GST Reverse Charge: Why a Books-Only Reconciliation Can't Catch What Was Never Flagged
Every GST audit programme has an RCM step. Most of them run the same way: pull the ledgers coded "reverse charge," check the tax was computed correctly, confirm it was paid under the right head, and move on. It's a sound procedure — for the transactions it looks at. The structural problem is what it doesn't look at: any inward supply that should have been flagged RCM but never was, in the books, in the first place.
That's not an edge case. It's the single most common way reverse-charge liability under Section 9(3) of the CGST Act goes unrecorded — not because the tax was computed wrong, but because nobody ever told the books to compute it at all.
The Structural Gap in a Books-Only Check
A books-only RCM procedure works like this: filter the purchase/expense register for ledgers or vouchers already tagged RCM, then test those entries for correctness — right rate, right period, ITC treatment, payment under the right challan. It's a well-formed test of accuracy. It says nothing about completeness.
The reason is mechanical, not a matter of auditor diligence. If an entry was never flagged RCM in the books — say a freight bill booked straight to "Transport Charges" with no reverse-charge tag, or a director's sitting fee booked to "Directors' Remuneration" with no GST treatment applied at all — a books-only recon will never surface it. There is nothing in the books pointing to it. You're testing the population the client's own bookkeeping already decided was RCM, which is precisely the population where the client's bookkeeping is least likely to be wrong.
This matters because the notified categories under Section 9(3) are exactly the ones that get miscoded in ordinary practice: goods transport agency (GTA) services, legal services from an advocate or firm of advocates, director sitting fees, security services (when supplied by anyone other than a body corporate, to a registered recipient other than certain exempt categories), sponsorship services, and import of services. None of these arrive from the supplier with GST charged — a GTA either charges under forward charge (having opted in) or the recipient self-assesses; an advocate's invoice typically has no GST line at all; a director's sitting fee is booked as remuneration, not "services received." The invoice looks, on its face, like a normal expense. Whether it's RCM at all is a fact about the category of the transaction, not something visible on the face of the bill — and that's exactly the judgement a bookkeeping team, working fast at month-end, is most likely to skip.
Section 9(4) — RCM on supplies from unregistered persons generally — has been substantially deferred since 2019 except for specified classes of recipients (notably promoters procuring construction inputs and cement). But 9(3)'s notified-category list has stayed in force the whole time, and it's the one that keeps producing this gap.
What a Portal-Side Check Adds
The fix isn't a smarter rule inside the books. It's an independent signal from outside the books — the GST portal itself, which doesn't care what the client's ledger called an entry.
Two portal-side data points matter here. First, GSTR-3B Table 3.1(d) — "Inward supplies liable to reverse charge" — is where the recipient self-reports RCM liability for the period. If a return period shows near-zero or suspiciously flat values in 3.1(d) across the year, while the ledger population contains a meaningful spend in GTA, legal fees, security services, director fees, or sponsorship, that gap is visible without opening a single voucher.
Second, GSTR-2A/2B carries a reverse-charge flag on invoices reported by the supplier's own GSTR-1 — where a registered GTA, security agency, or similar supplier has opted to bill under forward charge, or has separately indicated the supply attracts reverse charge. Cross-referencing this flag against what the books actually did with the corresponding purchase gives a second, independent read that doesn't depend on the books having tagged anything correctly in the first place.
Put together — the ledger population in notified categories, the portal's own 3.1(d) trend, and the 2A/2B reverse-charge flag — the check stops asking "is the RCM entry in the books correct?" and starts asking "does the volume of GTA/legal/security/sponsorship spend in the books line up with what the portal shows as reverse-charge activity for this GSTIN?" That second question is the one a books-only recon structurally cannot ask, because it has no view outside the books to compare against.
CORAA's GST reconciliation engine runs this as a cross-check: it reads GSTR-3B Table 3.1(d) and the reverse-charge indicator in GSTR-2A/2B alongside the books' own RCM flag on inward lines in the notified categories, and surfaces where the two diverge. It doesn't decide whether a given transaction is RCM-liable — that's a categorisation judgement for the auditor, informed by the nature of the supply and the recipient's status — it surfaces the population where books and portal disagree, so the auditor isn't relying on the books' own tagging to find the gap.
