Variant Perception

Where We Disagree With the Market

The market is anchoring its ROE skepticism to the wrong lever. Sell-side targets cluster at ₹17–21 (median ₹19.45 from 10 brokers — below the ₹22.30 spot) and price Yes Bank as a 7% ROE bank whose path to peer-level returns runs through a CASA-driven NIM walk that is structurally hard. Our reading of the report's evidence says the cleanest path from 7% ROE to roughly 10% ROE runs not through NIM expansion but through cost-to-income normalisation (already firing at 460 bps a year, with another ~600 bps of room to peer best) and through RIDF run-off (a disclosed, scheduled, regulator-mandated mechanical NIM tailwind worth 30–50 bps that does not depend on closing the CASA gap). Separately, the market is binary-pricing the ₹8,415 Cr Supreme Court AT-1 verdict as either "no material impact" (management line) or full ₹14,000–15,000 Cr liability + dilutive raise (Bear), while the most likely outcome path — a structured / partial settlement absorbed inside the 13.8% CET1 — sits uncovered between the two. None of these disagreements changes the fact that this is a sub-scale franchise with five live legacy legal threads; they change which signals a PM should weight when they resolve.

Variant Perception Scorecard

The scorecard captures a focused variant view, not a sweeping contrarian call. Consensus is unusually legible here — ten sell-side targets clustered ₹17–21, Alpha Spread DCF at ₹15.95, Emkay SELL at ₹17, ICICI Securities HOLD at ₹21 — which is why consensus clarity scores higher than evidence strength. The two operating-leverage disagreements (C/I trajectory, RIDF mechanics) carry strong, observable evidence and resolve on each quarterly print starting with Q1 FY27 (~17 July 2026). The AT-1 disagreement is weaker on evidence (legal outcomes are not modelable) but high on materiality, so it earns its place. The single most decisive print is Q2 FY27 (~17 October 2026) where two consecutive quarters of either confirmation or rejection of the slope land.

Consensus Map

The market's view is observable in five places — sell-side targets, valuation multiples, the price reaction to recent inflection prints, the buy-side flow into the SMBC transaction, and the language used in broker preview notes. The table compresses the readable consensus into the implied underwriting assumption embedded in each.

The cleanest read is the first row — ten sell-side targets averaging ₹19.45 against a ₹22.30 spot, with the lowest at ₹17 (Emkay SELL) and the highest at ₹24 (ICICI Sec HOLD raised TP). The desk is consistently bearish on the NIM walk. The buy-side has been more constructive — FII holding moved to 46% on SMBC accumulation and the stock has held a tight ₹17–25 range for six years — but the buy-side bid is strategic (SMBC, Carlyle/Advent exits) rather than fundamental, so the buy-side flow is not a clean consensus signal on the operating thesis. The market that prices Yes Bank's operating slope is the sell-side, and that desk is anchored to current ROE and the NIM-via-CASA narrative.

The Disagreement Ledger

Three ranked disagreements. Each is observable, quantified in the report, and resolves on a known event window. We are not contesting the broad valuation; we are contesting the source of the ROE accretion the market is pricing skepticism into.

Disagreement 1 — operating leverage as the cleaner path. A consensus analyst writing the bear summary on Yes Bank says, with reason, that NIM at 2.6% is 130–230 bps below the peer median and that CASA accretion at the rate Yes Bank has produced (35.1% versus Kotak 43%, IDFC First 47%) is too slow to materially close the spread gap inside three years. They are correct on those facts. But that bear synthesis treats ROE as a NIM-derivative when in fact ROE is the joint output of NIM, fee mix, opex, and credit cost — and Yes Bank's actual FY26 P&L shows operating leverage doing the heavy lifting (C/I down 460 bps, opex growth 4.6% versus income growth 11.4%) while NIM only moved 20 bps. If the market is right that NIM stays at 2.7–2.8% but we are right that C/I migrates to ~57–58% over three years on positive jaws, ROE walks from 7% to ~10% anyway. The market would have to concede that ROE accretion can come from a path it has not been pricing. The cleanest disconfirming signal is C/I stalling above 65% for two consecutive quarters despite revenue growth — that would prove the FY26 460 bps drop was a one-time reset, not the start of a trajectory.

Disagreement 2 — RIDF as a regulator-scheduled NIM tailwind. Consensus analysts who follow Indian banks know about RIDF — they will tell you it exists, that Yes Bank carries the PSL-shortfall penalty because of legacy under-compliance, and that it will run off over time. What they do not separately model is the size and the redeployment math: ₹27,931 Cr earning ~3% is roughly ₹600 Cr of below-market interest income relative to a redeployment at 9% advance yield; if half of that runs off into earning advances over FY27–FY28, the rate-spread alone adds ~₹300 Cr to NII without any deposit-franchise improvement. That is roughly 30 bps of NIM. Stack the saving-rate cuts already taken (~150 bps over the easing cycle, worth another ~10–15 bps of NIM in steady state as term deposits reprice down) and the mechanical floor for NIM expansion is closer to 2.9% than the sell-side's implied 2.7–2.8%. If we are right, consensus would have to concede that the durable portion of NIM expansion is regulator-scheduled rather than franchise-dependent. The disconfirming signal: RIDF balance flat or rising in any FY27 quarter despite management's guided ₹6,500–9,000 Cr run-off.

