energy-policy

The Ethanol Policy Is Right. The Rollout Is Not.

India hit 20 percent ethanol blending five years early, which is a genuine execution win. Then the government's own Attorney General called it an experiment in open court. The fuel is mostly fine. The way it was shipped to 300 million vehicle owners is the actual problem.

The Blog

In June 2026, India's Attorney General stood in the Supreme Court and said that <cite>"20 per cent ethanol blending is something that the government is experimenting with"</cite>, adding that results would be in by next year. His office clarified within hours that he was talking about a commercial dispute, not the national programme.

Maybe. But the clarification didn't land, because the sentence described the situation too accurately. Every petrol vehicle in India is running on E20. Nobody was asked. Nobody was told. And the study everyone keeps citing has not been published.

That's the story. Not the fuel. The fuel is mostly fine.

Where the government is right

Start here, because the criticism only works if you concede the wins first.

Ethanol is not exotic. Brazil runs E27 as standard petrol and has for years. Large parts of the US run E10 with E15 and E85 available. Henry Ford designed the Model T to run on it. Anyone claiming ethanol in petrol is inherently dangerous is arguing against a century of global practice.

The execution was genuinely impressive. Blending went from around 1.5 percent in 2014 to 20 percent in 2025. That's roughly 13x in eleven years, faster than any other ethanol-using nation has managed. The progression was staged: 10 percent in ESY 2021-22, 12.1 percent, 14.6 percent, 19.2 percent, then 20. It didn't happen overnight, whatever the discourse says.

The supply side is real, not a paper target. Installed ethanol capacity is now around 20 billion litres against roughly 11 billion needed to sustain E20. Dedicated ethanol plants with long-term offtake agreements and tripartite bank financing were the mechanism, and they worked. Building demand certainty first and letting capital follow is textbook industrial policy, executed competently.

The energy security logic holds. Roughly a fifth of every litre in your tank is now priced in rupees under an administered price, not in dollars against Brent. When West Asia goes sideways, that fifth doesn't move. India's retail fuel price increase over the last four years is the lowest among its neighbours and most major economies. Ethanol is not the only reason, but it is a reason.

The feedstock pivot was the right correction. Sugarcane alone could never have gotten India to E20 without wrecking sugar availability, and the government figured that out. Maize went from about 6 percent of ethanol production in ESY 2022-23 to roughly half by ESY 2024-25. Grain-based feedstock now makes up around two-thirds of output. That was a hard pivot done fast.

The catastrophic-failure story is not holding up. Maruti Suzuki serviced about 2.84 crore vehicles in FY 2025-26, including roughly 1.5 crore that were never certified for E20, and reported no ethanol-linked corrosion or abnormal component wear. Hero MotoCorp says the same. If E20 were eating engines, warranty claims would show it. They don't. The "my car seized because of E20" genre is mostly not survivable contact with data.

Where the government is wrong

Now the other column.

No label. This is the whole ballgame. Nobody was demanding that every pump stock three grades of petrol. The logistics argument against that is legitimate: over a lakh retail outlets, shared pipelines, shared depots. Fine. But a sticker on the dispenser and a line on the invoice cost nothing, break no supply chain, and would have defused most of this. The government chose not to. A PIL filed in July 2026 is now asking for exactly that, having explicitly declined to challenge the blending policy itself. When your critics have narrowed their demand to "please tell us what you sold us," you have lost the argument on process.

The study is not public. The government cites ARAI, IOCL, IIP, SIAM and an inter-ministerial committee. Those studies exist. The reports have not been released. You cannot run "trust the science" and "we're not publishing the science" simultaneously. In any regulated industry I've worked in, an unpublished validation report is treated as no validation report. That standard should not soften because the entity is a ministry.

The mileage numbers don't reconcile, and the physics says something interesting. Here's the range currently in circulation: ARAI says 2 to 6 percent. The government's own FAQ says 3 to 5. SIAM says 2 to 4. Carmakers privately say 7 to 8. Autocar India's road testing found 3.8 percent on a new Dzire, 5.2 percent on a BS4 Polo GT, 12.4 percent on an older Dzire, and 12.6 percent on a Creta N Line.

Run the arithmetic. Petrol carries about 32 MJ per litre, ethanol about 21. E20 therefore holds roughly 93 percent of the energy of pure petrol. The calorific floor for mileage loss is around 6.75 percent.

Two things follow. First, ARAI's 2 to 6 percent is better than the calorific floor, which is credible: E20's higher octane lets some engines run more efficiently and claw back part of the loss. That's a point in the government's favour and it deserves saying. Second, a 12.6 percent loss cannot be explained by fuel chemistry at all. Something else is happening in those vehicles, whether it's engines operating outside their fuel-trim calibration range, degraded components, or confounded measurement. Any of those is worth knowing. None of them is knowable from a single national band. The government should be publishing model-wise numbers and isn't.

The consumer pays twice, and NITI Aayog said not to do this. The 2021 NITI Aayog roadmap explicitly recommended that blended fuel be priced below regular petrol to compensate for the reduced calorific value. That recommendation was ignored. You now buy a litre with less energy in it at the same price, and burn more litres to cover the same distance. The government's defence is honest as far as it goes: maize ethanol is procured at around Rs 71.86 per litre, above the base price of petrol at 70-dollar crude, so E20 genuinely costs more to make right now. That explains why it isn't cheaper. It does not explain why the excise relief NITI recommended never materialised.

