US–India Remittance: The Two Perimeters an Operator Has to Satisfy

A US–India remittance operator sits inside two regulatory perimeters that were written independently, do not reference each other, and constrain different things. The US side governs what you must tell the sender and what rights the sender has. The India side governs how much can arrive, from whom, for what purpose, and in what form.

Most corridor write-ups cover neither properly, and instead spend their length on market-size figures. This is the operational version.

For the US-side licensing foundation — state money transmitter licensing, FinCEN MSB registration and BSA/AML programme structure — see the US–Mexico corridor reference, which sets it out in detail and applies equally here. This article covers what is specific to India.

The sending side: Regulation E subpart B

The CFPB's Remittance Transfer Rule sits in Regulation E subpart B. Three provisions carry most of the operational weight.

Disclosure (§1005.31). Before payment and again on a receipt, the provider must disclose the transfer amount, the fees, the exchange rate applied, and the total the designated recipient will receive. The exchange rate disclosure is the mechanism that makes an embedded FX spread visible — you do not have to publish your margin, but you do have to publish the rate you used and the resulting amount, which lets anyone compute it.

Cancellation (§1005.34). The sender can cancel and get a refund where the request reaches the provider no later than 30 minutes after the sender makes payment, provided the funds have not already been picked up or deposited. The refund is due within three business days.

Thirty minutes is short, but it is a real constraint on your architecture: it means you cannot treat funding as final at the moment of payment, and any design that pushes irrevocably to the India leg inside that window has to handle a reversal it may not be able to execute.

Safe harbour (§1005.30(f)). A provider is deemed not to be transferring "in the normal course of business" if it provided 500 or fewer remittance transfers in the previous calendar year and provides 500 or fewer in the current one. Above that, the full rule applies. The regulation also carries a transition provision for a provider that crosses the threshold mid-year.

Five hundred transfers is a very low bar. In practice any operator building a corridor product is inside the rule from the outset and should design to it rather than plan to grow into it.

The 2026 excise tax, and why its scope matters more than its rate

A 1% excise tax on remittance transfers took effect on 1 January 2026.

The rate is not the interesting part. The scope is: it applies to transfers funded by cash, money order or cashier's cheque, and exempts transfers funded from a bank account or by debit or credit card. The sender is liable; the provider collects and remits.

That scoping falls almost exactly along the line between two business models. Agent-network and walk-in flows — the ones that take cash over a counter — are inside it. Digital, account-funded flows, which dominate the cheapest end of this corridor, are outside it.

If you run both, this is not a uniform 1% cost. It is a tax on one product line, and it changes the relative economics between them. Model it against your funding mix.

The receiving side: two RBI channels, and they are not interchangeable

Inward personal remittance into India runs through one of two RBI-authorised arrangements. Choosing between them is a product decision made early and hard to reverse.

Money Transfer Service Scheme (MTSS)

Built for small personal transfers, and tightly bounded:

MTSS explicitly does not cover donations, trade transactions, property purchase or investment, and cannot be used to credit an NRE account. The recipient bank must KYC-verify before crediting a direct-to-account transfer.

A trap worth naming. A superseded 2003 RBI circular gives the per-beneficiary limit as 12 remittances per year. The current Master Direction gives 30. Both documents are findable, and the old one looks perfectly legitimate. If a compliance note or vendor deck quotes 12, it is citing an instrument that has been overtaken — check the Master Direction's own revision stamp before acting on any figure in this area.

Rupee Drawing Arrangement (RDA)

Structurally different. RDA is, in the RBI's own words, "a channel to receive cross-border remittances from overseas," under which Authorised Dealer Category I banks enter into tie-ups with non-resident Exchange Houses to open and maintain their Vostro Account.

It carries a broader permitted-purpose list than MTSS and a higher cap for trade transactions. But it comes with one hard product constraint: cash disbursement is not permitted under RDA at all.

So the decision resolves quickly. If your product hands cash to a beneficiary, RDA is unavailable and MTSS is your channel, with its USD 2,500 and 30-per-year ceilings. If you are crediting accounts and need purposes beyond personal maintenance, RDA is the wider channel — and the Exchange House relationship, not your own licence, is what admits you to it.

Both arrangements are authorised under FEMA, which is the statutory basis rather than a separate compliance layer to satisfy.

UPI is not a US–India remittance rail

This is the corridor's most persistent misconception and it is worth stating flatly.

