After the Gujarat High Court Read Down Rule 28(2): Re-Running the Surety Bond vs Bank Guarantee Sum

An Indian group bidding an EPC or infrastructure contract has to post security three or four times over the life of the job: bid security at tender, performance security at award, an advance payment guarantee when mobilisation money moves, and retention security through the defect liability period. Three instruments can carry that obligation. The parent company can issue a corporate guarantee to the employer or to the subsidiary's lender. A bank can issue a bank guarantee against sanctioned limits. An IRDAI-licensed general insurer can issue a surety bond under the IRDAI (Surety Insurance Contracts) Guidelines, 2022.

Treasury teams usually compare the three on headline price: guarantee commission per annum, surety premium per annum, and a corporate guarantee that looks free because no third party charges for it. GST sits on all three instruments at 18 percent, but on a different base in each case, and it is creditable in some hands and stranded in others. Once the tax and the collateral cost are added, the ranking between the three can invert.

Two events force a rebuild of that arithmetic now. The Gujarat High Court ruled in August 2026 on how related-party corporate guarantees are valued under Rule 28(2) of the CGST Rules. And simplification of GST on corporate guarantees is reported to be on the agenda of the 57th GST Council meeting on 12 September 2026. A contractor signing a five-year security structure this quarter is committing to a cost base that both events move.

Rule 28(2) prescribes a deemed value for a guarantee given by a person to a banking company or financial institution on behalf of a related person. The mechanism fixes the taxable value at 1 percent of the amount guaranteed per annum, or the actual consideration, whichever is higher. Groups that charged a token guarantee fee, or none, found themselves taxed on a value they never received.

The August 2026 judgment did three separate things, and they should not be collapsed into a single headline.

1. It upheld the levy in principle. GST applies to a corporate guarantee issued between related parties. The argument that a parent guaranteeing its own subsidiary is not making a supply did not succeed.
2. It read down the words "whichever is higher" as arbitrary. Per VATupdate's 22 August 2026 report, the court struck at the part of the mechanism that forced a deemed value even where a real, negotiated consideration existed.
3. It rejected retrospective application. The 1 percent valuation mechanism cannot be applied to periods before 26 October 2023, the date from which it operates. TaxO reported the same restriction on 14 August 2026.

The third holding has immediate cash value for groups already carrying show cause notices covering earlier years. The second changes forward planning, because it reopens the question of whether the tax should follow a genuinely arm's length guarantee fee rather than a flat deemed figure.

Note the scope. Rule 28(2) is written around guarantees given to a banking company or financial institution. A parent guarantee issued directly to a project employer is a different fact pattern from one backing a subsidiary's credit facility, and the two should not be assumed to attract identical treatment.

GST on a corporate guarantee is not automatically a wash. TaxO reported on 11 July 2026 that CBIC was likely to issue a circular clarifying or changing the valuation mechanism precisely because the 18 percent paid on corporate guarantees is often not recoverable as input tax credit. That is the practical grievance behind the litigation, and it is worth being specific about why the credit fails.

A credit is only useful where the recipient has taxable output against which to set it off, is registered in the state where the supply is treated as made, and is not carrying an accumulated balance it cannot use. Group guarantee structures break on all three.

• Holding companies with no taxable output. A pure investment holding company that issues a guarantee has little or no outward taxable supply to absorb the credit its subsidiary's invoice generates on the other side, and the subsidiary's credit may itself sit idle.
• Project SPVs in a build phase. An infrastructure SPV accumulates credit for years before it raises a single taxable invoice. Tax paid in year one on a guarantee is cash out now against a credit that unlocks much later, if at all.
• Recipients in exempt or partly exempt lines. Where the guaranteed entity's output is exempt, the credit is blocked at source and the 18 percent is a straight cost.
• Place of supply frictions. Multi-state groups regularly find the credit lands in a registration different from the one carrying the offsetting liability.

The same test applies to the other two instruments, and this is where the comparison turns. GST on a bank's guarantee commission and GST on an insurer's surety premium both attach to a service bought by an operating contractor with substantial taxable output. For that contractor the 18 percent is normally creditable. For the same group's guarantee flowing out of a holding company, it frequently is not. The instrument that looks free is the one most likely to carry a tax that never comes back.

A bank guarantee has three cost components, and only the first appears on the pricing sheet.

Commission. Charged per annum on the guarantee amount, priced off the contractor's internal rating and the tenor. GST at 18 percent applies to the commission and is ordinarily creditable to a taxable contractor, so the effective drag is close to the headline rate.

Cash margin. Banks routinely hold a margin against non-fund based limits, taken as fixed deposit or a lien on balances. The cost of that margin is the gap between the deposit return and the contractor's own cost of funds, applied to the margin amount for the full life of the guarantee. On a long defect liability period this can exceed the commission itself.

Limit consumption. A bank guarantee sits inside the sanctioned non-fund based limit and counts against the group's overall exposure with that bank. The opportunity cost is the next bid the contractor cannot support because the limit is full. That cost is real and does not appear in any ledger.

The practical failure mode is a contractor treating the margin as a recoverable asset rather than a cost. The deposit returns at release, so it feels like a cash movement. Over a seven year concession with retention security running to the end of the defect liability period, the funding differential on that margin is a live P and L item every year it is outstanding. Our note on replacing bank guarantees with surety on power sector contracts covers where employers have accepted the substitution.

A surety bond issued under the 2022 Guidelines is a three party contract of guarantee. An IRDAI-licensed general insurer guarantees the project owner that the contractor will perform. Its cost structure differs from a bank guarantee in ways that matter to the sum.

