Shinhan Financial Group (055550): Korea’s $39bn Bank Moved $7.3bn to Keep Buying Its Own Stock

Korea’s thirteenth-largest listed company — fourteenth by ticker, because Samsung Electronics’ preferred line counts twice — is a bank you have probably never read a research note on. It is worth $39.35 billion, it has retired 8.4% of itself in thirty-one months, and in March 2026 its shareholders voted 99.9% in favour of moving $7.31 billion from one line of the balance sheet to another. Total equity did not change by a single won. What changed was permission — and that is the whole story.

Foreign investors who own Korea usually own it through semiconductors. Samsung Electronics and SK hynix are the trade; the banks are the thing you skip. That is a defensible choice right up to the moment you notice that Korea’s four financial holding companies are together worth about $132 billion, that they are the most direct beneficiaries of the country’s Value-Up programme, and that they publish more about their own capital plans than almost any listed Korean company — in Korean.

This is the first of our four deep-dives on those holdings. We start with Shinhan because it is the one whose story is furthest from what an English-language screener shows you.

A note on units: every money figure in this piece is in US dollars, converted once at ₩1,350.36/$1 (the FX close of 22 September 2026). Share prices are shown in won first, because that is what you will see on a Korean order book, with the dollar equivalent in brackets.

🏢 What Is Shinhan Financial Group, and What Do You Actually Own?

Shinhan Financial Group is a bank holding company: a 197-person legal shell in central Seoul that owns fourteen financial subsidiaries outright, of which Shinhan Bank is by far the biggest. When you buy 055550 you are not buying a bank — you are buying the parent that collects dividends from one.

The holding company (Shinhan Financial Group Co., Ltd.) was created on 1 September 2001 by a share transfer from the shareholders of four companies — Shinhan Bank, Shinhan Securities, Shinhan Capital and Shinhan Investment Trust Management. It listed on the KOSPI nine days later, on 10 September 2001, and registered with the US Securities and Exchange Commission on 16 September 2003, listing American Depositary Shares on the New York Stock Exchange the same day. Its head office is at 20 Sejong-daero 9-gil, Jung-gu, Seoul, and its chief executive is Jin Ok-dong. (Source: DART H1 2026 semi-annual report, rcpNo 20260814003306)

The number that surprises people: the listed entity itself had 197 employees at 30 June 2026 — 184 on permanent contracts and 13 on fixed terms, with average tenure of three years and one month. Everything the group actually does happens one level down, inside 412 consolidated companies (one listed, 411 unlisted), of which the filing designates 115 as material.

Fourteen entities sit directly beneath the holding company. Ranked by what they contribute rather than by name:

• Shinhan Bank — commercial and corporate banking. Unlisted. The group’s engine, and a company with a much older legal history than its name suggests (see the history section below).
• Shinhan Card — credit cards, card loans, instalment finance and leasing. Unlisted.
• Shinhan Securities Co., Ltd. — brokerage, proprietary trading and underwriting. Unlisted. The subject of this year’s surprise.
• Shinhan Life Insurance — life insurance, formed in July 2021 by merging Shinhan Life with Orange Life (the former ING Korea business).
• Shinhan Capital — leasing and new-technology finance.
• Jeju Bank — a small regional bank, and the only listed subsidiary in the group (KOSPI code 006220). This matters when you search for a ticker; more on that below.
• Plus Shinhan Asset Management, Shinhan Savings Bank, Shinhan Asset Trust, Shinhan DS (group IT), Shinhan Fund Partners, Shinhan REITs Management, Shinhan Venture Investment and Shinhan EZ General Insurance.

For a foreign reader the practical translation is this: Shinhan is a diversified Korean financial group with a bank at its centre, a credit-card business, a life insurer, and — the part that has changed — a brokerage that has stopped being a rounding error.

📊 Where Does Shinhan’s Revenue Actually Come From?

A bank has no “revenue” line in the sense a manufacturer does, so the honest answer is by segment: in the first half of 2026 the banking segment produced 70.2% of group net income, down from 73.2% a year earlier, while the securities segment doubled its share to 16.8% — overtaking credit cards and insurance combined.

Interest income, fee income and trading income are the three engines, and the group reports them both ways. Shinhan publishes both, on a consolidated group basis, with net income stated on a controlling-interest basis. Here is the like-for-like half-on-half comparison:

Run the arithmetic on the increase rather than the level and the picture sharpens. Group net income rose $300.1M. Securities alone contributed $236.1M of that, and specialty finance a further $22.8M — together 86% of the entire increase. Banking added $143.6M; insurance subtracted $41.5M.

The fee-income split is where the shift is starkest, because fees are the part of a bank’s income statement that is supposed to be stable:

For the first time, Shinhan Securities is the group’s largest single source of fee income — $559.3M against the bank’s $495.4M. Two-thirds of the group’s entire $510.8M increase in net fee income came from the brokerage.

The company is admirably direct about the cause. Its own management discussion says Shinhan Securities’ first-half net income of ₩577.7bn rose 123.1% “as equity brokerage commissions increased on higher trading value amid a strong domestic stock market, and product trading income improved.” Shinhan Capital’s 48.1% rise is attributed to “increased securities-related gains on the stock-market boom,” with interest income actually falling. These are the company’s explanations, not ours, and they point at the KOSPI rather than at anything structural.

