Sumitomo Mitsui Financial Group, Inc. SMFG

Key data

The business

Sumitomo Mitsui Financial Group is the holding company over SMBC, the second-largest of Japan's three megabanks. It takes deposits from Japanese households and companies, ¥185.7 trillion of them, lends to corporates at home and around the world, and runs a large securities house in SMBC Nikko, a credit-card and consumer-finance arm, and a leasing business. It also sits on a vast portfolio of government and other bonds, the place idle deposits go when loan demand is soft.

For twenty years the model was hostage to one fact: the Bank of Japan pinned interest rates at or below zero, so ¥185 trillion of deposits earned the bank almost nothing. The franchise, cheap and sticky and enormous, was always there. What was missing was anything to earn on it. The 2024 exit from negative rates changed that, and it is the entire reason the ADR has nearly doubled off its low. The moat is the deposit base. The question this memo circles is whether the market has now paid, in full and in advance, for the good news that flows from it.

The numbers

Two things move this stock: how fast fee income compounds, and how violently the as-filed bottom line swings when bond markets move.

Fee income has climbed every year, ¥1.17T to ¥1.63T, about 9% annually, the quiet engine underneath the bank. Net income on the US-GAAP basis SMFG files with the SEC is a different animal, lurching with the marks on its securities book. That US-GAAP figure reported to the SEC differs from Japanese accounting standards, which treat securities marks and hedges differently. That gap is the whole point of linchpin one.

The quarterly ADR prints show the same volatility in miniature.

A 60% miss followed by a 73% beat on a bank whose underlying business barely moves quarter to quarter. That is accounting noise, not franchise noise, but it is the noise a rate bet has to sit through.

Credit, the thing that actually kills banks, is clean. The July 6-K shows reserves built modestly and the funding base steady. Non-performing loans sit near 0.97%.

Reserves rose about 6% to ¥1.0 trillion. The watch-item flagged last time, credit over headline EPS, held: the bank is provisioning a touch more off a tiny base, which is normalization, not deterioration. The larger point is the re-rate. The multiple went from about 0.4 times tangible book, its lifetime norm, to 1.8 now, and that expansion, not book growth, carried the ADR from the low-fifteens to $27. On a carried return on tangible equity near 11%, paying 1.8 times book hands a buyer roughly a 6% earnings yield on equity, ordinary for a bank. What the market has not fully separated is re-rating from compounding: the easy money is made, and forward return now leans on tangible book per share actually growing after the securities marks and above-book buybacks. The domestic interest spread in the next two 6-Ks settles it.

Management

CEO Toru Nakashima runs a shareholder-friendly capital story: a steady stream of buybacks followed by outright cancellation of the repurchased shares, documented across a run of summer 6-Ks. Cancelling stock is the honest version of a buyback, it permanently shrinks the count. The catch is price. Buying back at 1.8 times tangible book means every ¥1 spent retires only about ¥0.56 of tangible book, so the buybacks lift reported per-share earnings while quietly diluting tangible book per share, the one number the bank lens treats as truth. It is the right policy at 0.5 times book and a questionable one here. Insider activity is a non-event: one sale of about $170,000 in August, plan status not disclosed, nothing bought. Stock awards now carry malus and clawback provisions, a governance nod worth having.

How it fails or surprises you

The bond book marks against you. SMFG holds a large securities portfolio, and rising Japanese yields mark it down, a big part of why as-filed net income fell to ¥516B from ¥936B two years earlier. Another 50 to 75 basis points on the 10-year JGB widens the unrealized-loss line and pressures capital, with CET1 carried near 10.3%. Watch section four of the next 6-K.

You pay a record price for an ordinary bank. At 1.8 times tangible book against a 0.3 to 0.5 lifetime norm, there is no book-value floor under the price. A Bank of Japan pause reverses the re-rate toward book with nothing earned to cushion it. This is the number my clean-bank read cannot explain away: why pay 1.8x for 11% ROTCE when the bank has never sustained a multiple this high.

Rates keep normalizing for years (right tail). If the policy rate climbs from about 1% toward 1.5 to 2%, the spread on ¥186 trillion of near-zero deposits widens materially, cancellations keep shrinking the count, and a deposit-heavy bank is the most geared name to that path. The market may still under-price a multi-year climb. Watch the domestic spread and the buyback notices.

Closing thoughts

The market has already paid for most of the Japan rate normalization - the move from 0.4x to 1.8x tangible book did that. From here, returns depend on whether rates keep climbing and spreads keep widening, or whether the Bank of Japan pauses and the multiple contracts back toward book. No single quarter settles it; the domestic interest spread and the unrealized-securities line reveal it gradually across filings. The edge is small because the obvious re-rate already happened, and the fatter tail leans mildly down: a pause or a bond-mark air-pocket de-rates a stock at an all-time-high multiple faster than underlying earnings can cushion it.

The bet is still that Japanese rates keep normalizing and Sumitomo Mitsui earns more on its ¥186 trillion of cheap deposits. What breaks it is a BOJ that stops, or JGB yields that spike faster than loans can reprice and gut the bond book. The two numbers that tell you first, before headline profit does, are the domestic interest spread and the unrealized-securities line in the next 6-K. If the spread goes flat while the marks bleed, the bet is wrong no matter what per-share earnings print.