Every non-financial business you've analysed so far has the same deep structure. You raise capital and buy assets, then sell a product whose price exceeds its cost. How much debt and how much equity you use is a separate, second-order bit of plumbing. A bank turns that around. Damodaran puts it in one line, which you met in the special-situations capstone. For a bank, debt isn't financing. It's raw material.
How does a bank earn money from the spread?
Strip a bank to its skeleton. It buys money at one price (deposits at ~4–5% blended in India today), transforms it (maturity, credit assessment, diversification), and sells money at a higher price (loans at ~9–10%). The gross margin on this trade, the spread, is thin: 3–4 percentage points. No one gets rich earning 3% on their money. The entire business model is to run that thin spread across a balance sheet ten to twelve times larger than the owners' capital. A stylized healthy Indian private bank, per ₹100 of assets:
Advertisement
| Balance sheet (per ₹100 of assets) | Funded by | ||
|---|---|---|---|
| Loans (advances) | 65 | Deposits | 80 |
| Investments (mostly G-secs) | 25 | Borrowings | 9 |
| Cash, CRR balances, other | 10 | Other liabilities | 2 |
| Equity | 9 |
And the P&L that balance sheet throws off in a normal year:
| P&L (per ₹100 of assets) | ₹ |
|---|---|
| Interest income (loans ~9%, investments ~7%) | 7.40 |
| − Interest expense (deposits + borrowings, blended ~4.7%) | 4.20 |
| = Net interest income (NII) | 3.20 |
| + Other income (fees, distribution, treasury) | 1.20 |
| − Operating expenses (branches, people, tech) | 2.00 |
| = Pre-provision operating profit (PPOP) | 2.40 |
| − Provisions (credit cost) | 0.50 |
| = Profit before tax | 1.90 |
| − Tax (~25%) | 0.48 |
| = PAT | 1.42 |
ROA = 1.42%, which is a pathetic margin by FMCG standards. But equity is only ₹9, so ROE = 1.42 × (100/9) ≈ 15.8%. That multiplication, ROA × leverage = ROE, is the entire industry.
Now run it in reverse, because the market always does. Suppose just 1.5% of assets goes bad beyond provisions, say loans worth 1.5 are written off. That loss equals 17% of equity. A 9% asset-value error, one bad corporate cycle, wipes the owners out entirely.
Leverage is the product, and leverage is symmetric.
What are the three consequences of financial leverage?
Three consequences follow, and they organise everything below:
1. Deposits are raw material, and the franchise is the moat. A bank's cost of goods is its cost of funds. A bank that gathers current and savings accounts (CASA) pays 0% and ~2.5 to 3% respectively. A bank that has to buy bulk term deposits and market borrowings pays 7%+. On a 10×-levered balance sheet, a durable 100 bp funding advantage is roughly a 10-percentage-point ROE advantage, before competition claws any of it back. So analysts obsess over CASA ratios and deposit market share the way they obsess over brand strength at Nestlé. It's the same thing, brand strength expressed in basis points. The switching costs are real: salary accounts, auto-debits, habit. And the trust takes decades to build. That's a Phase 4 moat, and you can measure it.
2. Confidence is the foundation of the balance sheet. Deposits, the raw material, are repayable on demand at par, while the loans a bank makes are illiquid and long-dated. That maturity transformation is socially valuable and permanently fragile. If enough depositors ask for their money at once, no solvent bank can pay, because the assets can't be turned into cash at par on a Tuesday. Economists call this the Diamond-Dybvig problem. Once a depositor suspects others will run, running is the rational thing to do, so runs fulfil themselves. "The bank is fine" can be true on Monday and false by Friday with no change at all in the loan book. Society has patches. There's deposit insurance, with DICGC cover of ₹5 lakh per depositor per bank in India and $250,000 through the FDIC in the US. There's the central bank as lender of last resort, and there's liquidity regulation (LCR). They reduce the fragility without removing it. SVB, in the by-the-numbers worked example, is what happens when the patch doesn't cover the depositor base.
3. FCFF fails, so the valuation toolkit changes. Why can't you use the usual tools? Recall the three-line argument from the special-situations capstone. (a) Interest expense is a bank's cost of goods sold. So "operating income before financing costs" (EBIT) doesn't exist in any meaningful form, and without EBIT there's no FCFF and no sensible EV or EV/EBITDA. (b) "Capex" and "working capital" can't be defined. A bank reinvests by growing its loan book and the regulatory capital that has to come with it. (c) Cash and debt can't be netted. They're inventory and raw material. So you value the equity directly. Either you discount the dividends the capital rules permit (payout = 1 − g/ROE), or you take book value plus the present value of excess returns ((ROE − r) × book). And you price on P/B read against ROE, never P/E alone and never EV/anything. You'll see the full machinery run in the P/B against ROE valuation.
One more reframe before the anatomy. For an ordinary company, growth is presumptively good news. For a bank, growth is underwriting, and you can't see underwriting quality for years. The credit-cycle section explains why. A bank can grow loans 40% a year for five years just by saying yes to borrowers other banks turned down. Revenue is the one thing a lender can always buy. Hold on to that thought. Of all the instincts this playbook tries to build, it's the most important.