One structural item does sit in the same window and deserves a mention without being handed credit it has not earned: Shinhan Securities received a short-term finance (promissory-note issuance, 발행어음) licence on 17 December 2025 — a permission only a handful of Korean brokers hold, which expands low-cost funding capacity. The company’s own attribution for the first-half result is brokerage and trading, not the new licence. Watch it; do not yet count it.

🗓️ How Did Shinhan Get Here? A Short History

The entity called “Shinhan Bank” today is legally Chohung Bank, founded in 1943 from two banks dating to 1897 and 1906 — which makes the group’s core the oldest continuously operating bank lineage in Korea, not a 1980s challenger.

This is one of those facts that is entirely uncontroversial in the Korean filing and almost never appears in English. The semi-annual report states it plainly: Shinhan Bank Co., Ltd. was established on 1 October 1943 by the consolidation of Hanseong Bank (founded 19 February 1897) and Dongil Bank (founded 8 August 1906), and was originally named Chohung Bank. On 1 April 2006 Chohung Bank absorbed Shinhan Bank and the surviving company renamed itself Shinhan Bank. The old Chohung card division was simultaneously spun off and merged into Shinhan Card.

So the English shorthand — “Shinhan, the bank founded by Korean-Japanese investors in 1982” — describes the brand and the culture, but the legal person on the balance sheet is 1897 vintage. Turning points only:

🌍 Is Shinhan Actually a Leader, or Just Big at Home?

Shinhan is the number-two Korean financial group on both consolidated assets and profit, and it is genuinely more international than its peers — but overseas operations contributed only 13.7% of second-quarter profit, so roughly 86% of it still came from home.

Ranking the four holdings on their own FY2025 filings, all consolidated and all IFRS:

• Net income attributable to controlling shareholders, FY2025: KB $4,319.7M > Shinhan $3,681.7M > Hana $2,964.9M > Woori $2,313.7M.
• Consolidated total assets, 30 June 2026: KB $641.9bn > Shinhan $622.7bn > Hana $528.4bn > Woori $470.4bn.

Second on both counts, in other words — close behind KB, clearly ahead of Hana and Woori. (Sources: each company’s FY2025 annual report and H1 2026 semi-annual report on DART; consolidation scopes differ, particularly in insurance, so treat the asset ranking as indicative rather than like-for-like.)

On international reach the filing is specific and checkable: Shinhan Bank’s overseas network at 30 June 2026 was 163 locations in 20 countries, and the group’s overseas operations earned ₩250.7bn ($185.7M) in the second quarter — up 13.0% quarter on quarter, contributing 13.7% of that quarter’s profit. ⚠️ A basis note, because the filing’s own sentence is loose here. It attaches the 13.7% to 상반기 (“first half”), but only the quarterly reading reconciles: ₩250.7bn is 13.77% of second-quarter profit attributable to owners (₩1,820,135m, printed in the same report) and just 7.28% of the half’s ₩3,442,729m. The ₩250.7bn is itself given as a quarter-on-quarter change, which settles it. We therefore read both the amount and the 13.7% as second-quarter figures. The 2017 purchase of ANZ’s Vietnam retail business made Shinhan, in the company’s own words, the leading foreign bank in Vietnam; that is a company-reported claim we cannot independently verify and we label it as such.

What we are not going to do is quote Korean market-share percentages. Those numbers exist, but the ones that circulate in English are usually a mix of deposits, loans and “customers” from different dates, and Shinhan’s own semi-annual report does not publish a group market share. The verifiable version of “is it a leader” is the ranking above.

📈 Is the Business Getting Better or Worse?

Better on earnings, and mixed on credit: profit rose 13.3%, while the group’s allowance coverage of problem loans fell from 145.9% to 117.0% across the ten quarters the filing prints — a fall concentrated in a single step-down in the first quarter of 2026 rather than a steady slide, with the non-performing loan ratio range-bound at 0.71–0.84% throughout.

Note the wedge between the last two rows: earnings per share grew 18.4% while net income grew 13.3%. Roughly five percentage points of EPS growth were manufactured by there being fewer shares. That is the buyback showing up in the only place that matters to a shareholder. (The EPS numerator is also net of coupons on hybrid capital securities, which vary, so treat the five points as approximate rather than exact.)

On costs, the company reports a cost-to-income ratio of 36.6% for the second quarter of 2026, which it describes as stable, noting that second-quarter administrative expenses rose 8.1% on the prior quarter because of education and property taxes and performance pay at the securities arm. Computed on the filing’s own half-year subtotals — administrative expense divided by the sum of net interest, net fee and other operating income — the six-month figure is 41.0%. These are two different bases over two different periods and should not be compared to each other; we give both so that you know which one you are looking at when you see a number quoted elsewhere.

The semi-annual report credits the interest-income increase to an improvement in the bank’s net interest margin plus growth in average loan balances. It does not publish a group NIM figure, and we are not going to supply one — Shinhan’s NIM is disclosed in its quarterly earnings presentation and regulatory business disclosure, not in the semi-annual report we are reading here.

Now the part that does not appear in the headline. Shinhan’s semi-annual report carries a ten-quarter table of group capital and asset-quality ratios. We print all ten columns, because the story changes depending on which ones you keep:

Three readings matter here, and the third is the one a five-column extract would have hidden.