Disagreement 3 — AT-1 as continuous, not binary. This is the lowest-confidence variant view because legal outcomes are not modelable from financial evidence. But the two priced outcomes — "no impact" and "full ₹14–15,000 Cr liability + dilutive raise" — are both extreme. The Bombay HC quashed the write-down on procedural grounds (the bank did not follow the consent process set out in the AT-1 prospectus), not substantive bondholder-rights grounds, which leaves the Supreme Court room to remand with a framework or order a partial settlement rather than enforce the full ₹8,415 Cr plus ~6 years of compounding interest. Yes Bank's 13.8% CET1 versus an ~11% regulatory floor on RWA ₹3.28 lakh Cr is roughly ₹6,000–8,000 Cr of absorbable buffer. A partial-liability outcome in that range removes the ~16% book overhang without forcing a dilutive raise and unlocks the re-rating the bull case implicitly assumes will arrive only if NIM walks. If we are right, the market would have to concede that the right way to price the AT-1 binary is as a probability-weighted partial liability inside CET1, not as a tail event the disclosed balance sheet cannot absorb. The cleanest disconfirming signal is an SC ruling that enforces the full principal plus accrued interest, or one that restricts the RBI resolution authority broadly enough to force a sector-wide AT-1 re-pricing.

Evidence That Changes the Odds

Six pieces of evidence from the report that move the probability of the variant view either direction. Each one is auditable from disclosed sources and forces the consensus read to update.

The C/I and RIDF evidence (items 1 and 2) is the highest-confidence and highest-impact pair — these two together carry roughly two-thirds of the variant view. The CET1 buffer (item 4) is the structural anchor that converts the AT-1 binary into a continuous outcome. The Moody's upgrade (item 5) is corroborating, not load-bearing. The Q4 ₹341 Cr provision (item 6) is the most fragile — a second prudent buffer of similar size in Q1/Q2 FY27 would reverse the read and signal that management sees credit normalisation arriving faster than the operating leverage can compound.

How This Gets Resolved

Every resolution signal here is observable in a disclosure already on the calendar. The window concentrates between July and October 2026 — Q1 FY27 results (~17 July), the FY26 annual report and AGM (July–August), the SMBC RBI 24.99% approval expiry (~22 August), and Q2 FY27 results (~17 October). The AT-1 verdict is the wild card on an open clock.

The two operating-leverage signals (C/I and NIM/RIDF) resolve on the same quarterly clock and feed each other: a Q1 FY27 print showing C/I below 63% and NIM at 2.8%+ on attributed RIDF redeployment is two-thirds of the variant view validated in a single disclosure. A Q1 FY27 print with C/I above 65% and NIM flat at 2.6–2.7% breaks both legs at once and forces the variant view to retreat. The AT-1 timing is independent — verdict can land any session day — and resolves a separate binary that the operating-leverage view does not depend on.

What Would Make Us Wrong

The variant view rests on three fragile assumptions and we should name them before the market does.

The first is that the cost-to-income decline is structural rather than cyclical. The 460 bps drop in FY26 was helped by deposit growth (+12.1%) outpacing operating-expense growth (+4.6%) — a positive-jaws year. If FY27 deposit growth slows to 8–10% as the SMBC-driven anchor effect normalises, and opex growth re-accelerates as the bank ramps branches (82 added in FY26, similar pace likely in FY27), the C/I improvement could stall. The historical Yes Bank operating pattern under Prashant Kumar showed C/I oscillating between 71% and 74% for three years before the FY26 break — so the case for "structural" is one strong data point, not a multi-year trend yet. If C/I prints above 65% in either Q1 or Q2 FY27 despite revenue growth, the variant view has to scale back to "C/I is one of several modest tailwinds" rather than "C/I is the lever."

The second is that RIDF run-off translates cleanly into NIM expansion. The runoff itself is mechanical, but the redeployment is not — Yes Bank has to find ₹6,500–9,000 Cr of profitable advance growth in FY27 to convert the freed balance into earning assets. If credit growth slows or the bank reaches for unsecured retail to deploy the cash (where credit cost would rise), the spread benefit gets partially offset by higher provisioning. Management's stated 13–15% loan-growth guidance is consistent with absorption, but execution at this pace has been the bank's weak point for half a decade. If retail slippage breaks above 3.5% in any FY27 quarter — Warren's named disconfirming signal — the redeployment math gets uglier than the headline NIM gain.

The third is the assumption that the Supreme Court will deliver a continuous-outcome judgment rather than a binary one. The Bombay HC's procedural-grounds quash gives the SC room to remand, but it does not constrain it; the Court could equally enforce the full principal-plus-interest, especially if the bondholders' counsel argues that the procedural defect was substantive rather than technical. If the SC enforces the full ₹14,000–15,000 Cr liability, the CET1 buffer is not enough — a 16% book haircut and a 15–20% dilutive raise are the outcome the bear case prices, and the variant view on this disagreement is wrong.

Beyond those three, the broader structural risks remain in force: this is a sub-scale franchise with a 170 bps NIM gap to peers that the variant view does not attempt to close; five live legacy legal threads, any one of which can produce a tape-relevant headline; and a CEO five weeks into the job. The variant view is a refinement of how to read the slope of the recovery, not a claim that the destination is HDFC-like economics. If the recovery slope itself breaks — a second large prudent provision in Q1 FY27, an adverse AT-1 ruling, a NIM print below 2.6% — the variant view goes with it.

The first thing to watch is the Q1 FY27 cost-to-income print on ~17 July 2026 — if it lands below 63% on positive jaws of ≥ 5 pp, the variant view advances by a full grade; if it lands above 65% despite revenue growth, the operating-leverage leg is broken and the consensus read on Yes Bank is right.