surfaces later as a refund claim under one head and a demand under the other.

Q: What is reverse charge under GST and who actually pays the tax?

Reverse charge reverses the usual direction of payment. Ordinarily the supplier collects the tax from the recipient and pays it to the government; under Sections 9(3) and 9(4) of the central Act, and the corresponding provisions of the integrated Act, the recipient pays the tax directly to the government for notified categories of supply and for specified supplies from unregistered persons. The categories fixed by notification include goods transport agency services supplied to specified businesses, legal services by advocates to business entities, sponsorship, and services imported from outside India, where the Indian recipient pays the integrated tax. The recipient pays in cash, not from credit, and then takes credit of the tax paid where the supply is used in its business, so for a fully taxable business the mechanism is a cash-flow cost and a compliance checkpoint rather than a final burden. It exists to collect tax where the supplier is outside the administration’s easy reach.

Q: What is blocked credit under Section 17(5) of the CGST Act?

Section 17(5) lists inputs on which input tax credit is denied outright even though the tax was paid and the purchase was for the business. The list covers most motor vehicles for the transport of persons and the vessels and aircraft outside the transport and training exceptions, and with them the insurance, repair and maintenance of such vehicles; food and beverages, outdoor catering, beauty treatment, health services and cosmetic surgery outside the exceptions; membership of clubs and fitness centres; travel benefits extended to employees on vacation; goods and services for personal consumption; works contract services and goods for constructing immovable property other than plant and machinery; goods lost, stolen, destroyed, written off or given away as free samples; and tax paid under the fraud, detention and confiscation provisions. Where the employer is obliged by law to provide the benefit, or where the input falls inside the transport and manufacturing exceptions the section preserves, credit is available.

Q: What is the time limit for claiming input tax credit?

Section 16(4) of the central Act sets a long-stop date. Credit on an invoice or debit note pertaining to a financial year may be taken no later than 30 November of the financial year that follows, or the date on which the annual return for that year is furnished, whichever is earlier. The Finance Act, 2022 substituted this November date for the earlier deadline tied to the September return. Credit not claimed by the long-stop is lost, and the return for a later period cannot revive it, which makes the date one of the hard deadlines of the GST calendar and the November filing season the point at which a year’s purchase registers are finally reconciled against the auto-drafted statements. Credit wrongly blocked in the portal, or missed through a supplier’s late filing, must be chased before the date, not after it.

Q: How are exports taxed under GST?

Exports are zero rated under Section 16 of the Integrated Goods and Services Tax Act, which means the supply is taxed at nil while the exporter’s input tax credit is preserved and refunded rather than absorbed as a cost. An exporter chooses between two routes: supply under a bond or letter of undertaking without paying integrated tax, and claim refund of the unutilised input tax credit; or pay the integrated tax on the export and claim a refund of the tax paid. Supplies to a special economic zone developer or unit are zero rated on the same footing, being treated as inter-state supplies made in the course of inter-state trade. Zero rating is deliberately distinct from exemption: an exempt supplier charges no tax and loses credit on its inputs in proportion, while a zero-rated supplier charges none and recovers the whole of its input tax, keeping Indian tax out of export prices entirely.

Q: Can a composition dealer collect tax from customers or claim input tax credit?

No on both counts, and the two prohibitions define the scheme. A composition dealer pays tax at a flat rate on its own turnover and cannot recover that tax from the customer as tax: it issues a bill of supply rather than a tax invoice, must state on the bill that it is a composition taxable person not eligible to collect tax on supplies, and cannot charge any amount as GST on the bill. It also cannot claim input tax credit on its purchases, so the tax it pays on inputs becomes part of its cost, the deliberate price of leaving the credit chain. A customer buying from a composition dealer receives no credit either, which is why the scheme suits businesses selling to final consumers and penalises businesses selling into a supply chain of registered buyers.

Q: What is the Goods and Services Tax Network and what does it operate?