First, CET1 at 13.43% is comfortably above the company’s stated 13% floor and rose 13 basis points in the second quarter — that is the fuel gauge for everything in the next section. Note, though, that the series high is behind us: CET1 peaked at 13.62% in June 2025, and the total capital ratio peaked in the same quarter at 16.25%.

Second, the NPL ratio is range-bound, not deteriorating. Across ten quarters it has traded between 0.71% and 0.84%. The 0.71% we quoted as a starting point is the low of the entire series, and June 2026’s 0.79% sits below the 0.84% printed in both of the first two quarters of 2025. Any sentence that starts from March 2024 and ends at June 2026 is measuring from the best reading available.

Third, allowance coverage really has fallen — by 28.9 percentage points from March 2024 to June 2026 — but not gradually. It dropped through 2024, flattened around 124% for three quarters of 2025, rose to 126.55% at December 2025, then stepped down 13.7 points in a single quarter to 112.87% at March 2026 before recovering to 116.97%. That shape is a one-quarter event with a partial recovery, not a glide path, and it is worth knowing which of the two you are being sold. Provisions were also cut 7.6% year on year; the company describes its provisioning as “conservative and pre-emptive” and reports a credit-cost ratio of 0.42% within its annual plan.

Worth separating two Korean terms that look interchangeable in English and are not. The loan-loss allowance (대손충당금) is a contra-asset that reduces earnings — that is what the coverage column above measures. The loan-loss reserve (대손준비금) is something else entirely: an appropriation within equity that Korean supervisory rules require when the IFRS allowance falls short of the regulatory minimum. Shinhan discloses it separately, in note 20-2: ₩4,243,565 million ($3,142.5M) already appropriated at 30 June 2026, rising to ₩4,524,607 million ($3,350.7M) once the ₩281,042 million appropriation planned for this period is reflected. The same note shows what it does to earnings on a supervisory view — half-year profit of ₩3,442,729 million becomes ₩3,163,438 million after a ₩279,291 million reserve adjustment, and half-year EPS of ₩7,075 becomes ₩6,484. Mixing the two distorts both earnings and capital ratios, which is why we name the row the filing names it. (Source: rcpNo 20260814003306, consolidated note 20-2.)

One more asymmetry: at 30 June 2026 Shinhan Bank’s own NPL ratio was 0.31% and its delinquency ratio 0.34%, against the group’s 0.79%. The bank is clean; the credit risk in this group lives outside the bank — Shinhan Card’s delinquency ratio was 1.21%.

🔍 Why Did Shinhan Move $7.31 Billion Across Its Own Balance Sheet?

Because a Korean buy-and-cancel can only be funded out of distributable profit, which is measured on the holding company’s separate accounts — and in FY2025 Shinhan handed shareholders 105% of its separate-basis net income. In March 2026 shareholders voted to reclassify ₩9,865,923 million ($7,306.1M) of share premium as retained earnings, which is exactly where that pool is measured.

Here is the arithmetic behind that, because “running out of headroom” is the kind of claim that needs a number rather than an adjective — and it needs the right set of accounts. Article 462 of the Commercial Act measures distributable profit on the separate (별도) financial statements of the company declaring the distribution, not on the consolidated group. Shinhan’s FY2025 annual report prints both, side by side, in its dividend section:

• Consolidated net income (controlling interest): ₩4,971,561 million ($3,681.7M). Separate net income: ₩2,385,457 million ($1,766.5M) — the holding company collects dividends from subsidiaries, so its own income is roughly half the group’s.
• FY2025 cash dividends: ₩1,245,730 million ($922.5M). That is the 25.06% consolidated payout ratio the company reports — but 52.2% of separate net income.
• Add the buyback. The statement of changes in equity shows ₩1,250,001 million ($925.7M) of treasury stock purchased in FY2025, which is the ₩1.25tn the company’s own commentary cites. Dividends plus buyback is ₩2,495,731 million ($1,848.2M) — and that reconciles exactly to the 50.2% total shareholder-return ratio Shinhan reports, because 2,495,731 / 4,971,561 = 50.20%.
• Measured against the accounts that actually govern the distribution, the same ₩2,495,731 million is 104.6% of separate net income.

That is the constraint, stated precisely. Returning more than a year’s separate earnings is not illegal and does not by itself mean the pool was empty — distributable profit is a cumulative balance, not one year’s income, and Shinhan does not publish the balance. What it does mean is that the programme had stopped being funded by the year’s profit and had started drawing down the accumulated pool, one year at a roughly 105% run-rate, with a larger buyback already announced for 2026. (Sources: DART FY2025 annual report rcpNo 20260318000826, dividend section and consolidated statement of changes in equity.)

The vote itself is also completely unhidden: a numbered AGM agenda item with a recorded tally. What we could not find is any English-language report of it — not a wire story, not a broker note in English, not a filing summary. The event is public; the translation is what is missing. And it is not unique to Shinhan, which we deal with below.

Start with the mechanic. Every one of Shinhan’s cancellation filings contains the same sentence: the retirement is made under the proviso to Article 343(1) of the Korean Commercial Act, using treasury shares acquired “within the limit of distributable profit,” so that the number of shares outstanding falls but stated capital does not. That last clause is not boilerplate — it is the constraint. A Korean company cannot simply decide to shrink itself; it must have distributable profit to shrink itself out of.