The forex number is gross, not net. Official figures put cumulative savings at about Rs 1.97 lakh crore since ESY 2014-15, with roughly Rs 1.66 lakh crore paid to farmers. Those are gross displacement figures. They don't net out interest subvention on distillery capex, the GST concession, FCI rice sold to distilleries well below its procurement cost, or the extra litres consumed because of the calorific loss. And on displacement itself: E20 replaces 20 percent of volume, but only about 14 percent of energy. The honest crude-displacement number is 14, not 20. That's arithmetic, not opinion, and it should be in the published accounting.

Rice-to-ethanol is the weakest link and everyone knows it. Producing a litre of ethanol from rice takes roughly 10,790 litres of water. Maize is about 4,670. Sugarcane about 3,630. Rice is grown heavily in Punjab and Haryana, where the Central Ground Water Board classifies over 75 percent of districts as overexploited. FCI buys that rice at around Rs 42 a kilo on subsidised inputs and sells it to distilleries at Rs 22 to 23. Ashok Gulati has called the economics irrational once hidden subsidies are counted, and it's hard to argue with him. The buffer-stock defence is real: FCI rice stocks have run around three to four times the required buffer, and grain rotting in a warehouse helps nobody. But "we have too much of it" is a reason to fix procurement, not a reason to pour a groundwater-depleted crop into a fuel tank.

The politics have contaminated the evidence. The Attorney General told the court there was a lobby behind the PIL. The PIB fact-check attributes concerns to vested interests. Nitin Gadkari has called the campaign politically motivated and asked critics to produce a single damaged vehicle, while opposition parties have raised conflict-of-interest allegations over his family's ethanol-linked business interests. Meanwhile, in a Parliamentary Standing Committee meeting in late July 2026, officials reportedly told MPs that vehicle problems are a recent phenomenon and that tests are underway to find out why.

Read those two positions together. Either there is nothing to investigate, or there are tests underway. Both statements are being made by the same government in the same month.

Who wins and who pays

graph TD
    A["Feedstock\nSugarcane, maize, FCI rice"] --> B["Distilleries\n~20 bn litre capacity"]
    B --> C["OMCs procure at administered price\nRs 57.97 to Rs 71.86 per litre"]
    C --> D["Blended to E20 at depots"]
    D --> E["100,000+ retail outlets\nNo label. No E10 option."]
    E --> F["Consumer"]
 
    A -.-> I["FARMERS\nRs 1.66 lakh cr paid\nCane arrears down"]
    B -.-> H["MILLS AND DISTILLERS\nGuaranteed offtake\nInterest subvention\nImport protection"]
    C -.-> G["EXCHEQUER\nRs 1.97 lakh cr forex saved\ngross, not net"]
 
    A -.-> L["GROUNDWATER\n10,790 L water per litre\nof rice ethanol"]
    F -.-> J["CONSUMER\nSame price\n2 to 12 percent less mileage"]
    F -.-> K["OLDER VEHICLE OWNERS\n~80 percent of petrol fleet\nnever certified for E20"]
 
    classDef benefit fill:#14532d,stroke:#052e16,color:#ffffff
    classDef cost fill:#7f1d1d,stroke:#450a0a,color:#ffffff
    classDef chain fill:#334155,stroke:#0f172a,color:#ffffff
    class G,H,I benefit
    class J,K,L cost
    class A,B,C,D,E,F chain

The chain works. The distribution is the argument.

What a competent version looks like

None of this requires reversing the policy. Reverting to E10 now would strand about a lakh crore a year of financed distillery capacity and punish everyone who built to a published national target in good faith. That would be worse.

Six fixes, none of them expensive:

1. Publish the ARAI report. Raw data, test protocol, per-model results. Today.
2. Label the dispenser and the invoice. Ethanol percentage, printed. This is a sticker and a line of code in the POS software.
3. Ship a compatibility database. Model-wise, with a preventive replacement schedule for elastomer components and a capped-price parts programme. ARAI's own durability work found metals and plastics fine but older rubber hoses, seals and gaskets potentially degrading faster. That is a real, bounded maintenance cost. Bound it publicly. Maruti's roughly Rs 7,000 upgrade kit is one OEM doing what all of them should be required to do.
4. Price per unit of energy, or give the excise relief NITI recommended. Pick one. Charging full price for 93 percent of the energy is the single most avoidable own-goal in this programme.
5. Water-index the feedstock policy. Site distilleries against aquifer stress maps. Cap the rice pathway in overexploited blocks. Accelerate 2G ethanol from residues, which needs no additional land, crop or groundwater and reduces stubble burning as a bonus.
6. Publish the exit criteria before going to E27 or E30. BIS notified standards for E22 through E30 in May 2026. The government has since said no decision has been taken to move beyond E20, which is the correct call. Before that changes, say publicly what evidence would justify the next step and what evidence would stop it.

The actual lesson

I spend my working life in regulated systems. The rule there is not "don't ship." The rule is: instrument it, disclose it, and have a rollback path before you go to 100 percent of traffic.

India shipped a fuel change to roughly 300 million vehicles. No feature flag. No canary. No telemetry published. No opt-out. No stated rollback criteria. Then it defended the deployment by asserting that the monitoring, which nobody can see, shows no problems.

The ethanol is fine. The blend is defensible. The energy security case is real and the farmer economics are better than the critics allow.

The governance is the failure. And the maddening part is that the fixes are a sticker, a PDF, and a database.