1
EXECUTIVE SUMMARY
The RBI’s Draft Directions on Interest Rates on Loans and Advances, 2026, are more than a consolidation of existing instructions. They reshape the economics and operating model of lending by stripping discretion from the benchmark, shifting it into the spread, and then placing tighter constraints on how that spread can evolve.
Applies to: Commercial banks, including small finance banks and local area banks, regional rural banks, urban and rural co-operative banks, all-India financial institutions (Export-Import Bank of India (EXIM Bank), National Bank for Agriculture and Rural Development (NABARD), National Bank for Financing Infrastructure and Development (NaBFID), National Housing Bank (NHB), Small Industries Development Bank of India (SIDBI)), and Non-Banking Financial Companies (NBFCs), including housing finance companies. Domestic operations only.
Effective from: 1 April 2027
Existing book migrated by: 1 April 2029
Replaces: Five entity-specific Interest Rates on Advances Directions, 2025: commercial banks, small finance banks, local area banks, urban co-operative banks, and rural co-operative banks, together with paragraphs 60 and 61 of the NBFC Credit Facilities Directions, 2025
Status: Draft issued for comments
The implications extend well beyond pricing policy. CFOs, CROs, Business Heads, Treasury, and Operations will need to jointly rethink how loans are priced, repriced, migrated, and managed through their lifecycle.
Five shifts that matter:
1
The benchmark becomes simpler but less controllable
The internal benchmark shifts from a composite of funding cost, Cash Reserve Ratio (CRR) carry, return on net worth, operating cost, and tenor premium to a three-month moving average of the cost of fresh deposits and borrowings. Operating cost and tenor premium move into the spread.
Implication: The benchmark becomes a largely prescribed funding-cost construct. The ability to manage pricing economics, therefore, shifts from benchmark design to the initial calibration of spreads.
2
Pricing discretion moves to origination and becomes harder to exercise later
The draft caps benchmark resets at three months and fixes the chosen reset periodicity for the life of the loan. Non-credit components of the spread cannot be revised for three years, while changes to the credit risk premium require a documented change in the borrower’s credit profile and a comprehensive review.
Implication: Pricing becomes an origination discipline rather than a portfolio-management lever. Retention concessions, competitive pricing and subsequent margin management acquire a longer-term economic cost.
3
Faster repricing comes with fewer shock absorbers
Quarterly resets and more mechanical reset dates will accelerate transmission of funding costs and benchmark movements to the loan book. At the same time, tenor-linked benchmarks, annual resets, and disbursement-linked reset mechanisms that previously provided flexibility are withdrawn.
Implication: Earnings and margin sensitivity to funding costs may increase precisely as lenders lose some of the mechanisms historically used to smooth that transmission. Asset and Liability Management (ALM), pricing, and business strategy will need to be managed together.
4
The back book becomes a mandatory transformation program
From 1 April 2027, new lending must follow the new framework. By 1 April 2029, all existing loans linked to internal or external benchmarks must be migrated through a one-time mapping exercise. Migration requires borrower consent, cannot increase the borrower’s rate, and cannot carry a charge.
Implication: The two-year migration window is not merely an operational queue. Because the rate ceiling is determined by the rate immediately before migration, sequencing becomes a commercial decision. The exercise also creates a significant consent, documentation, data, and systems program.
5
A common rulebook will not create common economics
The framework brings banks, NBFCs, co-operative banks, and other regulated lenders under a common pricing architecture. However, competitive outcomes will still differ based on funding costs, risk-pricing capability, and the flexibility retained by different entity classes.
Implication: The draft may close regulatory arbitrage, but competitive differentiation will increasingly come from how efficiently a lender funds itself, how accurately it prices risk, and how well it calibrates spreads at origination.
The draft is still open for comments, and individual parameters may change. The underlying architecture, however, is less likely to change: a funding-cost-based benchmark, a floor beneath every loan, greater discipline around spreads, and mandatory migration of the legacy book.
2
A NEW BENCHMARK, AND A FLOOR UNDER EVERY LOAN
Two things together decide what a borrower pays: the benchmark a lender starts from, and the minimum it is permitted to charge. The draft Directions rewrite both. For banks, this is a rebuild of machinery they already operate. For non-banks, it is entirely new machinery.
This section deals only with how the rate is arrived at and how low it may go. How much of the rate can be changed and how often is addressed in Section 3.
2.1 The internal benchmark is stripped back to the cost of raising money
Today, a bank’s Marginal Cost of Funds-based Lending Rate (MCLR) is a composite of four things (existing paragraph 16): the marginal cost of funds, the negative carry on the cash reserve ratio (the cost of holding balances with the RBI that earn nothing), operating costs, and a tenor premium. The marginal cost of funds is itself a blend – 92 percent the cost of borrowings and 8 percent a return on net worth, the latter weight being tied to the common equity Tier 1 capital a bank must hold.