You can see the constraint on Shinhan’s own balance sheet. Look at the composition of equity attributable to controlling shareholders:

Capital surplus fell $7,306.0M in six months. Retained earnings rose $8,628.9M — far more than the $2,549.5M of half-year profit could explain. The statement of changes in equity names the transaction that reconciles them: a line item translating as “transfer of share premium to retained earnings,” ₩9,865,923 million ($7,306.1M), with an offsetting increase of exactly the same amount in retained earnings and zero effect on total equity.

One small precision, because we were nearly caught by it ourselves. The ₩9,865,777 million figure that appears in the same column of that statement is not the transfer — it is the column subtotal for all transactions with owners, and it differs from the transfer line by ₩146 million of “other consolidated capital adjustments” recorded in the same block. The transfer itself is ₩9,865,923 million; the balance-sheet movement in capital surplus is ₩9,865,777 million. Read the line item, not the total.

The other lines in that same “transactions with owners” block, for the six months to June 2026, are worth reading together because they are the full shareholder-return picture:

And the company states the purpose itself, which spares us any guessing about causation. In the dividend filing of 5 February 2026 — the same board meeting that approved the next buyback — Shinhan wrote that the board had resolved to transfer capital reserve to retained earnings under Article 461-2 of the Commercial Act, to be put to the March AGM, and that “this is part of executing the company’s shareholder-return policy in order to expand the dividend resource.” (Source: DART rcpNo 20260205800319, filed 5 Feb 2026)

The board minutes in the semi-annual report show the sequencing. At the regular board meeting of 5 February 2026, agenda item 3 was “reduction of capital reserve and transfer to retained earnings” and agenda item 4 was “acquisition and cancellation of treasury shares.” Both passed. The fuel and the engine were approved in the same sitting.

At the AGM of 26 March 2026 the item passed as ordinary business: of 477,157,399 voting shares, 396,728,371 voted, and 396,275,363 — 99.9% — voted in favour, with 453,008 against or abstaining. One outside director’s recorded rationale, reproduced in the filing, reads: “approved as proposed, as it helps the shareholder-return policy by securing stable distributable profit.”

The 477,157,399 figure is worth a footnote, because it is neither the issued share count nor today’s count. It is the number of shares in circulation at the AGM record date of 31 December 2025, and the FY2025 annual report prints the whole sum: Ⅳ. shares issued 485,494,934, less Ⅴ. treasury shares 8,337,535, equals Ⅵ. shares in circulation 477,157,399 (rcpNo 20260318000826, share-capital table). The 10,840,573 shares cancelled on 6 February 2026 were still outstanding on that record date, which is why the voting base is larger than the share count the company actually had on the day of the meeting — 474,654,361, the figure the semi-annual report uses for its own June-2026 shareholder percentages (rcpNo 20260814003306).

The AGM notice itself is where the company puts the number in round form, and it says something the balance sheet does not. The notice describes the reserve being transferred as ₩6.8tn of common-share premium plus ₩3.1tn of other share premium, ₩9.9tn in total — and then adds that “the portion of that amount which satisfies the requirements of the relevant law” may be used as a source of tax-free dividends in future. Not all of it, in other words. (Source: DART AGM notice rcpNo 20260303003230, filed 3 Mar 2026, agenda item 2.)

Throughout this article we use the executed figure from the financial statements rather than the round one from the notice — the transfer line in the consolidated statement of changes in equity (rcpNo 20260814003306) — and flag the notice’s ₩9.9tn only where the company itself uses it, as in the peer table below.

For completeness, the same equity table shows that Shinhan had never done this before: its capital surplus moved by just ₩3,590 million across the whole of FY2025 and ₩7,080 million in H1 2025. For this company, the transfer is new.

It is not, however, a Shinhan-only manoeuvre — and Shinhan was not first

All four of Korea’s big financial holdings have now used Article 461-2 to convert capital reserve into retained earnings. Shinhan’s is the largest; Woori’s was the earliest, by a full year.

We checked the 2025 meetings as well as the 2026 ones, and that is the only reason Woori is in this sentence rather than excluded from it. Search only the first half of 2026 — the window the other three sit in — and Woori returns zero hits, which reads as “Woori did not do this.” It did; it did it first. Three of the four passed the resolution in March 2026, within three days of each other. Woori passed its version at its March 2025 meeting, twelve months earlier. The lesson is the one that keeps recurring in Korean filings: a negative finding is only as wide as the window you searched.

So the finding is not that Shinhan invented something. It is threefold, and each part survives the correction: Shinhan’s is the biggest — ₩9.9tn against KB’s ₩7.5tn and Hana’s ₩7.4tn, the three figures that are stated on the same basis; its effect on total capital is precisely zero, which is what makes it invisible to anyone reading a capital-ratio screen; and the sector-wide version of this — four of Korea’s largest banks quietly rebuilding their legal capacity to distribute, across two AGM seasons — has not been reported in English either. That is a bigger story than one bank’s agenda item, and we would have missed it by writing the version that felt more exclusive.

🔁 Why Does Shinhan’s Buyback Retire Fewer Shares Every Year?

