The line that gets repeated about Korean banks is that they trade at half of book value, and for some of them that's still true. It stopped being true of KB Financial Group Inc. (KRX:105560, NYSE:KB) some time ago, and the gap between what a US screener shows and what the shares actually cost against common equity is now wide enough to matter.
At ₩166,600 the market capitalisation is ₩59.1tn. Total equity at the end of June was ₩62,701.8bn. Divide one by the other and you get 0.94x, which is the number a screener pulling the consolidated equity line will report. It's the wrong number, and the regulator already knows it's the wrong number.
Of that ₩62,701.8bn, ₩1,811.0bn is non-controlling interests — equity in subsidiaries owned by somebody other than KB. Equity attributable to owners of the parent is ₩60,890.8bn.
Inside that parent figure sits ₩4,206.7bn of hybrid securities, carried on the balance sheet under 신종자본증권 and issued in the form Korean regulators call write-down contingent capital. In Basel terms these are Additional Tier 1 instruments. They are equity for accounting purposes, they pay a coupon rather than a dividend, and they are explicitly not common equity tier 1. KB's own CET1 ratio of 13.74% already excludes them. A stock screener adding them to book value is doing something the Financial Supervisory Service would not do.
Take both out and common equity is about ₩56,684.1bn. Against a ₩59.1tn market capitalisation, that's roughly 1.04x. On 351,106,111 shares outstanding — the figure KB used in its July dividend resolution — book value per share works out near ₩161,400 against a ₩166,600 close.
So KB is not cheap on book. It's slightly expensive on book, and the case for owning it has to rest on the 14.09% return on equity rather than on a discount that isn't there.
The coupons are not free either. In FY2025 KB paid ₩304.9bn in hybrid distributions across the group, ₩202.4bn of it at the parent level, charged directly against retained earnings. That's a claim ranking ahead of the common dividend, and it's about 3.5% of parent net income. Basic earnings per share already deducts it, which is why the reported ₩10,670 for the first half doesn't reconcile cleanly to the ₩3,884.6bn attributable profit.
Korean companies have a long history of buying back shares and then parking them in treasury, where they can reappear. KB isn't doing that.
The FY2025 statement of changes in equity shows ₩1,480.0bn spent acquiring treasury shares, up from ₩820.0bn the year before, and ₩814.5bn of treasury shares cancelled against retained earnings. Over 2024 and 2025 combined that's ₩2.3tn bought and ₩1.33tn permanently retired. The treasury balance then fell from ₩1,901.5bn at the end of December to ₩176.0bn at the end of June, which means the great majority of what was sitting there has since been destroyed.
The July resolution continues it. On 23 July the board approved a ₩700bn buy-and-cancel, estimated at 3,997,715 shares against the ₩175,100 close the previous day, to be purchased on-market through Samsung Securities between 24 July and 16 December. The filing states plainly that the shares acquired will be cancelled in full, and notes that because the cancellation is made under the proviso to Article 343(1) of the Commercial Act — out of distributable profit — the total number of issued shares falls while paid-in capital does not. Issued shares stood at 354,687,734 when the resolution was filed.
The policy behind it is a capital threshold rather than a payout ratio. KB has said it will direct capital above a 13.5% CET1 ratio to shareholder returns, and the ratio was 13.74% at the end of June, up from 13.64% the quarter before. That's a rules-based commitment, which is more useful to a foreign holder than a target, because you can check it yourself each quarter.
Add it up for FY2025: ₩1,004.0bn in quarterly dividends, ₩298.3bn in the annual dividend, ₩1,480.0bn in buybacks. That's ₩2,782.3bn returned against ₩5,833.2bn of parent net income, a 47.7% total payout, with more than half of it coming through repurchase rather than dividend.
There's an item in the first-half balance sheet I can't account for and don't want to paper over. Capital surplus fell from ₩16,633.5bn at the end of December to ₩9,127.5bn at the end of June, down ₩7,506.0bn. Over the same six months retained earnings rose from ₩38,333.7bn to ₩46,129.3bn, up ₩7,795.5bn — far more than the ₩3,884.6bn of half-year profit could produce after dividends.
Total equity is unaffected. This is a transfer between two lines inside it.
Korean holding companies do this for a specific reason: buybacks and cancellations must come out of distributable profit, and moving capital surplus into retained earnings enlarges that pool. If that's what happened, it's a housekeeping step that makes several more years of the current buyback pace legally possible, and it's mildly bullish for anyone who cares about the capital return continuing.
But I'm reconstructing that from the shape of the numbers, not reading it. The summary statements in the half-year report don't say. The detail would be in the statement of changes in equity and its accompanying note, and the version I can pull through DART's structured feed covers FY2025, not the first half of 2026.
The obvious objection to my treatment is that stripping hybrid capital out of book value entirely is too harsh. These instruments do absorb loss — that's what the write-down trigger is for — and they support the balance sheet in a way pure debt doesn't. A holder is genuinely better protected with ₩4.2tn of AT1 beneath them than without it. Excluding it makes the book value conservative, not correct.
The counter is that it's exactly what the capital regime does, and the coupon is a real cash cost that recurs whether or not KB earns anything.
The bigger risk to the whole capital-return thesis is that the threshold cuts both ways. A CET1 rule that releases capital above 13.5% also withholds it below. Korean household credit and the commercial property exposures that have troubled Korean financials since 2023 are the two things most likely to push the ratio down, and if they do, the buyback stops without any change in policy or intent. There is no floor under the repurchase.
Then there's the funding mix. KB filed three separate board resolutions to issue write-down contingent capital in the five weeks to 24 August, alongside a shelf registration and multiple debt securities filings. In FY2025 the group issued ₩404.0bn of hybrids at the parent and redeemed ₩1,127.2bn, so the balance has been shrinking, from ₩5,082.6bn at end-2024 to ₩4,206.7bn now. A run of new issuance would reverse that, and it would mean part of what looks like capital returned to common shareholders is being replaced with a more expensive instrument sitting above them.
The third-quarter report, and specifically three lines.
The CET1 ratio against 13.5%. Anything at or below that number ends the discretionary buyback for a quarter, and the threshold makes it mechanical rather than a judgement call.
The hybrid capital balance against ₩4,206.7bn. If the three recent issuance resolutions land at scale and the number moves back above ₩5tn, the common equity denominator I used above shrinks further and the effective price-to-book rises again.
And the note explaining what moved ₩7.5tn from capital surplus to retained earnings. Until that's on the page it's the largest thing in KB's first-half accounts that nobody has written down.
kstock reads DART every morning and writes up what moved — the contract, the buyback, the number that does not add up. The daily post and a Saturday roundup, by email.