It is now a three-month moving average of the annualized, weighted-average interest cost on the volume of fresh domestic deposits and fresh borrowings raised during the month. Four cost layers leave the benchmark.
| Component | Existing Directions (banks) | Draft Directions, 2026 |
|---|---|---|
| Marginal cost of funds | Included – blended 92% cost of borrowings and 8% return on net worth | Included – and it is the whole benchmark |
| Negative carry on CRR | Included in the benchmark | Does not appear anywhere in the draft |
| Return on net worth | Included, at an 8% weight | Does not appear anywhere in the draft |
| Operating costs | Included in the benchmark | Moved out – named as a spread component |
| Tenor/term premium | Included, and the benchmark itself is tenor-linked | Moved out – named as a spread component |
Three mechanics inside the new calculation
1
Fresh flows replace outstanding balances.
Today, the rate used is a current one, but the weight applied to it is a stock – balances outstanding as a percentage of total funds other than equity, using rates offered on deposits as at the date of review. Under the draft, both the rate and the volume are based on what was actually raised during the month, and the interest figure is taken from the profit and loss account. For commercial banks, it points to Schedules 3 and 4 of the balance sheet for deposits and borrowings, and Schedule 15 of the profit and loss account for interest expended.
2
Tenor-linked publication ends.
Today, a bank publishes MCLR for overnight, one-month, three-month, six-month, and one-year maturities, with an option to publish longer ones, and the tenor is derived from a maturity-bucket rule. The draft requires the internal benchmark to be published on the first calendar day of each month, with no tenor-linked series. The maturity-bucket machinery goes with it, as does today’s freedom to choose a pre-announced publication date with Board or committee approval.
3
The rate is anchored to sanction, not disbursement.
Today, the MCLR prevailing on the date of first disbursement applies until the next reset. Under the draft, the benchmark published on the first of the month is the applicable benchmark for loans linked to it and sanctioned during that month.
A simpler calculation, and what is lost with it
Number of refinements disappear with it: the treatment of only the core portion of current and savings balances, the swap and hedge cost treatment for foreign currency deposits and borrowings, the option to use a published bank-bond benchmark yield as a proxy for long-term borrowings, the rule for floating-rate term deposits, and the option to reckon balances as at any day up to seven days before the rate takes effect. Fewer judgment calls, but also fewer levers with which to hold the number steady.
2.2 The external benchmark: a wider definition, the same mandate, one rule gone
Draft Directions names the Secured Overnight Rupee Rate for the first time, and the existing restriction to three-month and six-month Government of India Treasury Bill yields is dropped. The mandate itself is unchanged in substance. Floating-rate personal loans and floating-rate loans to micro, small, and medium enterprises (MSMEs) made by commercial banks must be linked to an external benchmark and may extend this to other categories if they wish.
Two changes sit underneath that continuity. The existing prohibition on a bank using more than one external benchmark within a single loan category has no counterpart in the draft. And the perimeter wording moves from “personal or retail loans (housing, auto, etc.)” to “personal loans,” a term the draft defines by reference to Banking Statistics I (Harmonized Definitions) rather than describing in the text.
2.3 The floor now sits under every loan
Draft Directions require both fixed- and floating-rate loans to be priced with reference to an internal or external benchmark plus a risk-based spread, and prohibit pricing any loan below the applicable benchmark. A floor is not new for banks. The existing circular already prohibits lending below the benchmark rate for a particular maturity. What changes is its reach and its level.
1
Reach.
Today, fixed-rate loans of tenor above three years are exempt from the benchmark provisions altogether. Fixed-rate loans with shorter tenors are floored at the four-component sum, and bill discounting and factoring facilities are excluded from that as well. The draft removes the tenor-based exemption and the bill discounting and factoring exception.
2
Level.
A bank’s floating-rate floor today is a tenor-matched MCLR that carries the reserve drag, operating costs, and a tenor premium. The draft’s benchmark carries none of them. The nominal floor, therefore, falls, even as it widens.
3
Reach, for everyone else.
For NBFCs, housing finance companies, co-operative banks, and all-India financial institutions, there is no pricing floor of any kind today. The concept arrives whole.
Hybrid loans get a cleaner rule than they have now: fixed-rate provisions apply to the periods when the rate is fixed, and floating-rate provisions apply to the periods when it floats.
Two provisions that give the floor teeth
- “Spread” is now a defined term, and it excludes charges and fees
- The credit risk premium can no longer be zero
Impact
The benchmark stops being a managed composite and becomes a straightforward measure of what money cost the lender recently. Published benchmarks will sit lower, and spreads will widen to hold the same all-in price.
Because the calculation runs on fresh flows rather than the existing stock of funding, a single month’s funding decisions move the lender’s own floor. The three-month moving average is the only smoothing left; the smoothing that came from weighting by outstanding balances is gone.
3
LESS ROOM TO REPRICE
Section 2 dealt with the rate a lender starts from and the floor it may not go below. This section deals with what happens next: how much of the price can be changed once a loan is on the books, and how often.