Because each tranche is sized in won, not in shares — so the higher the stock goes, the fewer shares the same money retires, and Shinhan’s share-count target gets more expensive precisely because the programme is working.

Every Shinhan cancellation follows the same three-step pattern, and each step is a separate DART filing. The board sets a won budget and converts it to an indicative share count at the previous day’s close. A broker then buys in the market over a window of one to six months. Finally a correction filing states the number of shares actually acquired and the cancellation date. Announcement and execution are different events, and the gap between them is information.

Four of the six completed tranches retired fewer shares than announced; two retired more. The pattern is not random and it is not execution skill. Look at the two that over-delivered — October 2024 (reference ₩56,100) and February 2025 (reference ₩50,700). In both windows the share price fell below the reference, so the fixed won budget bought more shares. In the four that under-delivered, the price rose. The worst gap was April 2024: a $222.2M budget struck at ₩43,500 was meant to retire 6,896,551 shares and retired 5,947,889 — 948,662 fewer, a 13.8% shortfall, for the same money.

Across the six completed tranches Shinhan announced 44,470,299 shares and retired 43,309,232 — 1,161,067 fewer, or 2.6% — while spending ₩2,649,999,960,500 ($1,962.4M), within ₩39,500 of the round ₩2.65tn it budgeted. The money was spent to the last basis point; the share count was not.

The share-count bridge closes exactly, which is how we know the ledger is complete:

Average price paid across the programme: ₩61,188 (~$45.31) per share retired. Against a 22 September close of ₩113,200 that is a 46% discount — the buyback has been strongly value-accretive. It is also why the next stretch is hard.

The target is in shares, and the price has doubled

Shinhan’s original Value-Up plan, filed 26 July 2024, set three 2027 targets: ROE of 10% and ROTCE of 11.5% on a CET1 ratio above 13%; a total shareholder-return ratio of about 50%; and — the third one — to reduce shares outstanding “to below 500 million in 2024 and to 450 million by 2027,” explicitly to lift tangible book value per share. The filing is unusually candid about why: the company judged that it needed “a larger reduction in share count than major domestic financial holding peers.” (Source: DART rcpNo 20240726800197, filed 26 Jul 2024. This document is labelled forward-looking information and says so in its first line: plans may change and actual results may differ.)

That candour was justified. Shinhan has 469,450,239 shares against KB’s 354,687,734 — 32% more shares for a smaller market capitalisation.

From 469,450,239 to 450,000,000 is 19,450,239 shares still to retire. What that costs depends entirely on when you ask:

• At the October 2024 reference price of ₩56,100: $808.1M.
• At the 22 September 2026 close of ₩113,200: $1,630.5M.

The same target, twice the money — because the plan worked. And the tranche currently in the market illustrates the squeeze in miniature. The board sized $518.4M on 22 July 2026 at ₩104,500, indicating 6,698,565 shares. Buy the same money at ₩113,200 and you retire about 6,183,700 — roughly 515,000 fewer shares. Finish it at that rate and the count lands near 463,266,494 — leaving about 13,266,494 shares still to retire by the end of 2027, a further $1,112.1M at today’s price.

The target itself changed wording — and the new wording is 12.8 million shares easier

The 2024 plan set an absolute floor of 450 million shares. Every restatement since sets a relative one: “reduce shares outstanding by 50 million.” Those are not the same target, and the second is the softer of the two.

The original July 2024 filing reads: “reduce shares to below 500 million in 2024 and to 450 million by 2027.” The April 2026 reset summarises the very same plan’s third target as “③ reduce shares outstanding by 50 million” — and so does the FY2025 annual report, and so does the H1 2026 semi-annual report. Applied to the 512,759,471 shares outstanding when the programme began, a 50-million reduction lands at 462,759,471, which is 12,759,471 shares above the 450 million the original filing named. At today’s price that difference is worth about $1,069.6M of buying the company no longer has to do to say it hit its target.

We are not asserting the company changed its target; it may simply be summarising loosely, and the April filing does say the existing 2027 targets “will continue to be kept.” But the loose summary is now the version that appears in three consecutive filings, and it is the easier one. This is the kind of drift a reader only catches by reading the 2024 filing and the 2026 filing side by side, in Korean. (Sources: rcpNo 20240726800197 — “2027년 450백만주까지 감수”; rcpNo 20260423800235 and rcpNo 20260814003306 — “발행주식수 50백만주 감속…감원”.)

There is a second, subtler thing in the April 2026 reset. The renamed plan — “Shinhan Value-Up +++” — leads with three targets: ROE 10%+ (managed in a 10–12% band), total payout 50%+ on a formula basis, and CET1 13%+. The share-count target is not among the three. The filing does say the existing 2027 targets “will continue to be kept” alongside the new guidance, so it has not been abandoned — but it has moved from the headline into a subordinate clause. That is a reasonable thing for a company to do when a target has become expensive. It is also the kind of thing a shareholder should notice.

The April 2026 Value-Up filing reports 2025 progress as: CET1 stable above 13%, ROE of 9.11% (up 67 basis points year on year, so short of the 10% target), total shareholder return of 50.2% (target beaten), and share count “reduced by about 25 million shares by January 2026.”

