NBFC financial ratios

Ten KPIs follow, each computed on Chenab Finance (synthetic, ₹ crore), a diversified retail NBFC that's ~60% secured / 40% unsecured. Here's its year, compactly.

• P&L: interest income 3,270, interest expense 1,510, NII 1,760, other income 300 (of which 120 is upfront DA gains), opex 780, PPOP 1,280, impairment 460, PBT 820, tax 205, PAT 615.
• Averages: AUM 23,000, total assets 24,300, borrowings 18,000, net worth 4,100.
• Closing balance sheet: gross loans 23,500 (Stage 3 of 590, total ECL provision 500, of which Stage-3 ECL 295), cash + liquid investments 1,600, other assets 500 = 25,100. These are funded by borrowings 19,600 (banks 8,800, NCDs 6,700, CP 2,300, ECB/securitization/sub-debt 1,800), other liabilities 1,100 and net worth 4,400.
• Capital: Tier 1 4,150, Tier 2 250, risk-weighted assets 22,300.
• AUM: opening 20,000, closing 26,000 (on-book 23,000 + assigned/co-lending pools 3,000), disbursements 13,500, write-offs 400.
• HQLA 850 against 30-day stressed net outflows of 800.

Here's the panel, with the thresholds you must know cold. They're rules of thumb, so calibrate them to segment and cycle as each entry below explains:

How do you read AUM growth and margins?

KPI 1: AUM growth (15 to 25% healthy, >40% risky). AUM = on-book loans + off-book pools the lender originated and services. For Chenab, (26,000 − 20,000)/20,000 = +30%, which puts it in the watch zone. Always rebuild the bridge: opening 20,000 + disbursements 13,500 − collections/runoff 7,100 − write-offs 400 = 26,000 ✓. The bridge exposes three lies a headline can tell. Growth can be manufactured by slowing runoff (tenor extensions, top-ups, moratoria, the book aging in place). It can be parked off-book to flatter gearing. It can also be disbursement-led growth into segments where the lender has no loss history. Why is >40% risky per se? Because at 40%+ three clocks run at once. Unseasoned vintages dominate the book before their losses can surface, the same denominator effect you saw with banks. The liability stack has to compound just as fast, usually through whatever money is easiest, which is CP. And adverse selection sharpens, since the marginal borrower a fast grower wins is often one a seasoned peer repriced or refused. Bajaj Finance compounded AUM ~25 to 30% for a decade with a deposit-and-NCD-heavy stack and got away with it. Almost every >40% grower of the 2016-18 cohort didn't.

KPI 2: NIM. NII ÷ average AUM. State your denominator, because some houses use on-book assets, which is fine as long as you're consistent. Chenab's NIM is 1,760/23,000 = 7.7%. Read it against the segment. A NIM of 2.5 to 3.5% is excellent for a prime HFC and catastrophic for a microlender, and the segment table gives the bands. Its trend tells you more than its level. A NIM falling while AUM accelerates is the classic "buying growth" signature.

KPI 3: Spread and risk-adjusted spread. Yield = 3,270/23,000 = 14.2%. CoF = 1,510/18,000 = 8.4%, so the spread is 5.8%. Risk-adjusted NIM = NIM − credit cost = 7.7 − 2.0 = 5.7%. This number survives contact with reality, and it's the only fair way to compare a 22%-yield microlender with a 9.5%-yield HFC. Worked example 2 makes that comparison. A spread squeeze, with CoF rising faster than yield, is the market repricing the NBFC's own credit faster than the NBFC can reprice its borrowers. It's the red flag that comes before the others.

How do funding mix and asset quality show risk?

KPI 4: Cost of funds & borrowing mix (CP-heavy = refinancing risk). Chenab's mix is banks 45%, NCDs 34%, CP 11.7% and other 9%, which puts CP in the amber band. The CP rule deserves its own paragraph, because the instrument is uniquely unforgiving. A 60 to 90 day tenor means a ₹2,300 stock at ~75-day average maturity forces ~₹11,000 of rollover events per year. That's eleven thousand crore of annual opportunities for someone to say no, with zero forgiveness. CP has no covenant-waiver dance. You pay at par on the day or you're in default. CP is also cheap, 50 to 150 bp inside NCDs, and the cheapness is the trap, because it converts term-premium savings into rollover risk. Read the mix with maturity and concentration in mind. Watch for reliance on a single channel. Even 45% from banks is concentration, and bank appetite for NBFC paper moves with the regulator's risk weights. Then check the rating rationale for banker names and undrawn lines. The thresholds: CP <10% comfortable, 10 to 20% watch, >20% red. For a lender with long assets, such as an HFC, even 10% deserves suspicion, while a gold lender with 4-month assets can defend more.