The draft answers that in two places that are usually read separately – the reset rules for the benchmark and the revision rules for the spread. Read together, they describe a book that reprices faster than it does today but on far less discretion.
3.1 A three-month ceiling on resets
Draft Direction requires that the benchmark on a floating-rate loan be reset at a periodicity chosen by the lender, not exceeding three months, and adds that, once fixed for a loan, that periodicity must remain unchanged for the entire tenor of the loan. It is worth discussing who things change for, because the three-month figure is not new everywhere.
| Loan type | Existing position | Draft position |
|---|---|---|
| Bank loans linked to MCLR | Reset periodicity of one year or lower, and it had to correspond to the tenor of the MCLR the loan referenced. | Three months at most, freely chosen within that cap, and fixed for the life of the loan |
| Bank loans linked to an external benchmark | Reset at least once in three months, so the cap was already three months old. | Same cap, but the periodicity now becomes irreversible, and the reset date is fixed |
| NBFC and housing finance company loans | No cap of any kind | Three months at most, unless the lender is in the Base Layer |
| Agricultural loans | Interest charged with reference to the crop season and due dates | Reset is linked to the crop season, up to 12 months, the one place a longer reset survives |
So, the substantive compression falls on MCLR-linked bank lending, where the outer limit moves from 12 months to 3 months, and on non-banks, which acquire a cap that previously did not exist. Externally benchmarked bank lending was already resetting quarterly; what it gains is rigidity.
One connected provision disappears. Today, the reset periodicity has to correspond to the tenor of the MCLR the loan was linked to. With the tenor-linked MCLR series abolished, that requirement has no counterpart in the draft, and reset periodicity becomes a free choice within the cap, decoupled from the benchmark itself.
3.2 Reset dates become mechanical
Today, a bank may specify its own reset dates and may offer loans with reset dates linked either to the date of first disbursement or to the date on which it reviews its MCLR. Reset dates, therefore, spread naturally across the calendar, loan by loan.
The draft removes that choice. Where the reset periodicity is less than a month, the benchmark is reset on the date the reset falls due under the loan agreement. In every other case, which will be most cases, it is reset on the first calendar day of the month in which the reset is due.
Directions completes the picture by requiring that the benchmark used for pricing, the reset periodicity, and the reset date be specified explicitly in the loan agreement. Today, only the reference benchmark and the periodicity have to form part of the loan contract.
3.3 The spread carries more and moves less
Two things happen to the spread at once. It takes on the cost layers that were left behind the benchmark, and it becomes considerably harder to change.
More components
Under the existing MCLR framework, the spread has two broad components: business strategy and credit risk premium. The draft sets out an illustrative list of four: credit risk premium, operating cost, term premium, and business strategy premium, and requires the spread to comprise the credit risk premium plus one or more other components. Operating cost and term premium are the two that have moved across from the benchmark.
They arrive without the rules that governed them inside the benchmark. Today, operating costs must exclude the cost of services that are separately recovered through service charges and must be calculated as a percentage of the marginal cost of funds. Neither requirement appears in the draft. Today, a change in tenor premium may not be borrower-specific or loan-class specific, and must be uniform across all loans for a given residual tenor. The draft describes the term premium simply as the premium associated with the loan’s tenor, with no uniformity requirement, so it may now differ from borrower to borrower.
The description of the business strategy also shifts slightly. Today, it reflects business strategy, market competition, embedded options in the loan product, and the loan’s market liquidity. The draft reflects competition, liquidity, expected returns, and other commercial considerations.
A longer lock, applied more widely
Paragraph 24 provides that, for a floating-rate loan, components of the spread other than the credit risk premium shall not be revised for a period of three years. A three-year rule is not new, but today it lives only in the external benchmark chapter and applies only to externally benchmarked loans. The draft extends it to every floating-rate loan.
Two features of the extended rule matter more than the headline.
- The clock restarts on any revision, in either direction.
- The retention lever sits outside the credit risk premium.
The lock is expressly limited to floating-rate loans. On a fixed-rate loan, the rate is by definition fixed for the whole tenor, so there is nothing for the rule to operate on.
3.4 The credit risk premium: a new trigger, and a narrower input set
This is the change most likely to be missed on a first reading, because the words look familiar.
Today, the restriction runs one way. The spread charged to an existing borrower shall not be increased except due to deterioration in the customer’s credit risk profile, supported by a full risk-profile review. Under the external benchmark, the credit risk premium may change only where the borrower’s credit assessment undergoes a substantial change, as agreed in the loan contract.
Draft reads differently in three respects. The credit risk premium may be revised, not merely increased, only when the borrower’s credit profile undergoes a change. The word “deterioration” is gone, so the restriction is symmetrical: a reduction now needs the same justification as an increase. And the qualifier “substantial” is gone, so any change in profile can support a revision, provided it is documented. A comprehensive review of the borrower’s credit risk profile must precede the revision in every case.
