Strip the model to its skeleton, the way you stripped a bank. An NBFC borrows at ~8 to 9%, above the risk-free curve, since the NBFC is itself a credit to the people lending to it. It lends at 14 to 18% to borrowers banks won't touch, and runs the spread across a balance sheet four to six times its equity. Set the two lenders side by side, per ₹100 of assets. The bank column is the stylized bank from the banking playbook, and the NBFC column is a stylized diversified retail lender:
| Per ₹100 of assets | Bank | NBFC |
|---|---|---|
| Loans | 65 | 90 |
| Investments + cash | 35 | 8 |
| Other assets | — | 2 |
| Funded by: deposits | 80 | — |
| Borrowings | 9 | 80 |
| Other liabilities | 2 | 2 |
| Equity | 9 | 18 |
And the P&L each throws off in a normal year:
| Per ₹100 of assets | Bank | NBFC |
|---|---|---|
| Interest income | 7.40 | 14.00 |
| − Interest expense | 4.20 | 6.60 |
| = Net interest income | 3.20 | 7.40 |
| + Other income | 1.20 | 1.00 |
| − Operating expenses | 2.00 | 3.40 |
| = PPOP | 2.40 | 5.00 |
| − Provisions / impairment | 0.50 | 1.50 |
| = PBT | 1.90 | 3.50 |
| − Tax (~25%) | 0.48 | 0.88 |
| = PAT | 1.42 | 2.62 |
| ROA | 1.42% | 2.62% |
| × Leverage (assets/equity) | 11.1× | 5.6× |
| ROE | 15.8% | 14.6% |
Advertisement
Why does an NBFC earn the same ROE as a bank?
Read the two columns until the symmetry clicks. The NBFC earns nearly twice the bank's ROA and arrives at the same ROE, because it can only safely run half the leverage. It has to. Its raw material costs ~240 bp more, since it buys wholesale money where the bank has deposits. Its customers pay ~500 bp more because they're riskier or less bankable. Its opex is heavier, because it collects in the field where a bank has branch queues. And its credit losses run 3× the bank's. Every line is the same machine with the risk dial turned up on both sides of the balance sheet at once. That's what "fragility squared" means here. The assets are riskier and the liabilities are flightier, and the only shock absorbers are the thicker equity slice and the fatter spread.
What are the structural consequences of this model?
That gives three structural consequences, and they mirror the three you met for banks:
1. The liability side is the franchise, and the moat if there is one. A bank's moat is CASA. An NBFC has no CASA, so its funding "franchise" is subtler. It's the willingness of banks, bond funds and insurers to lend to it, in size, at fine spreads, in a bad month. That willingness is built from decades of clean repayment and from granular retail books that can be securitized on demand. Conservative disclosed ALM helps, so does a AAA/AA+ rating defended like territory, and, bluntly, so do perceived too-important-to-fail relationships. Measure it the way you measured CASA, by cost of funds against peers and by how it behaves in stress quarters. In the crisis half-year after IL&FS the strongest franchises saw borrowing costs rise ~30 to 50 bp. Weak ones repriced 150 to 250 bp or lost access entirely. Both ranges are approximate, and the lesson is the gap between them more than either level. A lender whose CoF advantage survives a systemic squeeze has a real moat. One priced off its parent's name has a borrowed one, and one dependent on CP has none.
2. Confidence is still the foundation, but with no fire brigade. With banks you learned the Diamond-Dybvig run, where liabilities repayable at par on demand fund illiquid assets. For an NBFC the run moves slower, but it's more certain. Nothing is repayable on demand. Still, ₹2,000 crore of CP matures on Tuesday, and the mutual fund that holds it answers to its own investors. It has no deposit insurance, no RBI window and no obligation to roll. A depositor run needs panic. A wholesale run needs only prudence, with each lender deciding for itself, "why should I be the last one holding this paper?" So NBFC funding crises are set off by information events (a default elsewhere, a rating action, a newspaper exposé) and they run their course in weeks. Banks enjoy corrective machinery, insured depositors who don't run and a central bank that lends against collateral, and an NBFC has none of it. The ALM statement exists to answer the only question that matters: if new money stopped today, how long does the company live?
Why is fast growth a red flag for an NBFC?
3. Growth is a liability decision as much as an asset decision. For a bank, growth is underwriting, the master instinct of the banking playbook. For an NBFC, growth is underwriting plus a bet that somebody will fund the book's growing refinancing needs forever. Every extra ₹100 of AUM at 4.5× gearing means ₹80+ of new wholesale borrowing, which has to be raised and then rolled over again and again for the life of the asset. So the KPI panel treats AUM growth above ~40% as a red flag in itself. At that pace the lender is seasoning untested vintages and compounding its dependence on the kindness of markets, both at once. Indian analysts of the 2018 vintage learned to ask growth questions in liability language: who funds this growth, at what tenor, and what happens to the plan if they stop?