That last figure does not reconcile, and we are going to say so rather than pick whichever reading suits. The filing does not state the baseline date it is measuring from. Our own arithmetic on the cancellation ledger in the table above gives 27,264,537 shares retired between March 2024 and the end of January 2026; counting only the cancellations that happened after the plan was adopted in July 2024 gives 23,898,280, which is closer to “about 25 million” and is probably the intended basis. Neither figure appears in the Value-Up filing — both are sums we computed. The ledger reconciles to the exchange’s share count to the single share; the company’s one-line summary reconciles to nothing without an assumption.

A footnote that proves the mechanic

Shinhan’s paid-in capital has been frozen at ₩2,969,641 million through every cancellation. At the ₩5,000 par value the filings state, that corresponds to 593,928,200 shares — while only 469,450,239 exist. The 124,477,961-share gap is the accumulated legacy of retirements made under the Article 343(1) proviso, which shrinks the share count and leaves stated capital alone; 43,309,232 of those shares came from the 2024–26 programme documented above. If you ever wondered why a Korean bank’s “capital” line never moves while its share count falls every six months, that is why.

💰 What Do You Actually Pay for Shinhan Today?

₩113,200 a share buys you 11.5× FY2025 earnings and 0.88× June-2026 book — the middle of Korea’s big four on price, the bottom of the four on dividend yield, and the most buyback-weighted return mix of the group.

That 0.88× versus 0.95× line deserves a sentence, because almost every screener shows you the first number. Equity attributable to controlling shareholders includes $3,083.5M of hybrid capital securities — perpetual instruments that sit in equity under K-IFRS but belong to their own holders, and which cost $73.9M in coupons in the first half alone. Strip them out and common equity is $41,424M, so price-to-book is 0.95×, not 0.88×. Eight per cent is not a rounding difference when the entire bull case is “it trades below book.”

How Shinhan compares with Korea’s other three financial holdings

Two readings from that table, one flattering and one not.

The unflattering one: Shinhan is not the cheap one. Hana at 9.7× and 0.79×, and Woori at 8.8× and 0.68×, are both cheaper on both measures. If your thesis is “buy the cheapest Korean bank,” this is not it.

The interesting one: Shinhan and KB trade on almost identical earnings multiples — 11.5× against 11.4× — while Shinhan’s price-to-book is 13% lower. A persistent P/B gap at the same P/E is an ROE gap, arithmetically. But size that gap properly, on one period and one basis, because this is where a comparison goes wrong: on the two companies’ own reported FY2025 figures the gap is 0.82 percentage points — KB 9.93% against Shinhan 9.11% (KB: its H1 2026 report’s profitability indicators, rcpNo 20260814004200, FY2025 column; Shinhan: its April 2026 Value-Up filing, rcpNo 20260423800235). Recomputed ourselves on return on average controlling equity — FY2025 controlling net income over the mean of the December 2024 and December 2025 balances — the gap is a little wider, KB 9.98% against Shinhan 8.72%, and both companies sit below their reported figures because each defines its own average-balance denominator. Either way the direction is the same and the magnitude is under one and a quarter points. That is a thinner foundation for a 16% price-to-book premium than it first looks. The market is not discounting Shinhan’s governance relative to KB — it is discounting its returns, which is exactly what the Value-Up plan’s first target, ROE 10%+, is about. ⚠️ Do not compare either of these numbers with KB’s 13.11%, which is an annualised half-year figure; there is no comparable half-year number for Shinhan in the filings we are using.

The dividend line that matters most to a foreign holder

Shinhan has the lowest dividend yield and the lowest cash payout ratio of the four, and yet reports the highest total shareholder-return ratio. Its FY2025 total return ratio was 50.2% against a cash payout of 25.06%. The difference is buyback-and-cancel — though the company does not publish a line-by-line bridge between the two figures in this filing, so we will not invent one.

For a non-resident shareholder that mix is not neutral, it is better. A Korean dividend is taxed at source — 22% by default and 16.5% all-in for a US resident under the treaty. A share cancellation is taxed at nothing at all; it simply makes every remaining share a larger claim. Of Korea’s big four, Shinhan returns the largest proportion of its capital in the form a foreign holder keeps in full. Woori, with the highest yield and the highest cash payout, is the most tax-leaky of the four for the same reason. That is a genuine consideration and it is invisible on any yield screen.

The recent dividend record, all from the filings:

• 2025 Q1–Q3: ₩570 per quarter. FY2025 final: ₩880 (record date 20 Feb 2026, $309.3M). FY2025 total ₩2,590, $922.5M, which the company reported as a 3.0% yield on 2025 prices. FY2026’s total may differ, because the third-quarter amount is not yet declared.
• 2026 Q1: ₩740 (record date 30 Apr 2026, $258.4M). 2026 Q2: ₩740 (record date 30 Jul 2026, $257.3M).
• 2026 Q3: amount not yet declared. Record date is fixed at 3 November 2026, payment 27 November 2026; the board sets the amount after the quarter closes.

Two things follow. The quarterly dividend is up 29.8% year on year, ₩570 to ₩740 — a real increase. And the yield still fell, from the 3.0% the company reported for FY2025 to 2.29% on the same FY2025 dividend at today’s price, purely because the stock re-rated. If you came to Korean banks for income, the re-rating has already taken some of it.

🏷️ Which Shinhan Ticker Are You Actually Buying?