What may go into the credit risk premium also changes. Today, it is determined using a credit risk rating or scoring model that accounts for customer relationships, expected losses, and collateral. The draft’s list includes the probability of default, expected losses, available collateral security, and other risk mitigants, assessed for the borrower and the credit facility. Customer relationship, an express input today, is not among them.
Architecture becomes clear: relationship and competitive considerations belong in the non-CRP components, where they are subject to the three-year lock, and not in the credit risk premium, which answers only to credit.
One further provision does not carry forward. Today, the restriction on increasing the spread does not apply to loans under consortium or multiple banking arrangements. The draft contains no equivalent exclusion.
3.5 What now has to be written down
The draft moves several pricing decisions from practice into documents.
- Loan categories must be defined by the lender in its policy and approved. The draft expressly permits them to be built on product, borrower category, a combination of the two, or other criteria, such as whether the rate is fixed or floating.
- The methodology for determining the quantum of each spread component and the range of spread for each loan category must be in the policy.
- The benchmark, the reset periodicity, and the date of reset must be in the loan agreement itself.
- Any revision of the credit risk premium must be traceable to a documented change in credit profile and a comprehensive review, and must be consistent with the policy and the terms of the loan agreement.
Industry-wide impact
Repricing becomes faster but much less discretionary. Every floating-rate loan in the system, at every lender outside the small-entity carve-outs, will reflect a change in its benchmark within a quarter. Earnings become more sensitive to the benchmark and can no longer be managed through the timing of resets.
Because resets land on the first calendar day of the month rather than on dates derived from each loan’s disbursement, they concentrate. Large parts of a book will move on the same few dates, which changes the shape of interest income within a quarter and makes each month’s published benchmark consequential in a way it is not today.
The three-year lock changes the economics of defensive pricing. Because any revision restarts the clock, a rate cut offered to hold a client today removes the ability to reprice that category for three years from the date of the cut. Concessions acquire a cost that is not visible in the month they are given.
Credit risk premium discipline becomes symmetrical and evidence-based. Reducing a premium will require the same documented review as raising it, shifting real work onto credit teams, and making the rating or scoring framework the operative constraint on pricing.
Agricultural lending is the single carve-out in the other direction, with resets linked to crop season and permitted up to twelve months.
4
THE BACK BOOK
This section deals with the loans already on the balance sheet and with what is, in practice, the largest single piece of work the draft creates.
4.1 Two deadlines, not one
The draft comes into effect on 1 April 2027. However, all existing loans and advances linked to any internal or external benchmark are to be migrated to the new framework by 1 April 2029, through a one-time mapping exercise.
That produces two distinct obligations and a two-year overlap between them. From April 2027, all new lending has to be originated on the new architecture. Until April 2029, the legacy book may continue on its existing benchmarks. For those two years, a lender runs both frameworks at once.
This catches two older benchmarks. Base Rate is the internal benchmark for floating rate rupee loans sanctioned or renewed between 1 July 2010 and 31 March 2016; Benchmark Prime Lending Rate (BPLR) is the internal benchmark for loans sanctioned up to 30 June 2010. Both remain live today, and a bank is currently required to keep reviewing and publishing its Base Rate. Neither term appears anywhere in the draft, so loans priced off a benchmark first published in 2010 must be brought across by April 2029 and cannot be repriced upward in the process.
4.2 Three conditions attached to the migration
Draft attaches three conditions to the mapping exercise, and each of them constrains the lender rather than the borrower.
1
Borrower consent.
The mapping must be carried out with the borrower’s consent. The obligation to complete the exercise sits with the lender; the consent sits with the borrower.
2
A one-directional rate cap.
The borrower may not be placed in a disadvantageous position with respect to the applicable interest rate, and the lender must ensure that the revised rate does not exceed the rate applicable to that borrower immediately before the transition. There is no corresponding floor, so migration can leave a borrower’s rate unchanged or lower it, but never raise it.
3
No charges.
The lender may not levy any charges for the migration.
The cap is measured against the rate applicable “immediately before such transition”. Because the lender chooses when within the window to migrate each loan, the reference point and therefore the ceiling that the loan carries afterward move with the timing of the exercise.
4.3 The same mapping applies to a merger or acquisition
Draft Direction uses the same machinery for a different trigger. Upon an acquisition, merger, or amalgamation of regulated entities, in whole or in part, the transferee must undertake a one-time mapping of the loans transferred to it and determine the applicable interest rate, including the benchmark and spread, in accordance with its own policy.
The same no-disadvantage constraint applies, measured against a different reference point: the revised rate may not exceed the rate that applied to the borrower with the transferor immediately before the transaction.
The existing bank Directions address only a narrow version of this. Takeover of branches in rural and semi-urban centers, where transfers are on mutually agreed terms and existing borrowers may not be disadvantaged and may choose to stay with either bank.
