The Disaster Audit Bengal Didn't Want You to Read (Cyclone Amphan)

A Calcutta High Court bench ordered the Comptroller and Auditor General to audit West Bengal's Cyclone Amphan relief spending in December 2020. The CAG finished the job and handed its findings to the state government in February 2022. Almost nobody outside a small circle of officials saw that report for the next four and a half years. It finally reached the floor of the West Bengal Assembly on 25 July 2026, tabled by a government the Trinamool Congress no longer runs.

What Happened: Cyclone Amphan and the Fight Over Its Relief Money

Amphan made landfall near Sagar Island on 20 May 2020 as a super cyclonic storm, winds touching 133 kmph. It killed 99 people and tore through 16 of West Bengal's 23 districts, hitting North and South 24 Parganas hardest. By the government's own count, nearly 60 per cent of the state's population felt some effect, and losses ran across housing, agriculture, fisheries, forestry, power, irrigation and health infrastructure.

The state's Amphan memorandum put total damage at ₹1,02,442 crore and asked the Centre for ₹35,018 crore in central assistance. What came through was far smaller: the Inter-Ministerial Central Team recommended ₹2,500.29 crore, a sub-committee of the National Executive Committee revised it to ₹2,707.77 crore, and the Ministry of Home Affairs eventually released ₹2,250.28 crore from the National Disaster Response Fund, split between May and November 2020.

Almost as soon as the money started moving, so did the complaints. Several petitions landed in the Calcutta High Court alleging corruption and diversion of funds meant for Amphan relief, and on 1 December 2020 a division bench directed the CAG to conduct a full financial and performance audit within three months. The state government pushed back, filing a recall petition arguing the audit was premature. The Court disposed of that petition in January 2021, softening the deadline to a "reasonable time" but leaving the audit order intact. Auditors worked in two phases, February to March 2021 and mid-July to mid-September 2021, delayed further by the 2021 state election and the Covid-19 wave. Findings went to the state government and the Ministry of Home Affairs in February 2022. Then, for four and a half years, they went nowhere near the Assembly, until a new government tabled the report on 25 July 2026, weeks after the BJP took power in the state's 2026 election.

Finding One: A Damage Bill Nobody Could Verify

A gap between claim and payout isn't unusual on its own; NDRF norms are notoriously stingy and every state inflates its ask. What's unusual is what the CAG found when it asked to see the workings behind West Bengal's ₹1.02 lakh crore damage figure. Nothing. No district-wise damage reports, no norms explaining how the number was calculated. The department that produced the figure could not substantiate it to its own auditor.

Of the ₹1.02 lakh crore in damage West Bengal claimed to justify central assistance, auditors could not find a single district-level report to back the number. It was never verified, not even by the department that produced it.

Finding Two: The House Grant Math That Doesn't Add Up

House Building grant was West Bengal's largest single relief item: ₹1,948.51 crore paid to 21,03,384 beneficiaries across 16 districts between May and December 2020, at ₹20,000 for a fully damaged house and ₹5,000 for a partly damaged one. Compare that payout against the state's own memorandum, which had assessed 18,92,946 houses as fully or severely damaged. Only 5,69,597 households actually received the full grant, a shortfall of nearly 70 per cent. Meanwhile the partly-damaged category, assessed at 4,99,558 houses, ended up paying out to 15,33,787 beneficiaries, more than three times the assessed number.

Something happened between assessment and disbursement, and it wasn't a natural correction. In Purulia, the state's westernmost and least cyclone-exposed district, every single assessed house was recorded as fully damaged. Not one partly damaged house in the entire district. In South 24 Parganas, one of the two worst-hit districts, the fully-damaged count came in 77.21 per cent higher than what later field verification supported. Read together, these numbers point one way: the initial damage assessment was inflated to justify a bigger central ask, and the more careful verification that followed the money never squared with it.

The disbursement itself wasn't equal either. In the first four allotment orders, issued between 28 and 30 May 2020, before the state had even fixed a rate for partly damaged houses, ₹1,000 crore went out flat at ₹20,000 per house to five lakh beneficiaries. That's 51 per cent of the year's entire HB grant handed to roughly a quarter of the eventual beneficiary pool, days after the cyclone, before verification norms were tightened. The remaining three-quarters of beneficiaries split the balance ₹906.07 crore. On top of that, auditors sampling six districts found ₹251.38 crore paid to 3,12,088 people who didn't meet the state's own income cutoff for the grant, a monthly income above ₹2,500.

Note: In North and South 24 Parganas, the two worst-hit districts, authorities withdrew the entire ₹1,383.23 crore HB grant allocation from the treasury and parked it in commercial bank accounts instead, a direct violation of the West Bengal Treasury Rules, 2005. ₹34.68 crore of that sat unused in South 24 Parganas as of January 2022, yet utilisation certificates were filed for almost the full amount regardless.

Finding Three: Fabricated Voter IDs and Recycled Beneficiary Lists

The CAG's most damning findings weren't about assessment. They were about who got paid, and how many times. Cross-checking HB grant payments across six districts, auditors found 18,891 cases where the same bank accounts, belonging to 9,226 beneficiaries, received the grant more than once, in one case nine times over. That's ₹18.07 crore paid out with no duplication check in place, despite a department order eight months earlier explicitly requiring district authorities to run one.

