DuPont Analysis

Anchor the ROE and name the question

This is a change-over-time question, not a peer question, so the comparison set is Apple FY2015 versus Apple FY2023 and nothing else. FY2015: revenue about 234 billion dollars, net income about 53 billion, total assets about 290 billion, equity about 119 billion, so ROE is 53/119 = roughly 45 percent. FY2023: revenue about 383 billion, net income about 97 billion, total assets about 353 billion, equity about 62 billion, so ROE is 97/62 = roughly 156 percent. Note the shape of the puzzle before you decompose: revenue grew about 64 percent and net income about 83 percent over eight years, but ROE rose by 111 percentage points. Profit growth of that size cannot mathematically produce an ROE move of that size, so something in the denominator is doing the work.

Run the three-step identity for both years and tie it back

FY2015: net margin 53/234 = 22.8 percent; asset turnover 234/290 = 0.80x; equity multiplier 290/119 = 2.43x. Multiply: 0.228 x 0.80 x 2.43 = 44.3 percent, which reproduces the stated ROE. FY2023: net margin 97/383 = 25.3 percent; asset turnover 383/353 = 1.09x; equity multiplier 353/62 = 5.68x. Multiply: 0.253 x 1.09 x 5.68 = 156 percent, again a clean tie-back. Say the tie-back out loud in an interview — it takes fifteen seconds and it is the difference between quoting a formula and controlling it.

Attribute the 111-point rise, one lever at a time

Hold two levers fixed and move the third. Margin alone from 22.8 to 25.3 percent: 0.253 x 0.80 x 2.43 = 49.2 percent, so margin contributed about 4.5 points. Then turnover from 0.80x to 1.09x: 0.253 x 1.09 x 2.43 = 67.0 percent, so turnover contributed about 18 points. Then the equity multiplier from 2.43x to 5.68x takes you the rest of the way to 156 percent, contributing about 89 points. Headline sentence for the committee: roughly 80 percent of the ROE improvement came from the equity multiplier, and only about 4 points of 111 came from higher net margin. The analyst's claim that the business is three and a half times better is false as stated.

Ask why the multiplier moved — assets up or equity down?

This is the step most candidates skip, and it inverts the usual conclusion. A rising equity multiplier normally means assets grew on borrowed money, which is fragile. Here total assets rose only about 22 percent while shareholders' equity fell from about 119 billion dollars to about 62 billion, because Apple returned more cash to shareholders through buybacks and dividends than it earned — cumulative repurchases since the programme began run into the hundreds of billions, and buybacks retire equity directly. So the multiplier rose because the denominator was deliberately shrunk, not because the company loaded up on risk. Apple did also issue term debt of roughly 100 billion dollars at its peak, partly because pre-2018 much of its cash sat offshore, but that debt is comfortably inside the gross cash pile.

Go five-step on the margin and find the tax artefact

Split net margin into operating margin x interest burden x tax burden. FY2015: operating income about 71 billion on 234 billion of revenue = 30.5 percent operating margin; pre-tax income about 72 billion, so interest burden is about 1.02 (above 1.0 because Apple earns more on its investment portfolio than it pays on debt); net income 53 billion on 72 billion pre-tax = tax burden 0.74, an effective rate near 26 percent. FY2023: operating margin about 29.8 percent, interest burden about 1.0, and tax burden about 0.85 on an effective rate near 16 percent, largely reflecting the post-2018 US statutory cut from 35 to 21 percent plus mix. Read that carefully: operating margin actually drifted slightly down, and essentially the entire net-margin gain of 2.5 points is tax, not operations. Also flag that FY2024 net income carried a large one-off EU tax charge, which is exactly why you pick a clean year when comparing tax burden across time.

Benchmark honestly and judge the quality of the return

Compare Apple only against same-shape businesses, and use ROA when the multiplier is distorted: Apple's ROA is 25.3 percent x 1.09 = about 27 percent, versus a mass-market retailer like Walmart at roughly 2 to 3 percent net margin, about 2.4x turnover and about 3x leverage, giving roughly 6 percent ROA and high-teens ROE. That comparison shows Apple's operating engine is genuinely extraordinary — but that fact is visible in ROA, not in the 156 percent ROE. Also note the turnover improvement from 0.80x to 1.09x is itself half a balance-sheet story: Apple runs almost no factories, and it has been spending down its marketable-securities pile toward a net-cash-neutral target, which shrinks assets without any operational change. On risk, interest coverage is enormous (EBIT of roughly 114 billion against interest expense in the low single-digit billions), so this is a capital-return artefact, not distress leverage.