4.4 If a benchmark disappears mid-loan
Where a benchmark is discontinued during the term of a floating-rate loan, the lender must change the benchmark without putting the borrower at a disadvantage on the rate. The draft adds that the lender may incorporate a fallback mechanism into the loan agreement to address such a scenario.
The permission is worth reading; it requires that the benchmark, reset periodicity, and reset date be stated explicitly in the loan agreement, alongside the widened definition of the external benchmark. A lender that does not build a fallback into its documentation will be managing a discontinuation without one.
Impact
- Legacy book no longer falls out of scope by amortizing; it has to be actively remediated on a fixed timetable. For most lenders, this will be the largest workstream the draft creates, and it runs in parallel with originating new business on the new framework.
- The economics of the migration are asymmetric by design. The rate cap runs one way only, so every migrated loan is at worst neutral to revenue and potentially negative. There is no mechanism by which the exercise improves the yield on the existing book, and no ability to recover its cost from borrowers.
- Because the cap is fixed at the rate applying immediately before each loan’s transition, the sequencing of the migration determines the ceiling each loan carries afterward. Migrating a portfolio in a month when rates are low locks in a lower permanent ceiling than migrating it later. Sequencing becomes a commercial decision, not purely an operational one.
- Consent turns a compliance deadline into an outreach exercise. The duty to complete the migration rests with the lender, while the decision to consent rests with the borrower, across the whole benchmark-linked book. Contactability, documentation, and evidence of consent become the binding constraints rather than the rate mechanics.
- Merger and acquisition economics change. An acquirer inherits a mapping obligation on the target’s book with the ceiling set by the transferor’s pricing, so an acquired portfolio cannot be repriced upward on integration. That belongs in diligence and in valuation, not in post-completion planning.
5
WHERE THE COMPETITIVE MAP MOVES
This section deals with the consequences: who competes with whom, how joint arrangements work, and the products in which the draft imposes limits rather than mechanics.
The draft applies one set of pricing rules to commercial banks, regional rural banks, urban and rural co-operative banks, all-India financial institutions, and NBFCs, including housing finance companies. The substantive constraints become common: no loan may be priced below its applicable benchmark, the calculation conventions are the same, and the small-ticket ceilings apply equally.
5.1 Co-lending and loan transfers
Transfers turn on who the lender on record is
Where a loan exposure is transferred, and the lender on record for the borrower does not change, the interest rate, including the benchmark, the spread, and, for floating-rate loans, the reset mechanism, continues to be governed by the contractual terms agreed between the transferor and the borrower. Where a new agreement is entered into between the transferee and the borrower because the lender on record has changed, the interest rate after transfer is governed by the transferee’s own framework.
So the pricing consequence of a transfer follows its legal form. Neither branch of the rule has a counterpart in the existing bank Directions.
Co-lending is brought inside the framework.
In a co-lending arrangement, a regulated entity must, in addition to these Directions, comply with the Transfer and Distribution of Credit Risk Directions applicable to it. Today, the equivalent cross-reference exists for NBFCs, but in the digital lending context rather than as a pricing rule, and the existing bank Directions do not address co-lending at all.
The significance is less in the cross-reference than in the words “in addition to”. Each co-lender must satisfy its own benchmark, floor, reset, and spread obligations on its share of the exposure, and these obligations are not identical across entity classes or proportionality tiers.
5.2 Two product-specific rules
1
Working capital: each drawdown may stand alone
Directions permit working capital demand loans, including those extended as part of a working capital facility, to be treated as separate loans for the purpose of determining the interest rate and spread, where each drawdown has a fixed tenor.
A drawdown treated as a separate loan uses the benchmark of the month in which it is sanctioned, has its own reset periodicity fixed for its tenor, and starts its own three-year clock on the non-credit components of its spread. Treating a facility as a single loan produces a very different result from treating each drawdown as its own loan. The provision is permissive, so this becomes a design choice.
2
Foreign currency lending
Direction requires that the rate on a foreign currency loan be set in accordance with the lender’s policy, with reference to a market-determined external benchmark plus a risk-based spread. For banks, that is substantially the existing position, which allows freedom under a Board-approved policy, requires reference to a market-determined external benchmark, and builds the rate by adding spread components. For non-banks, it is new.
One connected change is worth noting. Today, loans linked to a market-determined external benchmark are expressly exempted from the benchmark, spread, and reset chapters. That exemption does not appear in the draft’s list.
5.3 Limits on what a small borrower can be charged
Every regulated entity needs to set an explicit ceiling on the Annual Percentage Rate, including the interest rate and all other charges, for microfinance loans and small-value loans, while ensuring these are not usurious. A small-value loan is a personal loan to an individual with a principal amount not exceeding ₹50,000. The Reserve Bank does not set the number; the obligation is to set one, and the measure is the all-inclusive annual percentage rate as defined in the Responsible Business Conduct Directions.
Two changes are embedded here.
