Finding 01 · Data quality · FY2025-26
One tag, two meanings: the unit problem in India's first machine-readable climate filings
In the first year of mandatory machine-readable (XBRL) BRSR filings, 22 of the 49 Nifty-50 companies whose FY2025-26 filings we parsed tagged their Scope 1 greenhouse-gas emissions in "MtCO2e" — million tonnes of CO2-equivalent. For 21 of them, the figure is actually in tonnes — a million-fold labelling error, confirmed against each one's own PDF filing — while one company, Tata Steel, uses the same tag correctly for genuine megatonnes.
How 49 Nifty-50 filings tag Scope 1 emissions
Unit reference (unitRef) on the TotalScope1Emissions fact, FY2025-26 XBRL
Source: NSE BRSR XBRL filings, parsed 15–19 Aug 2026. Gold = the tag under examination.
If you trust the tag, Reliance's Scope 1 is
36.35 trillion tonnes
≈ 700× the entire world's annual greenhouse-gas emissions
What the filing actually reports
36.35 million tonnes
matches the figure in Reliance's own PDF filing
What we found
Each XBRL fact carries a machine-readable unit reference. Read literally, 18 of the 22 MtCO2e figures are physically impossible — several would individually exceed global annual emissions. Checking every flagged company against the units stated in its own PDF filing:
- 21 companies confirmed mistagged — their PDFs state "Metric tonnes of CO2 equivalent" for the same figures the XBRL tags as megatonnes. Eight match digit-for-digit (Dr Reddy's 1,16,841; ONGC 8,379,783.82; BEL 8,109.76; IndiGo 10,419,376.30; Jio Financial 1.73; Apollo Hospitals 13,167.90; Eternal 25.62 — matched in both the current and prior year; UltraTech Cement 8,88,84,246, one of India's largest emitters), and Coal India matches exactly as the sum of the two entity-level tables in its PDF.
- 1 company confirmed correct — Tata Steel's MtCO2e-tagged values are genuine megatonnes: its prior-year Scope 3 of 23 MtCO2e matches the 23 Mt figure in its FY2024-25 PDF.
The 22 MtCO2e-tagged filings, verified one by one
Each square is one company, checked against its own PDF filing
Verification: unit stated in each company's own PDF filing; eight exact digit-for-digit value matches, including one matched in both current and prior year.
The precise defect is therefore not a wrong unit — it is an unreliable unit field. The identical tag carries tonnes for most filers and megatonnes for one. No consumer of this data — researcher, data vendor, or index provider — can normalize it by trusting the tag.
This is not a Nifty-50 quirk — it runs across the whole first vintage
We then read the Scope 1 unit tag on every FY2025-26 BRSR filing in the exchange window — 768 companies, not just the fifty largest. The pattern is not confined to big caps; it is systemic.
How all 728 filers tag Scope 1 emissions
Every FY2025-26 BRSR XBRL in the exchange window that reports a Scope 1 value
Source: NSE BRSR XBRL filings, parsed 19 Aug 2026. Gold = the tag under examination.
Filers tagging Scope 1 in MtCO2e
432 of 728 — 59%
the same tag Reliance, ONGC and Coal India use for tonnes-scale values
…whose value is physically impossible as tagged
263 of those 432
read literally, each exceeds ~1.3 billion tonnes of CO₂-equivalent
The population census is deterministic — it only reads the tag; it makes no judgement. The conclusion that these MtCO2e values are really tonnes rests on the PDF checks, which we extended beyond the Nifty 50 and which keep holding: Hindustan Zinc's PDF states "Metric tonnes of CO2 equivalent" for a Scope 1 of 47,82,041 — exactly its XBRL value tagged MtCO2e; Tata Chemicals' PDF reads 48,50,029, again digit-for-digit. Across every one of the 24 companies checked against their own PDFs — spanning cement, metals, power, banking, pharma and IT — the MtCO2e-tagged figure is tonnes, the sole exception being Tata Steel's genuine megatonnes. The 169 smaller MtCO2e values that are not magnitude-impossible are left unclaimed; only the deterministic census and the PDF-verified sample are asserted here.
Likely mechanism
ONGC's own PDF labels its emissions unit "MTCO2e." In greenhouse-gas accounting convention, "MT" reads metric tonnes; a filing tool mapping the abbreviation to the taxonomy's megatonnes unit would produce exactly the pattern observed — a shared software-side misreading rather than 21 independent data-entry errors. We flag this as a plausible mechanism, not an established fact.
Why it matters
This is the first machine-readable vintage of Indian corporate climate data. Whatever consumes it downstream inherits the error unless it re-verifies every company by hand — which removes the point of machine-readability. It is also fixable at the source: one validation rule in the filing utility would catch essentially every instance at filing time. And in every verified case the number matches the company's PDF — this is a metadata defect, not misreporting. No company is accused of misstating its emissions.
FAQ
Does this mean companies misreported their emissions?
No. In every verified case the numeric value matches the company's own PDF. The defect is in the machine-readable unit metadata, most plausibly introduced by filing software.
Could the MtCO2e tags simply all be read as tonnes?
No — Tata Steel's genuinely-megatonne figures would then be wrong by a factor of a million in the opposite direction. That is the core problem: the tag cannot be trusted in either direction without per-company verification.
What is the fix?
A plausibility validation in the filing utility (no single Indian company's Scope 1 plausibly exceeds ~500 MtCO2e), and a one-time corrected re-filing window. The underlying disclosures need no restatement.