1. Why blend percentage moves the state tax number at all

Petrol sold at the pump in India carries two state-relevant levies: central excise (the Union’s cut, currently around ₹19.90/litre) and state VAT, charged ad valorem on top of the pre-tax price. VAT is where states make their money on fuel, and headline rates vary enormously — from about 13% in Tamil Nadu and 13.7% in Gujarat to 29–31% in Karnataka, Kerala, Andhra Pradesh and Madhya Pradesh, and over 35% in Telangana.

Ethanol procured by the OMCs for blending is a different animal entirely: it is taxed under GST, not VAT, at a flat 5%, split roughly evenly into 2.5% CGST and 2.5% SGST. At the pump, the blended fuel is still sold and taxed as “petrol” at the state’s full VAT rate — the retail tax treatment hasn’t been redesigned around the blend. But economically, a rising share of every litre dispensed is, on the input side, 5%-GST ethanol rather than VAT-rated petrol. The state didn’t collect VAT on that ethanol fraction when it was purchased by the OMC; it only gets the SGST sliver.

That mismatch is the entire story of this article: as blend percentage rises from E10 toward E20 and beyond, a growing slice of the fuel pool sits in the low-tax GST lane instead of the high-tax VAT lane, and the gap between “VAT the state would have collected on that volume as pure petrol” and “SGST the state actually collects on that volume as ethanol” widens — state by state, in proportion to that state’s own VAT rate and its own petrol consumption.

2. The production map: sugarcane states vs. grain states

India’s ethanol supply comes from two feedstock families, and which one a state leans on shapes its distillery investment case, not its SGST take:

• Sugarcane / molasses-based — the traditional route, concentrated in the cane belts: Maharashtra, Uttar Pradesh, Karnataka, Tamil Nadu, Gujarat and Bihar host most of India’s C-heavy and B-heavy molasses distilleries built off existing sugar mills.
• Grain-based — maize and damaged/surplus FCI rice, the fastest-growing segment as the sugar-diversion cap bites. Uttar Pradesh has been adding grain capacity fastest (roughly 50 crore litres/year of new capacity in 2025 alone, much of it around the Gorakhpur GIDA industrial cluster), alongside Punjab, Bihar and Madhya Pradesh.

Installed national capacity stood at roughly 2,000 crore litres/year for ethanol supply year 2025–26 (CareEdge, May 2026), with about 400 crore litres/year of further additions expected by FY27. The Department of Food & Public Distribution’s sanction register is far larger — roughly 4,530 crore litres across 1,212 approved projects — meaning built capacity still lags what has been cleared on paper.

Capacity is not evenly distributed relative to demand. Regional modelling (built off the CareEdge May-2026 dataset) puts Maharashtra around 277 crore litres/year in surplus — its distillery base outproduces its own state fuel-blending requirement — while Tamil Nadu runs roughly a 77 crore litre/year deficit, importing ethanol from surplus states to meet its blending obligation. Surplus states like Maharashtra therefore matter more to the national supply chain than to their own SGST line, precisely because of the destination-based GST point above: the ethanol they distil and ship out earns SGST for whichever state burns it as fuel.

Feedstock, not steel, is increasingly the binding constraint as blends rise. Molasses supply is capped by cane-diversion politics and the sugar-diversion cap; the marginal litre needed to push blends past E20 is increasingly grain — principally maize, the costliest feedstock slab in per-litre terms — which ties ethanol economics to poultry and animal-feed markets. A parallel route via FCI surplus rice covers only an estimated ~211 crore litres/year at typical lifting rates (about 3.9 percentage points of national blend), and carries real policy risk: rice-for-ethanol allocation was suspended once before, in July 2023.

3. Modelled SGST impact, state by state, across E10–E30

The table below scales India’s current tax architecture (5% GST on ethanol, 2.5% notionally to the state; full state VAT foregone on the petrol volume that ethanol displaces) across four blend levels. It is built from PPAC Ready Reckoner FY2025–26 H1 state-wise petrol (MS) sales volumes and headline state VAT rates, with a fixed ₹78/litre pre-VAT petrol base and ₹60/litre ethanol procurement price as editable levers — not certainties.

3.1 National picture

E10 is derived by linear scaling of the E20 model (ethanol volume moves in exact proportion to blend fraction, all else equal) — it is not separately re-derived from PPAC tables. For scale: the Union also forgoes roughly ₹10,760–₹32,270 crore/year of central excise across the same E10→E30 range on the same displaced volume — a bigger number than the state VAT loss at every blend level, but one the Centre absorbs, not the states.

3.2 Top 15 states by net loss, today’s E20 blend

Two forces set a state’s exposure: its VAT rate (Telangana, Kerala, Andhra Pradesh, Karnataka, Rajasthan and Madhya Pradesh all sit near or above 29%) and its petrol volume (Uttar Pradesh, Maharashtra, Tamil Nadu and Gujarat are simply large fuel markets). States that score high on both — Maharashtra and Karnataka above all — carry the largest absolute exposure. Low-VAT, high-volume Tamil Nadu and Gujarat are a useful counterpoint: big markets, but their light VAT rates cap the loss relative to their size.