Premium, with GST at 18 percent. The premium is a general insurance premium bought for business purposes by an operating contractor, so for a taxable contractor the tax is normally creditable in the same way commission tax is. The comparison against bank commission should therefore be run on tax exclusive numbers, then the credit position confirmed rather than assumed.

Little or no cash margin. Surety insurers generally do not hold the cash collateral banks hold, which is the entire commercial case for the product.

Indemnity exposure instead. The insurer's recourse runs through an indemnity agreement signed by the contractor and usually by its promoters. Where a bank debits the contractor's account on invocation and recovers instantly, the insurer pays the employer and then pursues an already distressed contractor under the indemnity. That difference in recourse explains why surety appetite tends to be binary: underwriters decline a marginal contractor rather than price the risk up. Read the surety bond fundamentals for infrastructure contractors for how that credit assessment is run.

Bond wording risk. Whether the bond is conditional or unconditional, and how closely its trigger tracks the underlying contract, decides how easily an employer can call it. A cheaper bond with a wider trigger is not cheaper. The policy wording is a cost input.

Capacity is also part of the price. IRDAI reduced the solvency control level for surety business to 1.5x from 1.875x and removed the 30 percent per contract exposure limit, and amendments to the General Financial Rules recognise insurance surety bonds as equivalent to bank guarantees for central government procurement. Both changes widen the set of contracts on which the substitution is available at all. The IRDAI operational framework for surety bonds sets out the constraints that remain.

Price a corporate guarantee as if it were bought from a third party, because after Rule 28(2) the tax authority effectively does.

Start with the deemed value. On a guarantee of INR 100 crore, a 1 percent per annum deemed value produces a taxable value of INR 1 crore per year, and GST at 18 percent on that is INR 18 lakh per year. Where that 18 lakh is creditable in the recipient's hands, the economic cost is small. Where the recipient is a build phase SPV or an entity with blocked credit, the full amount is cash out every year the guarantee is outstanding, on an instrument the group thought was free.

Against that, add the items that never reach a pricing sheet:

• Balance sheet disclosure. A guarantee is a contingent liability that lenders and rating agencies read.
• Cross default reach. A parent guarantee pulls a subsidiary's project failure onto the parent's covenants in a way a bank guarantee or a surety bond does not.
• Employer acceptance. Many public employers will not accept a parent guarantee for performance security at all, which limits the instrument to intra group and lender facing use.
• Valuation dispute risk. Even after the read down, a guarantee priced at nil or nominal consideration remains contestable, and a notice cycle carries its own cost.

Build one table per instrument, per tranche of security, over the full tenor including the defect liability period. Use the contractor's own numbers, because the two costs that decide the outcome (the funding differential on margin, and whether the tax credit is usable) are entity specific.

1. Headline charge, tax exclusive. Guarantee commission per annum, surety premium per annum, or deemed guarantee value for a corporate guarantee.
2. GST at 18 percent on that charge, then a separate line for the portion that is genuinely non-creditable in the paying entity's hands. This is the line most models omit, and for a corporate guarantee out of a holding company it is often the whole amount.
3. Collateral cost. Margin amount multiplied by the gap between deposit yield and the contractor's marginal cost of funds, multiplied by the years outstanding. Zero or near zero for surety.
4. Limit displacement. The value of the non-fund based headroom consumed, expressed as bid capacity foregone.
5. Recourse and indemnity exposure. For surety, the promoter indemnity and its practical reach. For a corporate guarantee, the cross default consequence at parent level.
6. Wording risk. Conditional versus unconditional, and how far the trigger departs from the underlying contract.

Run the total for each instrument across each tranche. In many EPC structures the outcome is mixed rather than uniform: a bank guarantee where the employer insists on one and the tenor is short, a surety bond where the tenor is long and the margin cost dominates, and a corporate guarantee confined to lender facing use where the group can actually absorb the credit. Brokers earn their fee on this line by mapping insurer appetite and constructing a credit shaped submission, which the placement economics on surety sets out in detail. The indemnity position and the engineering insurance programme sitting alongside it should be reviewed in the same exercise, since the employer usually reads them together.

Business Today reported on 29 August 2026 that simplification of GST on corporate guarantees is expected on the 57th GST Council agenda for 12 September 2026, and the Economic Times reported on 30 August 2026 that the same meeting carries easier input tax credit access for buyers. Both items point the same way for this decision, and neither is a reason to defer it.

Before the meeting:

• Quantify the exposure on existing guarantees. For each outstanding related party guarantee, compute the deemed value, the tax at 18 percent, and the share that is non-creditable in the paying entity. That number is the size of the position the Council could move.
• Separate pre and post 26 October 2023 periods. Any demand covering earlier periods now runs against the retrospectivity holding. Segregate those years in working papers before responding to a notice.
• Document real consideration where it exists. The read down of "whichever is higher" only helps a group that can show a genuine, evidenced guarantee fee. A guarantee priced at nil has nothing to fall back on.
• Do not sign a five year security structure on the assumption that relief arrives. An agenda item is not a notification.

After the meeting, re-run step 2 of the comparison above. If input tax credit access widens, the corporate guarantee line gets cheaper for operating companies and stays expensive for holding companies and build phase SPVs. If the valuation mechanism is simplified, the deemed value input changes and the tranche level ranking may move with it. Margin cost, limit displacement, indemnity exposure and wording risk do not move at all, which is a reminder of how much of this decision was never a tax question.