Two different things get conflated:

NRI UPI access. NPCI permits non-resident Indians holding international mobile numbers in around ten countries — the United States among them — to link that number to an NRE or NRO account and use UPI. The sources reviewed state repeatedly that this facility cannot be used for cross-border payments. It is domestic Indian functionality, applied to money already sitting in an Indian account. It moves nothing across a border.

NPCI International (NIPL) corridor linkages. These are genuine cross-border real-time payment arrangements, and several exist — Singapore's PayNow linkage being the most operationally significant. No United States arrangement was found in the sources reviewed, despite targeted searching. That is an absence of retrieved evidence rather than proof none exists or is planned, but it is enough that you should not build a roadmap on one.

The practical consequence: payout into India from the US runs on bank credit, cash pickup through an agent network, or wallets. Not UPI. Any product plan, pitch deck or article promising "send to India via UPI" from a US origin is describing something that was not evidenced in the sources reviewed.

For UPI's genuine cross-border role — which is real, but sits on the merchant-acceptance side rather than the remittance side — see the UPI infrastructure reference.

Which model you actually operate under

Neither channel is something a US operator holds directly. Both are Indian-side authorisations, and that shapes market entry more than any US licence does.

Under RDA, the gating relationship is the Exchange House. The Master Direction describes Authorised Dealer Category I banks entering into tie-ups with non-resident Exchange Houses to open and maintain Vostro accounts. The Exchange House is the overseas counterparty in that structure. A US operator therefore reaches RDA by becoming, or contracting with, an entity that holds such a tie-up — not by applying to the RBI on its own account.

Under MTSS, the structure is Overseas Principal and Indian Agent. The Overseas Principal contracts with an Indian Agent that handles domestic distribution and payout, and the RBI sets financial standing requirements on both sides of that pair.

Two practical consequences:

1. Your India-side capability is inherited from a partner, and so are its limits. If your partner operates MTSS, you have MTSS ceilings, whatever your own volumes or balance sheet. Discovering that after launch is expensive.
2. The channel decision precedes the partner decision. Work out whether you need cash payout and whether your purposes extend beyond personal maintenance first, because those answers determine which kind of partner you are looking for.

This mirrors the pattern in the US–Mexico corridor, where most operators reach the receiving side through a licensed local partner rather than direct authorisation — but the Indian structures are named, bounded and purpose-restricted in a way that makes the choice less reversible.

Payout, failure and returns

Domestic Indian rails carry the last leg. IMPS runs continuously; NEFT is the batch-oriented alternative that has since moved to continuous settlement.

Returns discipline is where the operational detail sits. RBI's NEFT procedural framework establishes that a transaction that cannot be credited must be returned to the originator within a defined window, with the returning bank liable for penal interest on delay, and enumerates valid reasons for non-payment: no such account, account-number mismatch, or a court-order restraint.

The two-hour window survives the move to continuous settlement. RBI's current NEFT FAQ, updated September 2024, still states that where credit is not possible the destination bank must return the transaction to the originating branch within two hours of completion of the batch in which it was processed, and that a bank failing to credit or return inside that window is liable for penal interest to the affected customer.

For your own reconciliation, the practical requirement is that a failed India-side payout must map back to a sending-side transaction that may already have passed its Regulation E cancellation window. Those two clocks are not aligned, and nothing in either regime aligns them for you.

Cost, and what the number actually is

The World Bank's Remittance Prices Worldwide survey put the US→India corridor at 3.68% average total cost on a USD 200 send, in its 4–28 August 2025 survey window.

That is competitively priced by any standard, with a long tail of digital providers well under 1% and legacy bank channels an order of magnitude above them.

Two honest qualifications. This is a live corridor snapshot with a stated survey window, not the figure from the formal quarterly report PDF, which could not be retrieved — methodology and snapshot dates may differ. And a corridor average blends channels that are not substitutes: an account-funded digital transfer and a cash-over-the-counter payout are different products with different costs and, since January 2026, different tax treatment.

What an operator should take from this

• Two perimeters, no bridge. Regulation E governs your disclosure and the sender's rights; MTSS or RDA governs what may arrive and how. Neither anticipates the other.
• The cash question decides your India channel. Cash payout means MTSS and its USD 2,500 / 30-per-year ceilings. No cash means RDA is open to you.
• The 500-transfer safe harbour is not a growth runway. Design to the rule from day one.
• The excise tax is a funding-mix question, not a blanket 1%.
• UPI is not the answer to this corridor, whatever the pitch deck says.
• Verify RBI figures against the current Master Direction. The superseded circular quoting 12 remittances a year is still out there and still looks authoritative.