There are only two listed Shinhan entities — the holding company 055550 and its small regional subsidiary Jeju Bank 006220 — and everything else with “Shinhan” in the name is either unlisted or a fund.

Worth knowing for the sector as a whole: three of Korea’s four financial holdings have NYSE-listed ADRs — Shinhan (SHG), KB (KB) and Woori (WF) — and Hana has no NYSE or Nasdaq listing. We state it that narrowly on purpose: the SEC’s company-ticker-exchange file, which lists issuers with an exchange-listed class, returns all three of the others and no entry for Hana Financial Group. That file would not show an unsponsored over-the-counter depositary receipt, so we are not claiming Hana has no US-traded line at all — only that it has no exchange listing. If you are building the sector from a US brokerage account, Hana is not reachable the way the other three are. Our Korean ADR list covers the depositary mechanics, and our guide to buying Korean stocks covers the local route.

🌏 Is Foreign Money Buying or Selling Shinhan?

Selling — consistently, across 5, 20 and 60 sessions. And the two big-name 5% filings that look like a contradiction are not one, which is the more useful lesson.

We show every flow claim over at least two windows, because a single window invents narratives. Here are three:

All three windows point the same way, so there is no sign reversal to report and no window-shopping to confess: foreigners have been consistent net sellers, and Korean institutions have been the buyers on the other side. Note that the selling is into a strong market, at prices between roughly ₩87,700 and ₩114,600 against an average buyback price of ₩61,188. This looks like profit-taking, not a verdict.

Why the two 5% filings do not rebut that — and what they actually tell you

Because a Korean 5% large-holding report measures the change since that holder’s previous report — which for BlackRock was September 2018, and for Capital Group was the day before.

Over the same stretch, two large-holding reports landed on DART. Read only their headline lines and you get a tidy contradiction: the tape says foreigners are leaving, BlackRock and Capital Group say they are arriving. We nearly wrote that. Then we opened the filings.

Neither filing describes buying inside our flow window. Korea’s 5% rule requires a report when a holding crosses a threshold or moves by one percentage point — so a filing’s “increase” is the gap since that holder’s last report, which can be any distance in the past.

• BlackRock’s +4,475,332 shares is an eight-year figure. Its previous report was dated 14 September 2018. The filing also breaks the increase down, and the breakdown is not what the headline implies: only +928,863 of it is common stock held by BlackRock Fund Advisors itself. 3,426,797 shares — 76.6% of the entire increase — are booked against Aperio Group LLC, whose entire holding appears in the change schedule, and the affiliate count in the same table has gone from fourteen to fifteen. The filing lists “change in affiliated parties” (특별관계자 변동) among its stated reasons. Of Aperio’s 3,426,797 shares, 3,411,238 are depositary receipts rather than ordinary shares. And the report carries a note — our translation — saying that certain positions reported as shares in the previous report are being re-classified as depositary receipts and shown as movements for that reason, “not because the holdings changed through actual share trades” (실제 주식 거래로 인하여 내역이 변동된 것은 아님).
• Capital Group’s “new” filing covers a single day. Its previous report base date is 25 August 2026, at 23,453,086 shares and 4.99%. The next day it held 23,568,346 and 5.02%. The filing exists because a threshold was crossed, and the purchase it discloses is 115,260 shares — 0.03% of the company.

Size that against the tape. Capital’s 115,260 shares, valued at that day’s ₩107,400 settled close, is about $9.2M. The aggregate foreign net sale in 055550 over the 60 sessions is −$249.5M, or 3,100,749 Shinhan shares — not a market-wide number. The named buying we can actually locate inside the window is roughly 27 times smaller than the aggregate selling, on either measure. There is no contradiction to resolve. The tape is simply right.

Which leaves the honest version: the aggregate is a net of everything — index rebalancing, fast money and long-only flow all cancelling out — and for Shinhan, over these windows, the net is selling. Our guide to reading Korean investor flows explains why aggregates behave this way, and the Foreign Flow Watch series tracks them weekly.

🏛️ Who Controls Shinhan, and What Do Minority Holders Get?

Nobody controls Shinhan — and for a large Korean company that is genuinely unusual. There is no founding family, no chaebol parent, no holding-company pyramid above it and no circular shareholding. The largest shareholder is the National Pension Service with 9.45%.

Something quietly instructive is happening in that first row, and it happens twice. The company prints three National Pension Service holdings in the same filing: 43,846,070 shares and 9.03% at 31 December 2025; 43,630,195 shares and 9.19% at 30 June 2026; and 44,370,524 shares and 9.45% at the 30 July 2026 dividend record date. (Source: rcpNo 20260814003306, largest-shareholder section and its note on the Q2 record date.)

Now do the subtractions, which are ours. Between December and June the NPS stake rose while its holding fell by 215,875 shares — it sold, and its percentage went up anyway, because the denominator shrank faster. That is what a buy-and-cancel does to every holder who does nothing: it makes them bigger by standing still. Then, in July, the opposite. The share count did not change between 30 June and 30 July, so the move from 43,630,195 to 44,370,524 is a real purchase of 740,329 shares — bought in a month when the foreign tape in Shinhan was selling. One percentage went up for an arithmetic reason and the next went up for a real one, and nothing in the percentage itself tells you which. Always check whether the share count moved.