How Large the Gap Can Get
To put a sense of scale on it: it's entirely possible for a books-only RCM check to show a few thousand rupees of reverse-charge liability for a period — exactly what the ledgers happened to tag — while a portal-side cross-check against the same period's notified-category spend implies an exposure two to three orders of magnitude larger. Not because the books-tagged entries were wrong, but because almost none of the actual GTA, legal, security, or director-fee spend for the period was ever tagged RCM to begin with. This is illustrative of the order of magnitude a books-only check can miss structurally, not a specific cited engagement — the gap between "what the books flagged" and "what the portal implies should have been flagged" compounds silently, invoice after invoice, all year.
The Exposure Once It's Found
Once a genuine RCM gap is confirmed, the exposure runs on three tracks. First, the tax itself — Section 9(3)/9(4) liability doesn't disappear because it wasn't self-assessed in time; it remains payable, typically discovered well after the return period has closed. Second, interest under Section 50 accrues from the original due date, not the date of discovery — for a gap spanning a full year, that's a meaningful compounding cost by the time it surfaces. Third, because RCM tax paid in cash is available as ITC only in the period it's actually paid (not the period the liability arose), a late-discovered gap can also mean ITC that should have offset the RCM cash outflow was never claimed in time, or was claimed against the wrong period.
None of this is a judgement CORAA makes. Whether a specific director's fee arrangement, a specific security contract, or a specific freight arrangement actually falls under the notified RCM categories — and what to do about a confirmed gap — is the auditor's call, informed by the facts of the engagement. What a portal-side check does is put the right population in front of the auditor to make that call on, instead of leaving it dependent on whether the client's bookkeeping happened to flag it months earlier.
Even without automated cross-referencing, the same logic is worth building into any GST audit programme: pull the full ledger population for GTA/freight, legal & professional fees, director remuneration/sitting fees, security charges, and sponsorship, independent of whether the books tagged them RCM — then check each category's GST treatment against Section 9(3) applicability, rather than starting from the books' own RCM-tagged subset. The moment the starting population is "what the books already flagged," the check has already conceded the one failure mode it most needs to catch.
Frequently Asked Questions
Why does a books-only RCM reconciliation miss so much liability?
Because it tests correctness of entries already tagged reverse charge in the books, not completeness of the population that should have been tagged. If a GTA bill, legal fee, or director's sitting fee was never flagged RCM at the point of booking, a books-only check has no way to notice it — there's nothing in the books pointing to the gap. The check is only as good as the tagging it inherits.
What does GSTR-3B Table 3.1(d) show?
Table 3.1(d) is where a registered person self-reports inward supplies liable to reverse charge for the return period. Comparing the trend in 3.1(d) against the volume of notified-category spend in the books (GTA, legal services, security services, director fees, sponsorship, import of services) is one of the clearest portal-side signals that RCM liability is being under-reported.
Does GSTR-2A or 2B show reverse-charge transactions?
Yes — where a registered supplier files GSTR-1 and marks a supply as attracting reverse charge, that flag carries through to the recipient's GSTR-2A/2B. Cross-referencing this flag against how the books treated the same purchase gives an independent, portal-derived check that doesn't depend on the books' own RCM tagging.
Is Section 9(4) RCM still relevant for most businesses?
Largely no — RCM on supplies from unregistered persons under Section 9(4) has been deferred since 2019 for most taxpayers, with exceptions for specified recipient classes such as promoters procuring certain construction inputs. Section 9(3)'s notified-category list (GTA, legal services, director fees, security services, sponsorship, import of services) has remained continuously in force and is where most books-only gaps arise.
Reconciling RCM liability by re-checking only what the books already flagged will always miss the entries that were never flagged to begin with. CORAA's GST reconciliation module runs the portal-side cross-check — GSTR-3B Table 3.1(d) and the 2A/2B reverse-charge indicator against the books' own RCM tagging — so the gap surfaces before a notice does, with the auditor still deciding what each flagged difference means.