1
The population widens.
Today, an explicit ceiling is required only for microfinance loans. Small value and personal loans are governed by a softer standard – interest must be “justifiable having regard to the total cost incurred in extending the loan and the extent of return that could reasonably be expected”. That standard is replaced by a hard requirement to set a ceiling.
2
The measure changes.
The existing microfinance ceiling is expressed as a ceiling on the interest rate and all other charges. The draft presents it as a ceiling on the annual percentage rate, a single, all-inclusive figure that is harder to circumvent by shifting costs between interest and fees.
The draft also clarifies something that causes practical confusion: the date on which interest is applied to a loan account and the date on which repayment falls due need not coincide. A lender may set due dates under its policy and the terms agreed with the borrower.
5.4 How interest is calculated
The draft prescribes calculation conventions that the existing Directions largely leave alone. Interest is charged at monthly rests, with agricultural exceptions: annual rests for a long-duration crop, and due dates set with reference to the crop season for a short-duration crop. Interest is computed on a daily reducing balance basis. And an Actual/Actual day-count convention applies.
For banks, monthly rests and the agricultural treatment are broadly the existing position. The Daily reducing balance and the Actual / Actual convention are new and do not appear in any of the existing Directions. The existing requirement to round interest on rupee advances to the nearest rupee is not carried forward. For non-banks, all four are new; the existing NBFC Directions refer to a reducing-balance method only in the worked illustration accompanying the annual percentage rate disclosure, which is a convention for computing the disclosure rather than a rule about how interest is charged.
Impact
- Co-lending becomes a compliance intersection rather than a commercial one. Because each partner must meet its own obligations on its own share, a bank and an NBFC lending to the same borrower may face different reset caps and different spread-revision constraints on the same exposure. In practice, arrangements will have to be built to the stricter of the two.
- The legal form of a transfer now determines its pricing outcome. Where the lender on record remains unchanged, the original contract survives, and the transferred pool retains its pricing; where a new agreement is entered into, the transferee’s framework applies, and the rate may change. Structuring decisions that were previously about servicing and accounting now carry pricing consequences.
- Working capital can be re-architected deliberately. Because reset periodicity, the three-year spread lock, and the benchmark month all attach at the loan level, the choice to treat each fixed-tenor drawdown as a separate loan materially changes how flexibly a facility can be repriced. The provision is permissive, so it rewards lenders who think about it rather than defaulting.
- Small-ticket pricing acquires a hard ceiling where it had a soft standard, and the ceiling is set on an all-inclusive measure. Moving costs between interest and fees no longer help. The obligation applies equally to banks and non-banks, removing what is today a structural difference in how the two are held to account for personal lending under ₹50,000.
- Calculation conventions become uniform. Daily reducing balance and an Actual/Actual day count remove a quiet source of yield differences between lenders and between products, and both require system changes. For non-banks, monthly rests and the bar on compounding before a payment is overdue are additional.
6
PREPAREDNESS ROADMAP
This section consolidates the work implied by Sections 2 to 5 into seven workstreams and three phases. It assumes the draft is adopted broadly as published; where it is not, the sequencing holds even if individual requirements move.
A The three phases
The draft sets two dates, creating three distinct periods of work rather than two.
| Period | What it is | What has to be finished by the end of it |
|---|---|---|
| Now to 31 March 2027 | Design and build. No new obligation is live, but everything has to be ready. | Benchmark computation built and parallel-run; Board policy approved; spread architecture recalibrated; systems changed; documentation templates redrafted; back-book inventory complete |
| 1 April 2027 | Go live with the new business. The first benchmark is published on this date. | All new lending is priced on the new benchmark, floor, reset, and spread rules; the monthly publication begins; dual running of two frameworks starts |
| April 2027 to 31 March 2029 | Migrate. Two frameworks operate side by side throughout. | Every existing loan linked to any internal or external benchmark migrated, with borrower consent, at no charge and at no higher rate. |
B Seven workstreams
01 Benchmark construction and publication
- Build the marginal cost of funds calculation basis fresh deposits and fresh borrowings raised in the month, interest actually expended on each, annualized and weighted, then averaged over three months.
- Map the source data and fix the evidence chain. The draft requires the computation to be system-generated and independently verifiable, so decide now who owns the number, who reviews it, and what the audit trail looks like.
- Parallel-run for several months to quantify two things separately: how far the new benchmark sits below the present one, and how much more it moves month to month on a fresh-flow basis.
- Decide how many internal benchmarks will be maintained and document the basis for each, because the floor operates by reference to the benchmark applicable to the particular loan.
- Decide what replaces the refinements that do not carry forward the core-balance treatment of current and savings deposits, swap and hedge costs on foreign currency funding, the bond-yield proxy for long-term borrowings, the floating-rate term deposit rule, and the seven-day lag option.