The Goods and Services Tax Network is the not-for-profit company that builds and operates the common portal through which the GST Acts are administered. It was incorporated in 2013, ahead of the tax, to create a shared information technology platform for the Centre and all the states, and it was later brought into full government ownership shared between them. It operates the systems for registration, the filing of returns, the payment of tax, the electronic cash, credit and liability ledgers, the auto-drafted statements on which recipients’ credit depends, the refund processing interface and the e-way bill and e-invoicing platforms. Every registered person in the country deals with the tax through it, which makes it the operating system of the statute family: when the portal changes a facility, the practical law for millions of taxpayers changes with it, without any amendment to any Act.

Q: What happens if goods are transported without an e-way bill?

Moving goods without a required e-way bill, or under a bill that has expired or does not match the consignment, exposes the goods and the vehicle to detention or seizure under Section 129 of the central Act at the point of interception. The goods are released on payment of a penalty calibrated to the tax payable on the goods where the owner comes forward to pay, and on a higher measure, linked to the value of the goods, where the owner does not. Where the officer forms the view that the movement was made with intent to evade tax, Section 130 permits confiscation of the goods and the conveyance, with an option to pay a fine in lieu of confiscation in addition to the tax and penalty. The courts have released goods in many cases where the lapse was documentary, an expired bill or an incorrect vehicle number with the tax documents in order, and have sustained action where the bill was a cover for unaccounted goods.

Q: What is an input service distributor under GST?

An input service distributor is an office of a supplier, registered separately as such, that receives invoices for input services used by more than one registration of the same business and distributes the tax credit on those services to the registrations that used them. Headquarters commonly contract for legal, audit, software, advertising and brand services consumed across every branch; the invoice reaches the head office, while the benefit is spread across the states in which the business operates. Section 20 of the central Act requires the distributor to issue a prescribed document and to allocate the credit by a turnover-based formula set out in the rules, central and state tax credit being distributed as such where the recipient is in the same state and as integrated tax credit across state lines. Amendment has made this route compulsory for the distribution of common third-party service credit, ending the practice of head offices simply cross-charging branches for such services.

Q: What interest applies when GST is paid late?

Interest is the price the Acts put on delay, and it runs by the day. Under Section 50 of the central Act, a person who fails to pay the tax within the period prescribed pays interest at eighteen per cent per annum for the period of delay in the ordinary case; the interest is charged on the net cash liability, the part of the liability that remains after the credit in the electronic ledger is used, a restriction the Finance Act, 2022 made retrospective to the tax’s commencement. Interest also attends input tax credit that was wrongly availed and utilised, a charge that was aligned to eighteen per cent with retrospective effect after a period in which a higher rate was notified. Interest is compensatory in the courts’ reading, not penal, and it is not waived by the settlement of the tax itself; late fees for delayed returns under Section 47 are a separate and smaller charge, computed per day of delay subject to caps fixed by notification.

Q: What is a GSTIN and how is the number structured?

The GSTIN, the Goods and Services Tax Identification Number, is the fifteen-character registration number issued to every registered person under the central Act and the state Acts, one for each state in which the person is registered. Its structure encodes the design of the tax. The first two digits are the code of the state of registration. The next ten characters are the person’s permanent account number under the income tax law, so that the GST and income tax identities of a business are joined at the root. The thirteenth character records the number of registrations the same person holds within that state on the same permanent account number, the fourteenth is a fixed character, and the fifteenth is a check digit that lets systems detect a mistyped number. A business operating in twelve states therefore holds twelve GSTINs sharing a common core, each with its own returns, ledgers and administration.

Q: How is a supply between two branches of the same business in different states taxed?

It is taxed as a supply between strangers, because the Acts treat each registration of a person as a distinct person. Section 25(4) of the central Act provides that a person with more than one registration, whether in one state or in several, is treated as distinct persons for each registration, and Schedule I treats supplies between distinct persons in the course or furtherance of business as supplies even though no consideration passes. A stock transfer from a company’s Maharashtra branch to its Karnataka branch is therefore an inter-state supply bearing integrated tax, invoiced by the sending registration and taken as credit by the receiving one, so the tax is neutral in money and significant only in cash flow and record. The same logic taxes head office services supplied to a branch. Valuation is governed by the valuation rules, which accept the invoice value where the receiving registration is entitled to full credit, confining valuation disputes to cases where its credit is blocked or exempt.