3.3 How the loss scales as blend rises, top 8 states

The loss scales almost exactly with blend fraction — doubling volume from E10 to E20 roughly doubles the shortfall, and pushing to E30 roughly triples it against E10. No state escapes this trend under the present tax split; blending mandates simply make each state’s existing exposure bigger in direct proportion.

3.4 A hypothetical: what if states got a dedicated ethanol SGST?

One policy idea worth modelling — explored in the companion analysis this article draws on — is a dedicated state SGST levy on ethanol, layered on top of the existing 5% GST split, funded from the roughly ₹15/litre cost gap between cheaper ethanol and the higher-taxed petrol base it replaces. Because that gap comfortably covers a levy of a few rupees per litre, such a rate could in principle be added without moving the pump price. This is a modelled policy construct, not an enacted rate or a GST Council decision.

Even a full 5% dedicated SGST recovers only 3–18% of the VAT states forgo, depending on rate and blend — it offsets part of the shortfall in Section 3.1, it does not close it. Full parity recovery would need a rate near 27%, which stops being a modest, price-neutral levy and becomes a real tax-policy fight.

3.5 Where that hypothetical levy would land: top 5 states

4. Winners and losers

Under the tax architecture that exists today — petrol VAT plus a plain 5% GST on ethanol, no dedicated ethanol SGST — there are no state winners. Every one of the 36 states/UTs matched in this model shows a net loss at every blend level, because the SGST half-share of ethanol’s 5% GST is structurally smaller than the VAT that same volume would have earned as petrol, at every observed headline VAT rate from 13% to over 35%. Higher blending is, in this narrow fiscal sense, a headwind for every state simultaneously — it just isn’t the same size of headwind everywhere.

The one lever that meaningfully changes this picture, at least partially, is a dedicated ethanol SGST of the kind modelled in Section 3.1 — but even at 5%, no state actually crosses into net-positive territory; it merely reduces how negative the number is. As things stand, higher blending is a genuine centre–state fiscal transfer question as much as an energy-policy one: the Union’s excise loss is larger in absolute terms than the states’ VAT loss at every blend level, but it is the states, without a redesigned tax split, that see almost none of the ethanol-side GST offset their own foregone VAT.

5. Segment-by-segment cost-per-km reality check

The sections above model what higher ethanol blending does to state revenue. None of it touches the other side of the transaction: what a household actually pays per kilometre, and how that varies by the vehicle they drive and the fuel they choose. That's a different model — this blog's vehicle_fuel_mileage cost-of-ownership repo, built independently from ARAI/SIAM fuel-efficiency declarations and OMC-published city fuel boards — but it belongs alongside the state-tax numbers above, because the fiscal argument only matters if it changes what a driver actually experiences at the pump or the charging point.

5.1 Cost per km, by vehicle segment and powertrain

Two things fall out of the segment breakdown that the national-average figures alone hide. First, the EV-vs-CNG gap widens going up the segment ladder in absolute terms (₹0.46/km on a two-wheeler, ₹2.15/km on a compact SUV) even though EV stays cheaper at every size — the bigger and thirstier the vehicle, the more a fuel-cost switch is worth in rupee terms. Second, CNG's lead over E20 petrol is remarkably stable in relative terms across segments — roughly a 55–58% saving per km whether it's a two-wheeler or a compact SUV — because both fuels scale with the same underlying engine displacement and load.

5.2 How much the city you live in changes the answer

The petrol/km spread across these six metros is almost entirely a VAT effect — the same ~14% spread in pump price (₹101.5 in the cheapest city to ₹115.7 in the costliest, per the model's headline finding) that this article's own state-by-state VAT map explains for petrol shows up again here, just expressed per kilometre instead of per litre. CNG and EV costs move less between cities because electricity tariffs and CNG city-gate prices vary less than VAT-loaded petrol/diesel does — which is itself a reason a driver's fuel-switching incentive is stronger in high-VAT states than low-VAT ones.

5.3 Payback: how fast the switch pays for itself

CNG payback clusters tightly around 1.7–1.8 years across every segment except two-wheelers (under a year), which is the single best explanation, alongside the sales-mix data in 5.4 below, for why CNG rather than EV is the default fuel-switch for a private car buyer today: it's a materially faster payback than EV at every segment, including two-wheelers, on a much smaller upfront premium. EV payback only gets competitive with CNG at high annual mileage (the model's "High" 25,000 km/yr scenario, not shown here, pulls hatchback EV payback down to ~3.4 years) — consistent with EVs suiting high-utilisation fleets and commercial runs better than average private ownership.