Where banks and non-banks differ
Commercial banks, regional rural banks, Tier 3 and 4 urban co-operative banks, and rural co-operative banks with assets above ₹1,000 crore must publish the benchmark itself on the first calendar day of every month and retire the tenor-linked MCLR series and the maturity-bucket machinery. NBFCs, housing finance companies, all-India financial institutions, and smaller co-operative banks must instead build the benchmark from scratch and publish the methodology on their digital interfaces or at branches where they do not have one.
02 Spread architecture and pricing governance
- Rebuild the spread to absorb operating cost and term premium and decide explicitly where the cost of capital and the reserve drag are now recovered.
- Define loan categories and have them approved. The draft allows them to be built on product, on borrower category, on a combination, or on whether the rate is fixed or floating.
- Set the methodology for the quantum of each spread component and the range of spread for each category, and confirm the credit risk premium is positive on every loan.
- Confirm no fee or charge is being counted as spread, since the draft defines spread to exclude them; a fee cannot be used to lift an at-benchmark loan above the floor.
- Build the clock. Non-credit spread components cannot move in either direction for three years, measured from the first disbursement or the last revision. Hence, the tracking has to sit at the loan and category levels and survive both increases and decreases.
- Build the retention-concession path under the draft’s exception: justifiable grounds, non-discriminatory, within policy. Make sure the approver knows it restarts the three-year clock.
- Refresh the Board policy to cover methodology, the definition of the internal benchmark, spread components, loan categories, and delegation of pricing powers, and put it on an annual review cycle.
03 Reset mechanics and balance sheet management
- Choose a reset periodicity for each product within the three-month cap, knowing it is fixed for the life of every loan written against it.
- Reconfigure reset dates to the first calendar day of the month in which the reset falls due, or to the agreement due date where the periodicity is shorter than a month.
- Re-model the repricing gap and earnings sensitivity for quarterly resets, landing on concentrated dates rather than spread across the year by disbursement date.
04 Systems, data, and calculation
- Confirm monthly rests, and that interest compounds only after a repayment becomes overdue.
- Build or extend the annual percentage rate engine, since the ceilings on microfinance and small-value loans are set on an all-inclusive annual percentage rate.
- Build benchmark publication, versioning, and retention; loan-level reset scheduling with immutable periodicity; and spread-clock tracking.
05 Documentation and contracts
- Redraft loan agreements to state the benchmark, the reset periodicity, and the date of reset explicitly.
- Add a fallback mechanism for benchmark discontinuation. The draft permits rather than requires it, so the choice is between contracting for it now and managing a discontinuation without a contractual answer.
- Make the terms on which the credit risk premium may be revised consistent between the policy and the loan agreement, since the draft requires revision to be in accordance with both.
- Design the migration consent instrument and the evidence retention that goes with it, before the volume arrives.
Where banks and non-banks differ
NBFCs and housing finance companies must also keep their Responsible Business Conduct obligations aligned, as those Directions have not been repealed. Disclosure of the approach to grading risk in the application form and sanction letter, the annualized rate and its method of application, prospective-only rate changes, the Key Facts Statement and annual percentage rate disclosures, the reset options on installment-based personal loans, and the quarterly borrower statement all continue, and will now sit on top of the new pricing architecture.
06 The back book
- Inventorize every loan linked to any internal or external benchmark, and confirms what the inventory excludes and why.
- Design the mapping methodology, and capture for each loan the rate applying immediately before its transition, and that figure becomes a permanent ceiling on that loan.
- Decide the sequencing deliberately. Because the ceiling is set on the migration date, a portfolio’s position within the two-year window determines the ceiling it carries thereafter.
- Plan the consent exercise as an outreach program: contactability, channels, evidence, and a defined path for borrowers who do not respond.
- Build the mapping obligation and the transferor-rate ceiling into acquisition diligence and pricing models, since an acquired book cannot be repriced upward on integration.
Where banks and non-banks differ
For banks, the population includes the Base Rate and BPLR vintages, which the existing Directions allow to run to maturity, and that a bank is currently required to continue publishing. Both disappear, so loans priced off a benchmark first published in 2010 have to be brought across. For NBFCs and housing finance companies, the exercise is narrower and turns on which loans are, in fact, linked to a benchmark, a question worth settling early because it determines the size of the program.
07 Products, portfolios, and joint arrangements
- Re-test every product against the floor, starting with fixed-rate and long-tenor lending, bill discounting, and factoring, all of which lose an exemption they have today.
- Set the Board-level annual percentage rate ceilings for microfinance loans and for personal loans up to ₹50,000, and put the disclosure in place.
- Recompute the agricultural cap to include charges and fees as well as interest, and apply the one-year definition of short-term.
- Decide the working capital demand loan treatment – whether each fixed-tenor drawdown is priced as its own loan, with its own benchmark month, reset periodicity, and spread clock.
- Align the legal form of loan transfers with the intended pricing outcome, since the framework that governs a transferred loan turns on whether the lender on record changes.