Recast DuPont for a lender before you compute anything

The classic three-step of margin x asset turnover x leverage breaks for a financial institution, and saying so is worth real credit. For an NBFC, assets are not a cost of doing business, they are the revenue-producing product, so revenue over assets is simply the portfolio yield, and net margin over that yield double-counts the same spread. The market-standard form is ROE = ROA x Equity Multiplier, where ROA already nets interest income, interest expense, fees, opex and credit cost. State the identity explicitly: Sarvodaya 2.4 percent x 6.9 = 16.6 percent; peer 3.2 percent x 6.1 = 19.5 percent. Both tie back, so you can start attributing.

Attribute the 2.9-point gap one lever at a time

Start from Sarvodaya at 2.4 x 6.9 = 16.6 percent. Move only ROA to the peer's 3.2 percent: 3.2 x 6.9 = 22.1 percent, a favourable swing of 5.5 points. Then move only leverage down to the peer's 6.1x: 3.2 x 6.1 = 19.5 percent, an adverse swing of 2.6 points. The two deltas net to the 2.9-point gap, and the sentence that matters is this: Sarvodaya's underlying earning power is 5.5 points of ROE worse than the peer's, and it is currently hiding roughly 2.6 points of that hole by running about 13 percent more leverage. The founder's lean-cost story is not visible in the headline ROE gap because leverage is masking the real spread.

Drill ROA into its five component spreads

Break ROA into yield, cost of funds, fees, opex and credit cost, all as a percentage of average assets, and pre-tax. Sarvodaya: yield 15.0, funding cost 7.8, so NIM 7.2; fees 0.5; opex 3.6; credit cost 0.9; PBT 3.2; at a 25 percent tax rate, ROA 2.4. Peer: yield 15.8, funding cost 6.9, so NIM 8.9; fees 0.6; opex 4.3; credit cost 0.9; PBT 4.3; ROA 3.2. So the founder is right about opex — Sarvodaya is 70 bps cheaper to run — but that saving is swamped by 90 bps of worse funding cost and 80 bps of lower yield.

Convert the two spread gaps into business causes

On funding, Sarvodaya is rated A+ against the peer's AA+, and roughly 68 percent of its borrowings are bank term loans priced off MCLR, with no NCD programme, no securitisation or direct-assignment desk, and no public deposit licence — that mix, not negotiating skill, is the 90 bps. On yield, about 40 percent of the book is refinance and top-up lending to existing customers, written at 200 to 300 bps below fresh used-vehicle rates to stop attrition, and the fleet mix skews to 4-to-6-year-old vehicles rather than the 8-plus-year-old, small-operator segment where the peer prices highest. The DSA-led sourcing that produces the opex saving is itself the reason yields are thin: a third-party sourcer competes on rate, an owned branch competes on relationship and turnaround time. Flag one more thing to the committee — identical 0.9 percent credit cost with a materially DSA-sourced book is suspicious, because DSA vintages typically season worse, so ask for static-pool loss curves by sourcing channel before believing it.

Judge the quality of the leverage, not just its level

A 6.9x equity multiplier means equity is about 14.5 percent of assets, which for an NBFC sits uncomfortably close to the regulatory CRAR floor of 15 percent — so there is almost no headroom to grow the book without fresh equity, and any ROE built on that leverage is not repeatable at scale. The asset-liability profile compounds the fragility: floating-rate bank borrowings against a fixed-rate loan book means a repo-hiking cycle widens the funding gap immediately while the book reprices only as it runs off. The reference case for an Indian audience is the 2018-19 IL&FS and DHFL sequence, where wholesale-funded NBFCs with thin equity and short liabilities lost access to funding in weeks. Conclusion for this lever: Sarvodaya's leverage is not a strength to be extended, it is a liability to be reduced.

Rank the fixes by size and time to close

One, fix the liability side first — build an NCD programme, start co-lending with a bank under the 80:20 structure, and run direct assignment on the seasoned book; co-lending is doubly useful because it moves 80 percent of new loans off balance sheet while retaining servicing fee income, which lifts ROA and cuts the multiplier at the same time. A 50 bps funding-cost improvement is worth roughly 37 bps of pre-tax and about 28 bps of post-tax ROA, which at 6.9x is roughly 1.9 points of ROE, achievable in 12 to 18 months. Two, reprice or shrink the low-yield top-up book and push DSA sourcing toward the older-vehicle, higher-yield segment, worth perhaps 30 to 40 bps of yield and another 1.5 to 2 points of ROE, but over 24 months as the book turns. Three, explicitly reject the tempting arithmetic answer of levering to 7.5x, because it adds about 1.8 points of ROE while eating the entire regulatory buffer.