The flip side of dispersed ownership is that Shinhan’s board answers to institutions rather than to an owner. The recorded voting on the 26 March 2026 AGM suggests that is working reasonably well by Korean standards: the financial statements passed with 99.2% support, the capital-reserve reduction with 99.9%, and an amendment to the articles of association — which needed a special resolution — with 99.9%. The audit committee is composed entirely of outside directors, the company has adopted electronic voting, and nine outside directors attended every buyback and cancellation resolution we read.

On shareholder returns, the summary is:

• Policy: total shareholder-return ratio of 50%+, now on a formula basis, with CET1 13%+ as the stated constraint. FY2025 delivered 50.2% against the 50% target. (Company-reported, and the Value-Up filing is explicitly labelled forward-looking information.)
• Dividends: quarterly, with record dates fixed a year ahead — an unusually shareholder-friendly practice in Korea, made possible by a January 2025 amendment to Article 165-12 of the Financial Investment Services and Capital Markets Act, which let boards set quarterly record dates themselves. Shinhan’s board pre-announced all three 2026 quarterly record dates at its February meeting.
• Buybacks: $1,962.4M spent and 43,309,232 shares permanently retired since February 2024, with a further $518.4M buying in the market until 22 October 2026. Every cancellation is a filed, dated event — not a discretionary programme.
• Tax status: Shinhan qualifies as a “high-dividend company” under Article 104-27 of the Restriction of Special Taxation Act, which the company flags in its Value-Up filings. That designation certifies the payout, not its durability.
• Related-party dealings: the compliance officer’s recorded activity includes quarterly review of intra-group transactions, all reported as “appropriate.” That is the company’s own attestation, not an independent finding.

One discordant detail from the same filing: the employee stock ownership association’s holding fell from 24,470,282 shares at the start of 2026 to 22,363,830 at 30 June — 2,106,452 shares, 8.6%, sold by staff into the rally. Employees are not a governance signal, but they do work there.

For context on why any of this moves a Korean share price at all, see our explainers on the Korea Discount and the Value-Up programme.

🧮 How Much Capital Is Left to Return?

Above its own 13% CET1 floor, Shinhan held about $1,165M of excess common equity at 30 June 2026 — and the buyback currently in the market will consume roughly 44% of it.

This is the calculation the whole Value-Up story rests on, and it is simple arithmetic on two numbers in the filing. CET1 capital was ₩49,361,300 million against risk-weighted assets of ₩367,595,900 million. Thirteen per cent of those RWAs is ₩47,787,467 million. The difference — ₩1,573,833 million, or $1,165.5M — is the capital sitting above the company’s stated floor.

The $518.4M tranche buying until 22 October 2026 uses about 44% of that. Which sounds tighter than it is, because the buffer refills: CET1 capital grew ₩1,256,600 million in the second quarter while risk-weighted assets grew ₩5,793,000 million, absorbing roughly ₩753,090 million of it at a 13% marginal charge — leaving about ₩503.5bn ($372.9M) of net CET1 generation in that one quarter. That is our arithmetic on one quarter, not a company figure and not an annual run-rate. We are deliberately not multiplying it by four: RWA growth is lumpy, FX moves it (the company explicitly credits FX management for the second-quarter CET1 improvement), and a credit cycle would move both terms at once. One quarter’s generation covering roughly three-quarters of one tranche is the right order of magnitude to carry away.

📚 Lingo Check

🎯 Why It Matters for K-Export Stars

This site exists on a simple premise: the decisive facts about Korean companies are filed in Korean, on DART, and never translated. Shinhan is the cleanest demonstration we have published.

Everything in this article was public. The $7.31 billion capital-reserve transfer was a numbered AGM agenda item with a recorded vote of 396,275,363 to 453,008. The seven-tranche buyback ledger, with its announced-versus-executed gaps, is seven pairs of filings. The 123.1% jump at the securities arm is in a segment table, with the company’s own attribution to the stock market printed beside it. The National Pension Service selling shares while its stake rose is two lines in the shareholder section.

None of it required a model, an analyst call or a source. It required reading the filing in the language it was filed in — and then doing the arithmetic the filing does not do for you. That is the entire edge, and it is available on a $39 billion company in the KOSPI top fifteen because the English-language coverage stops at the press release.

Shinhan is the first of four. KB, Hana and Woori run the same playbook with different arithmetic, and we are applying one consistent valuation rule across all four so the comparison actually means something.

Disclaimer: This article is for informational and educational purposes only and does not constitute investment advice, a recommendation, or an offer to buy or sell any security. All figures are sourced from public filings (DART), exchange data (KRX) and the US Securities and Exchange Commission as of the dates stated, and time-sensitive figures — prices, market capitalisations, multiples, yields and flow data — are dated to 22 September 2026 and will change. Value-Up targets quoted here are the company’s own forward-looking statements, which its filings expressly note may change and may differ from actual results. Any calculation labelled as ours is an estimate, not a company figure. Disclosure of interest: the author holds no position in Shinhan Financial Group (055550), and none in KB (105560) or Woori (316140). The author does hold small long positions in Hana Financial Group (086790) and Samsung Electronics (005930), both named in this article, and holds KODEX 200, a KOSPI 200 index ETF that gives indirect exposure to all four bank holding companies discussed. Do your own research and consult a licensed adviser before investing. See our Corrections Policy.