5.4 What this looks like in actual sales, not just modelled cost

The payback math above is a prediction; India's Vahan registration data and FADA's retail figures are the check on whether buyers are actually following it. This blog's companion piece on the EV market pulls those numbers together (CEEW-GFC Electric Mobility Dashboard on Vahan registrations, JMK Research, IESA, FADA FY26):

• National EV penetration: 8.5% of all FY26 vehicle sales, up from 6.8% in FY24 — growing, but still a single-digit share of the market.
• Two-wheelers carry the EV volume but not the penetration: e-2Ws are 58% of all EV units sold (1.4 million), yet only 6.5% of all two-wheelers sold are electric — the segment with the fastest CNG/EV payback in Section 5.3 is, paradoxically, the one where EV adoption lags its own economics most, likely on upfront price and financing access rather than running cost.
• Three-wheelers already crossed the tipping point: over 60% of all 3Ws sold nationally are now electric, with L5 passenger EV share at 30.5% — the segment that runs the highest daily mileage, where running-cost savings dominate total cost of ownership fastest, has electrified with the least subsidy dependence of any segment.
• CNG, not EV, is the default fuel-switch for private cars: FADA's FY26 data puts CNG at roughly 22% of passenger-vehicle sales nationally — well ahead of EV's 8.5% economy-wide penetration — which lines up with Section 5.3's finding that CNG's payback beats EV's at every car segment under average usage.

Sales-mix figures sourced to CEEW-GFC Electric Mobility Dashboard (Vahan registrations), JMK Research, IESA and FADA FY26 retail data, via this blog's EV market piece; note the EV and CNG shares above are not strictly comparable (CNG's 22% is passenger-vehicle-specific; EV's 8.5% spans all vehicle categories including two- and three-wheelers, where EV penetration runs far higher) — treat both as directional market-mix indicators, not a single like-for-like race.

5.5 The ethanol penalty, translated to a per-vehicle number

This is the consumer-side mirror of the state revenue shortfall computed in Sections 3–4: the same E20 mileage penalty that costs states an estimated ₹17,863 crore/yr in net foregone VAT (Section 3.1) also costs the median petrol hatchback owner about ₹4,000/yr directly, before any tax-policy change at all — a cost this article's SGST modelling does not offset, because none of the proposed dedicated-levy scenarios in Section 3.4 return money to consumers; they only change which government (state vs centre) collects it.

6. What this can’t tell you

• No official dataset exists. State finance departments, the GST Council and the CAG do not publish a state-wise, blend-attributed split of SGST collected on ethanol versus VAT foregone on displaced petrol. Every figure in Sections 3–4 is reconstructed from national PPAC tables plus modelled assumptions — not sourced from a state budget document or treasury filing.
• Headline VAT only. The VAT rates used are the headline percentage from the PPAC Ready Reckoner; several states layer on fixed cesses or “whichever is higher of % or ₹/L” floor clauses not captured here, so real losses in those states are likely understated, not overstated.
• Interstate ethanol movement isn’t modelled. Section 1’s destination-based GST point — that a surplus state like Maharashtra doesn’t automatically keep the SGST on ethanol it ships to a deficit state like Tamil Nadu — is described qualitatively, not quantified. The actual IGST/settlement mechanics of interstate ethanol stock transfers would need a genuinely different model.
• Fixed prices, not a market. A constant ₹78/litre pre-VAT petrol base is used throughout; the ethanol price is ₹60/litre for the E10–E20 rows but rises across the E22–E30 rows and the top-5 state table — roughly ₹61.5/litre at E22 climbing to about ₹68.1/litre at E30, a grain-slab-weighted price that increases with blend level rather than a single flat figure, reflecting the costlier feedstock mix needed to push past E20. Real ethanol procurement prices vary by feedstock slab — C-heavy molasses, B-heavy molasses, damaged grain, maize — and move with OMC differential-pricing decisions year to year.
• The dedicated SGST is hypothetical. The 1–5% ethanol-SGST levy in Section 3.4 is a modelled policy proposal explored for this analysis. It is not passed law, not a GST Council decision, and would require a Council amendment (or a separate legal route) to exist at all.
• Production capacity ≠ production data audited here. State feedstock and distillery-capacity figures (Section 2) draw on industry reporting (CareEdge, DFPD sanction registers) rather than a verified state-wise capacity census; treat the sugarcane-belt/grain-belt characterisation as directional, not a precise inventory.
• No fleet or engineering constraints. Blend-wall compatibility, E20-rated vehicle penetration, and actual DFPD ethanol offtake against target are outside the scope of this state-tax model; they were covered, at the national level, in the companion OMC-books article.

Verifying any of this against reality would require, at minimum: state-wise VAT/SGST collection data broken out by fuel type and blend fraction from state commercial-tax departments; the actual (not headline) VAT schedule including cesses and floors; and a GST-settlement study of interstate ethanol movement. None of that is public today, as far as this research could establish.