Part 01 — Foundations & Market

Unit Economics

SME lending looks attractive because headline interest rates can be high. It becomes less magical once the income statement is decomposed. The lender earns yield, pays for funds, spends heavily on origination and servicing, absorbs credit losses, pays tax, and then hopes the remaining return on assets (RoA) is enough to justify leverage and equity risk.

A simple loan-level tree is:

portfolio yield - cost of funds = net interest margin

net interest margin + fees - operating expense - credit cost = pre-tax RoA

pre-tax RoA - tax = RoA

RoA x leverage = return on equity (RoE)

For a bank, the tree starts with cheap current-account and savings-account deposits. For an NBFC, the tree starts with borrowings, securitisation, co-lending and assignment. That difference explains most pricing differences in Indian MSME lending.

Yield

Yield is the annualised interest and fee income on the loan book. It varies sharply by product:

Product	Indicative borrower rate in practice	Why
Bank cash credit/overdraft to established MSME	9-14%	Secured, monitored, deposit relationship, lower cost of funds
Bank/NBFC LAP for business	11-18%	Property-backed, longer tenor, legal/valuation cost
Machinery/equipment finance	12-20%	Asset-backed but residual value and resale vary
NBFC unsecured business loan	18-36%	High opex, thin files, higher default risk
Merchant cash advance/settlement-led loans	18-30% equivalent	Daily/weekly collections reduce some risk, but borrower is small
TReDS invoice discounting	Often single to low double digits annualised	Buyer/anchor risk, short tenor, competitive financier bidding

These are market ranges, not regulatory caps. RBI generally requires transparent pricing, board-approved interest-rate policy and fair disclosure rather than prescribing one MSME lending rate. Published lender policies show the stack. Kinara Capital discloses an indicative base rate built from 14.5% weighted-average borrowing cost, 7.5% operating cost and 4.0% base RoA, producing a 26.0% benchmark, with loan rates from 14% to 36% reducing balance (Kinara Capital interest-rate policy). Five-Star Business Finance said its FY2024-25 incremental seven-year loan rates were reduced to 21.5-22.5% depending on risk profile, from 24.5%, and its average ticket size was ₹3-5 lakh (Five-Star Business Finance FY2024-25 directors’ report).

Cost of funds

For PSBs and private banks, cost of funds is largely deposit cost plus treasury transfer pricing. For NBFCs, it is a blended cost of:

term loans from banks and financial institutions;
non-convertible debentures (NCDs);
market-linked debentures and commercial paper for stronger issuers;
securitisation or direct assignment of pools;
co-lending/co-origination with banks;
SIDBI refinance or bank lines linked to PSL strategy;
external commercial borrowings for eligible, stronger lenders.

NBFC funding is confidence-sensitive. A lender with granular secured pools, low delinquency and a strong rating can borrow materially cheaper than a young unsecured fintech NBFC. Rating reports often focus on this. CRISIL’s March 2025 rationale on U GRO Capital noted funding diversification, public-sector bank funding, external commercial borrowings, and co-lending/co-origination at 44% of assets under management as of December 2024 (CRISIL Ratings, U GRO Capital, 12 March 2025).

Co-lending changes the unit economics because the bank funds most of the loan at bank cost while the NBFC sources, services and keeps a smaller share plus fees/spread. It can improve RoE if the NBFC’s sourcing and collections costs are under control. It can also create operational complexity: split ledgers, escrow, borrower communication, regulatory reporting, delinquency allocation and partner reconciliation. See co-lending business models for the operating mechanics.

Operating expense

Opex is why a ₹3 lakh MSME loan cannot be priced like a ₹3 crore corporate loan. Acquisition may involve field sales, DSA payout, bureau pulls, GST pulls, bank-statement analysis, KYC, document collection, field investigation, FI/RCU checks, tele-verification, legal/valuation for secured loans, and repeated follow-up for missing documents.

A high-quality digital flow can reduce opex, but it does not eliminate it. Someone still handles exceptions: Aadhaar/PAN mismatch, proprietor current account in personal name, GST inactive, Udyam classification mismatch, banking password failure, bounced mandate, undelivered notice, or borrower not reachable after first EMI bounce. The lowest opex is usually in anchor-led or repeat lending where the lender already sees settlement, invoice or transaction data.

As a practical benchmark, unsecured micro-SME NBFCs often need 5-9% of average assets in opex at scale; secured branch-led lenders can be lower if ticket sizes are larger; banks allocate opex differently because branch and relationship costs are shared across liabilities, payments and loans. A lender that prices at 22% but spends 8% on opex and 5% on credit cost is not earning a 22% spread.

Credit cost

Credit cost is the annualised loan-loss charge: write-offs plus provision movement, net of recoveries. It is not the same as gross non-performing assets (GNPA). A portfolio can show low GNPA because of write-offs, restructuring, sale, or fast collections; credit cost reveals the P&L damage.

MSME credit cost varies by segment:

Secured LAP SME books may run low credit cost in benign periods, but recoveries are slow and legal-cost heavy.
Unsecured business loans can produce attractive yield but can move from 3-5% normalised credit cost to double digits when underwriting weakens or a vintage goes bad.
Very small enterprise loans can show high early bounce but recover through field collections; roll-rate design matters.
Invoice finance has lower borrower-default risk when buyer acceptance is strong, but fraud, dilution and buyer dispute risk remain.

SIDBI’s July 2026 MSME Pulse is a useful warning: unsecured business-loan accounts originated in March 2025 had 2.9x higher “ever 90+ days past due within 12 months” delinquency than the overall level, and the ₹2 lakh to ₹10 lakh enterprise exposure segment had 2.1x higher delinquencies (SIDBI MSME Pulse, July 2026). A lender chasing NTC borrowers purely for growth can quickly convert yield into credit cost.

RoA and RoE examples

Consider an unsecured NBFC business-loan book:

Item	Example
Portfolio yield and fees	25.0%
Cost of funds	12.0%
Net margin	13.0%
Operating expense	6.5%
Credit cost	4.5%
Pre-tax RoA	2.0%
Tax	0.5%
RoA	1.5%

At 4x assets/equity, this is a 6% RoE. If the same lender lowers credit cost to 2.5% and opex to 5.5%, RoA becomes 3.5% and RoE about 14%. If credit cost spikes to 9%, equity returns disappear. The business is therefore a risk-and-operations business, not simply a high-rate business.

A secured SME lender may look different: 15% yield, 9.5% cost of funds, 2.5% opex and 1.0% credit cost can still produce acceptable RoA because ticket sizes are bigger and losses are collateral-supported. A bank may lend at 11% because its marginal funding cost and opex allocation are lower, and because PSL, deposits and current-account float improve relationship economics.

Funding and ALM discipline

NBFCs do not create deposits; they transform borrowed money into loans. Asset-liability management (ALM) therefore matters. Funding five-year LAP with one-year bank lines creates rollover risk. Funding short-tenor merchant loans with longer NCDs is safer but may reduce margins. Securitisation and direct assignment release liquidity but require clean pools, seasoning, credit enhancement and investor confidence.

The strongest MSME lenders treat treasury as a product function. Pricing grids should update when borrowing cost changes. Sanction systems should know whether a loan is intended for own book, co-lending, assignment or guarantee cover. Collections systems should report vintage roll rates quickly enough to adjust scorecards before a bad cohort becomes a P&L event.

For a deeper, data-backed extension of this page, read How Lenders Make Money for revenue architecture and worked P&L structures, Cost Structures & What the Data Says for lender benchmarks, and Evaluating Growth: The AUM Math and Rational CAGR for growth scenario formulas.