- Reconcile co-lending arrangements partner by partner. Each lender meets its own obligations on its own share, so where the two sit in different entity classes or proportionality tiers, the arrangement has to be built to the stricter.
- Re-map the exemptions, noting they now release a loan from the Directions as a whole rather than from the pricing chapters alone.
Where banks and non-banks differ
Small finance banks and local area banks should treat this workstream as a priority, because moving into the definition of Commercial Banks brings them within the external benchmark mandate for floating-rate personal and MSME lending, which accounts for a large share of their book. For NBFCs and housing finance companies, foreign currency lending acquires an express requirement to price off a market-determined external benchmark plus a risk-based spread for the first time.
7
CLOSING THOUGHTS
Read provision by provision, the draft looks like a tidying exercise: six instruments folded into one, a simpler benchmark formula, a few dates. Read as a whole, it does something more particular. It changes where, in a lending business, the pricing decision actually sits.
Discretion is relocated, not removed
Almost every element of judgment that today lives inside the benchmark is pushed out of it. The negative carry on reserves, the return on net worth, operating costs, and the tenor premium all leave. What remains is a prescribed arithmetic on last quarter’s funding, computed from the general ledger and required to be independently verifiable. There is very little a lender can decide about it.
All of that judgment reappears in the spread. And the spread is then fenced: the credit risk premium must be positive and may move only on a documented change in credit profile; the other components cannot move for three years; fees are excluded by definition; and the whole structure has to be pre-approved by category in a Board policy.
The effect is that pricing becomes an origination discipline rather than a portfolio-management lever. The decisions that matter are the spread build, the reset periodicity, and the loan category a borrower is placed in; these are set once at the front end and are then difficult or impossible to revisit.
That is a significant change in where capability has to sit. A business that today manages margin by adjusting the back book will find that lever largely gone. The equivalent capability under the draft is getting the initial calibration right, at scale, across a category framework of one’s own design.
Convergence closes an arbitrage but squeezes the middle
The obvious reading of a single rulebook is that it levels the field between banks and non-banks. The more useful reading is that it changes what a lender competes on. Once the benchmark is a prescribed function of funding costs and the floor is absolute, the two factors that determine competitive pricing are how cheaply an entity funds itself and how well it rates its credit. Both reward scale and balance-sheet strength. The largest lenders have the funding advantage.
The pressure therefore lands hardest in the middle – on entities large enough to be caught by every restriction but without a large bank’s cost of funds. Middle Layer NBFCs, Tier 3 and 4 urban co-operative banks, regional rural banks, and small finance banks are in that position, and small finance banks additionally pick up the external benchmark mandate for personal and MSME lending, which makes up much of their book.
Faster transmission, fewer shock absorbers, and a front-loaded bill
Every floating-rate loan outside the carve-outs will reflect its benchmark within a quarter, with the benchmark itself moving with last quarter’s funding flows. Policy and funding-cost changes will pass through to borrowers faster than they do today. That is plainly the intent.
The corollary is that the instruments a lender used to smooth that transmission are withdrawn at the same time. Annual resets, tenor-linked benchmarks, reset dates anchored to each loan’s disbursement, and weighting by the existing stock of funding rather than the month’s fresh flows: all of them absorbed volatility. All of them go. Earnings become more sensitive to the benchmark precisely as the ability to manage that sensitivity is removed.
And the bill arrives first. Systems, documentation, governance, and the benchmark build all land before April 2027. The back-book migration then runs for two further years, cannot raise any borrower’s rate, and cannot be charged for. There is no provision in the draft by which a lender recovers any part of that cost. The exercise should be planned as an investment with a compliance return, not as a project with a payback.
Three questions worth putting to your own organization
These are the questions whose answers will determine whether the transition is managed well or merely completed.
1
If our spread framework had to be explained, component by component, to a borrower and to a supervisor, could it be?
The draft’s architecture assumes a price can be decomposed and each part justified.
2
Who owns the benchmark?
It is a single monthly number that sets the floor beneath every loan, has to be independently verifiable, and will visibly fall when the framework changes. Treasury, Finance, and Internal Audit all have a claim. Leaving that unresolved will stall the workstreams that depend on it.
3
Are we treating the two-year migration window as a queue or as an opportunity?
Every legacy loan has to be touched, consent obtained, and a permanent rate ceiling fixed at the moment it moves. A portfolio that is going to be handled loan by loan anyway is also one that can be re-segmented, re-categorized, and re-documented in the same pass.
A final practical point. This is a draft issued for comments, with an effective date two years out, and individual parameters may well change: the reset cap, the length of the spread lock, the thresholds for the proportionality carve-outs, and the small-value loan limit. The architecture is far less likely to change: a benchmark built from the marginal cost of funds, a floor beneath every loan, judgment concentrated in a documented spread, and a legacy book that has to be migrated rather than run off. Building to that architecture is a safe investment now. Calibrating to any specific number in the draft is not.