KPI 5: GNPA / Stage 3 and coverage. Stage 3 ratio = 590/23,500 = 2.5%. Stage-3 coverage = 295/590 = 50%, and net Stage 3 = 295/23,205 = 1.3%. Bands depend on the segment. A 2.5% ratio is mediocre for prime housing. It's respectable for vehicle finance, and excellent for microfinance in a bad year. Then run the NBFC-specific checks. Look at coverage by stage, because a lender disclosing only blended coverage is hiding the stage-2 pile. Look at how aggressive the write-off policy is. Fast write-offs flatter GNPA, so rebuild the roll-forward the way you did for banks. And look at the impairment reserve. A big one means the ECL model runs lighter than the IRAC floor.

KPI 6: Credit cost. Impairment ÷ average AUM = 460/23,000 = 2.0%. Judge it against the yield premium the book earns. Chenab collects ~500 bp over an HFC's yield and pays ~170 bp of it in credit cost, so the spread survives. The failure pattern is credit cost eating the entire yield premium at the bottom of the cycle, and unsecured books are where you'll see it. Always pair this KPI with growth. Credit cost falling while growth runs >30% is the denominator effect wearing makeup. Don't mistake it for underwriting genius.

What do gearing and capital adequacy tell you?

KPI 7: Gearing (3 to 6× ok, >7 to 8× risky). Borrowings ÷ net worth = 19,600/4,400 = 4.5×, comfortable. The band exists because gearing is the fragility multiplier you met in the business model. At 4.5×, a loss of 3% of loans ≈ 15% of net worth, which is painful but survivable. At 8× the same loss is ~27% of net worth. Then the rating goes, and the funding goes with it. There's also the prime-HFC exception. Individual mortgages lose so little (through-cycle credit cost of 20 to 50 bp) that 8 to 10× has historically been tolerated. But the exception belongs to the asset class, and a sector label doesn't earn it. An "HFC" running a 30% developer book at 9× gearing, as DHFL did, has borrowed the leverage tolerance of an asset class it no longer holds. Wherever you see the exception claimed, re-underwrite the asset mix first.

KPI 8: CRAR (min 15%, check the current Master Direction). CRAR = (4,150 + 250)/22,300 = 19.7%. Tier 1 = 18.6%, against floors of 15% total / 10% Tier 1 for the middle layer up. Now the growth headroom. Surplus capital = (19.7% − 15%) × 22,300 ≈ 1,050, which at a 15% requirement supports ~7,000 of new RWA. Call it ~31% asset growth before any retained earnings, or ~45% with a year's retention. So capital doesn't bind Chenab. Liquidity binds first, and it almost always does. That asymmetry is the most misread fact in the sector. IL&FS group entities and DHFL both reported CRARs comfortably above the floor within months of dying. Capital measures loss-absorption. It says nothing about whether Tuesday's CP gets paid, so never accept a capital ratio as an answer to a liquidity question.

How do you read returns and liquidity?

KPI 9: ROA (>2.5 to 3% strong) and ROE. ROA = 615/24,300 = 2.5%, or 2.7% on average AUM, so state the convention you use. ROE = 615/4,100 = 15.0%. Leverage is why the threshold is set where it is. A bank manufactures 15% ROE from 1.3% ROA at 11×, while an NBFC has to reach it from ~2.5 to 3% ROA at 4 to 6×. That's the business model working as designed. So an NBFC printing 15% ROE from 1.5% ROA is running 9 to 10× leverage, and you should go straight back to KPI 7 and the ALM note. Run a quality check before applauding Chenab's 2.5%. Strip out the upfront DA gains (120 × 0.75 = 90 post-tax) and ROA drops to 2.2%. Still real, just flattered.

KPI 10: ALM gaps and LCR. Chenab's first-month cumulative gap is +4% of outflows. It's positive because Chenab's book is shorter than Tapti's, and all its buckets are inside board limits. LCR = 850/800 = 106% ✓. Classify it as a pass. This row comes last on the panel but first in the decision order, because a lender that fails here is uninvestable whatever the other nine say.

Which regulatory layer does an NBFC belong to?

The SBR row. The last row is a classification, and it has consequences. Chenab is non-deposit-taking with assets of ₹25,100 crore > ₹1,000 crore, so it's Middle Layer. That brings 15% CRAR, the full ALM/LCR framework and board-approved limits. The layer a lender belongs to tells you which rulebook you may rely on, and how deep its disclosure will go.