If your business exports under an Advance Authorization and claims a refund of unutilised input tax credit, you may have run into a specific and confusing corner of the GST refund rules: the difference between Rule 89(4) and Rule 89(4B) of the CGST Rules. For several years, this wasn’t a minor technicality. It was a genuine, unresolved ambiguity in the law itself, one that produced real DGGI summonses and real refund recovery demands.
The two rules, and the gap between them
Rule 89(4) is the general formula for refunding unutilised ITC on zero-rated supplies made without payment of tax: a proportionate calculation based on turnover.
Rule 89(4B) applied in a narrower situation: where a supplier had used certain notified duty-exemption benefits (including those linked to Advance Authorization) on the inputs used to manufacture the exported goods. Where it applied, the refund of ITC on those inputs was meant to be based on actual, one-to-one correlation between inputs used and goods exported, not the proportionate turnover-based formula.
The problem was that Rule 89(4B) described what should happen in such cases but never laid out a computation formula for doing it. Rule 89(4), by contrast, had a clear, ready-to-use formula. Two overlapping provisions, one of them incomplete. Exporters filing refunds against Advance Authorization exports had no unambiguous method to follow, and many filed under Rule 89(4) simply because it was the only complete formula available.
Where that gap turned into real exposure
This wasn’t a hypothetical problem. DGGI summonses were issued to exporters on precisely this basis: a refund filed under Rule 89(4) for exports linked to Advance Authorization, with the department’s position being that Rule 89(4B) should have applied instead, and that the mismatch alone made the refund suspect.
That argument has a real flaw worth naming directly: filing under the wrong sub-rule is, at most, a classification issue. It is not, by itself, evidence that an excess refund was granted. The two provisions can produce different quantum in some cases, but establishing an actual excess requires an actual recomputation under the correct method, not an inference from which box was ticked on the refund portal.
How the courts settled it: Filatex India
This exact question reached the Gujarat High Court in Filatex India Ltd. v. Union of India. The department had demanded recovery of a refund on the theory that it should have been computed under Rule 89(4B) rather than Rule 89(4).
The Court didn’t accept a blanket “wrong rule, therefore excess refund” position. Instead, it quashed the recovery order and remanded the matter, directing the authorities to actually recompute the eligible refund using the input-output ratio of raw materials consumed in the exported goods, the method contemplated (but never formularised) under Rule 89(4B). In other words: work out the real number under the right method, rather than treat a classification mismatch as proof of wrongdoing on its own.
That shifts the burden in a way that matters. It moves the department from “you used the wrong category” to “here is the actual excess we’ve calculated, and here is how we calculated it.” Those are very different things to have to defend.
Update, October 2024: the rule no longer exists
The section below was added after this post was first published, once the rule at the centre of this dispute was removed from the statute book.
Effective 8 October 2024, via Notification No. 20/2024-Central Tax, the CGST Rules omitted sub-rules (4A) and (4B) of Rule 89 entirely, along with the consequential references to them elsewhere in Rule 89. This followed a recommendation of the 54th GST Council meeting.
The reasoning behind the omission is worth noting, because it validates the substance of the argument above: the one-to-one input-output correlation requirement under the old 89(4A)/89(4B) regime was seen as unduly restrictive on exporters. The government’s fix wasn’t to sharpen that formula. It was to scrap the separate mechanism altogether and fold all zero-rated export refunds, regardless of whether notified exemption benefits were used on inputs, into the single proportionate formula under Rule 89(4).
Two practical consequences if you have exposure from an earlier period:
If you have a pending demand or show-cause notice built on Rule 89(4B), the fact that the sub-rule has been omitted, without an accompanying savings clause preserving pending proceedings under it, is itself a live argument. Whether that omission causes such proceedings to lapse entirely is being tested, but it is a serious point to raise, not a footnote.
If you’re still filing refunds for older periods where Advance Authorization or similar notified benefits were used on inputs, check which rule applied at the time of export and structure your working papers accordingly. The current, simplified position under Rule 89(4) doesn’t retroactively rewrite what applied in FY 2019-20 or FY 2020-21.
The broader lesson
A demand built around “you used the wrong refund category” sounds procedural, but it can carry real financial exposure if the underlying burden of proof isn’t examined closely. The department has to establish an actual excess, properly computed, not merely infer one from a filing choice. Where the underlying law itself was genuinely unsettled, as it was here for several years, that context is part of the legal argument, not background noise.
If you’ve received a summons or notice referencing Rule 89(4), 89(4A) or 89(4B) for an older export period, the analysis above is a starting point, not a substitute for looking at your specific numbers. Get in touch if you’d like it reviewed.