The photographic evidence tells the same story. Every HB grant claim required a photograph of the damaged house with its owner, so auditors pulled 3,70,491 such photographs across five of the six districts (South 24 Parganas produced none at all). Of a sample of 31,457 photographs examined, 3,126 turned out to be identical images filed against different beneficiaries. In one case from Mahisadal block in Purba Medinipur, the same photograph backed two separate "fully damaged" claims, one filed under Moriam Bibi's name and the other under Rejak Dutta's. Auditors couldn't put a rupee figure on how much this cost, because the photographs were never cross-linked to Form-B serial numbers, but three district administrations blamed it on clerical error. None of them produced records showing the error was ever fixed.

Crop assistance fared no better. Under the Krishak Bandhu Assured Income Scheme, farmers are supposed to be uniquely identified by their Electoral Photo Identity Card (voter ID) number, and the portal is designed to reject duplicate enrolments. Auditors found 2,697 beneficiaries who enrolled multiple times anyway, using either altered EPIC numbers or numbers belonging to someone else entirely. Of 5,406 transactions tied to these accounts, 3,011, worth ₹46.94 lakh, used EPIC numbers that don't exist anywhere in the Election Commission's database. Someone typed in fabricated voter identities to draw disaster relief money, and the system let it through.

Then there's the betel vine (paan boroj) payout. Horticulture assistance worth ₹2.82 crore went to 5,627 farmers using a beneficiary list built for an entirely different cyclone, Bulbul, which struck seven months before Amphan, with no fresh application or field check for the newer disaster. 1,727 of those same farmers then applied separately for Amphan assistance and were paid again. And while the state's own memorandum showed betel vine accounted for only 4.67 per cent of the total damaged horticultural area (8,704 of 1,86,362 hectares), betel vine growers were the only horticulture beneficiaries to receive any input subsidy at all: ₹81.29 crore of it, ₹65.62 crore more than the assessed damage for that crop alone justified. Farmers growing vegetables, fruit, and flowers, whose fields made up 95 per cent of the damaged area, got nothing.

It wasn't only households and farmers. WBSEDCL initially claimed ₹1,235.25 crore in cyclone damage to its power infrastructure. A later, documented assessment put the real figure at ₹240 crore, just 19 per cent of the original claim.

Finding Four: A System Built Without Safeguards

It would be easy to read this audit as a straightforward corruption story and stop there. That undersells what went wrong. West Bengal ran a relief operation covering nearly 47 lakh beneficiaries with no mandatory Aadhaar or unique identifier, no standard application form (departments, even individual blocks, used different formats, some handwritten), no register logging incoming applications, and no audit trail connecting an application to its verification to its payment. Supervising officers appointed specifically to catch these problems visited the field in June 2020 and reported finding no anomalies at all. Every irregularity documented above happened after that clean chit.

That's the real lesson for anyone working in disaster governance, not just in Bengal. A relief system without a de-duplication mechanism will produce duplicate payments whether or not anyone intends fraud. A system without a uniform, verifiable application process will always be exploitable by whoever's local network is best positioned to game it, regardless of which party is in office. Amphan is not the first Indian disaster relief operation with these problems. Without structural fixes, it won't be the last.

To be fair, the Trinamool Congress has pushed back hard on the political framing, arguing that a compliance audit documenting deficiencies isn't the same as a finding of fraud, and that no criminal wrongdoing has been established against any individual. That distinction matters, and it's technically true: the CAG report is an audit, not a chargesheet. But an administration doesn't need a chargesheet to have failed the people it was supposed to serve. Not fixing a duplication check for a year and a half after ordering one, not requiring a single verifiable ID across lakhs of transactions, not producing a single damage report to justify a ₹1.02 lakh crore claim: these were administrative choices. Someone made them.

Conclusion: Table the Report, Then Fix the System

West Bengal's new government has spent the weeks since 25 July naming the previous one. That's the easy part. The CAG's 2022 recommendations, a unified application format, mandatory Aadhaar-linked de-duplication, a trackable single-window system, and recovery of every rupee paid to ineligible or duplicate beneficiaries, are sitting in the same report that just got tabled. None of them require an election to implement.

Table the report. Then fix the system.

Recovering ₹18.07 crore in duplicate HB payments and correcting a ₹65.62 crore horticulture subsidy skew is a start, not an end. West Bengal's next cyclone won't wait for the political fight over this one to finish. The state needs a working, ID-linked disaster relief architecture in place before it does.

Further Reading

• Inside The CAG Audit Of West Bengal's Amphan Relief — Outlook India
• CAG Flags Major Irregularities In Bengal's Post-Amphan Relief — The Federal
• CAG Flags Inflated Amphan Damage Claim, Duplicate Relief Payments In Bengal — United News of India
• Special Session To Discuss CAG Reports, Says Swapan Dasgupta — The Pioneer
• 2026 West Bengal Legislative Assembly Election — Wikipedia, for background on the change of government behind the report's tabling