Between the end of FY2022 and the end of FY2024, DB Insurance Co., Ltd. (KRX:005830) earned ₩3,595.6bn of net income and its shareholders' equity fell by ₩2,733.1bn.
Both of those are from the same set of consolidated statements. Equity went from ₩12,124.3bn to ₩10,303.7bn to ₩9,391.2bn across those two years. Retained earnings over the same span went up, from ₩9,007.2bn to ₩10,414.2bn to ₩11,950.1bn, and dividends were paid in both years.
If you value this company off book value, you should know that its book value is partly a rates instrument. It has since recovered to ₩14,041.2bn at 30 June 2026, which is a ₩4.6tn swing in three years at a company whose annual profit is under ₩2tn.
FY2024 is the clearest case. Net income was ₩1,853,217m. Total comprehensive income was negative ₩626,719m. The difference is ₩2,479,936m that never appeared in the income statement and went straight to equity.
It runs the other way too. FY2022 net income was ₩2,039,482m against comprehensive income of ₩4,234,415m. The first quarter of 2025 posted ₩431,364m of net income and negative ₩831,631m of comprehensive income. The second quarter of 2026 posted ₩876,864m of net income and ₩1,970,458m of comprehensive income.
Quarter to quarter, the two series are barely related.
For a US reader used to a P&C carrier where net income and comprehensive income differ by a manageable amount of unrealised bond gains, this looks broken. It is not. It is what happens when IFRS 9 and IFRS 17 arrive together at a company holding roughly ₩80tn of assets against multi-decade insurance liabilities.
Both sides of the balance sheet get remeasured. A large part of the investment portfolio is carried at fair value, and the insurance contract liabilities are discounted at market rates that also move. When rates fall, the bonds gain and the liabilities also grow. When rates rise, both shrink. In principle these offset. In practice the durations do not match exactly, the discount curves are not the same, and the portion of each revaluation that is routed to other comprehensive income rather than profit and loss is determined by accounting elections rather than by economics.
Whatever does not offset lands in accumulated other comprehensive income, which is a component of equity. Hence the gaps.
The half-year report contains the OCI note that would break this into its components, and I have not pulled that apart here. What matters for a reader is the direction of the conclusion, not the decomposition: the equity line on this balance sheet is not a running tally of retained profit, and treating a price-to-book ratio as a stable valuation anchor for a Korean insurer will mislead you at exactly the moments when rates are moving.
That is also why the value-up plan the company published on 28 August keys its dividend policy to K-ICS solvency bands rather than to book value or earnings. The regulator's capital measure is the one management can actually steer.
Underneath the accounting noise there is a real trend, and it is not flattering.
Interest expense was ₩78,985m in FY2021. It was ₩230,690m in FY2022, ₩381,226m in FY2023, ₩465,818m in FY2024 and ₩609,742m in FY2025. That is 7.7 times in four years.
Interest income over the same period went from ₩1,399,169m to ₩2,183,195m, up 56%.
Net interest, therefore, went from ₩1,320,184m to ₩1,573,453m. Nineteen percent, over a period when total assets went from ₩64.8tn to ₩74.0tn, up 14%. The investment margin has been roughly flat while the funding cost has compounded.
The first half of 2026 continues it. Interest expense of ₩387,157m against ₩291,160m a year earlier, up 33%, while interest income rose about 6%.
The cause is visible in the filings. DB has been issuing debt securities that count as regulatory capital: three separate issuance decisions filed on 16 June 2026, one of them amended on 29 June, following a debt securities registration finalised on 2 June. Financing activities brought in ₩1,500,530m of cash in FY2025, against outflows of ₩1,058,392m in FY2023 and ₩449,898m in FY2024.
This is the cost of the strategy. Hybrid capital lets an insurer hold its solvency ratio up without diluting shareholders, and it funded a ₩2.5tn acquisition. It is not free, and the coupon runs forever.
Worth flagging for anyone working from the same data. The quarterly cash flow figures in DB's interim filings are cumulative year to date, not discrete three-month values, whatever a data note may say.
The check is simple. Cash was ₩1,713,419m at December 2025, ₩1,022,439m at March 2026 and ₩1,155,694m at June 2026. The stated first-quarter net change in cash is negative ₩690,980m, which matches December to March. The stated second-quarter figure is negative ₩557,725m, which matches December to June — not March to June, which was positive ₩133,255m.
So first-half operating cash flow was ₩6,293,966m, of which about ₩3,890,654m fell in the second quarter alone. First-half investing outflow was ₩7,311,778m, and the second-quarter portion includes the cash that went out the door for Fortegra.
Get this wrong and you will double-count the second quarter.
The strongest counterargument is that OCI volatility in an insurer is noise around an economic position that is roughly hedged. If assets and liabilities both reprice with rates, then the equity swings are an artefact of imperfect accounting symmetry rather than a real change in the firm's value. Nobody's claim got better or worse when DB's book value fell ₩2.7tn.
There is something to that. It is why regulators built K-ICS on an economic-value basis in the first place, and why the market prices Korean insurers well below book without much apparent distress about it.
The interest expense point has a rebuttal too. Capital securities are cheaper than equity for a company trading at 0.86 times book, and issuing them to buy a US specialty insurer earning around twelve times is a defensible trade even at current coupons. The alternative — issuing shares below book — is worse.
What I would not accept is the version where none of it matters at all. Distributable profit is a statutory calculation, the dividend coverage ratio the company just adopted as a policy variable is built on it, and a perpetual coupon is a permanent prior claim on cash ahead of shareholders. At ₩609.7bn a year and rising, the interest bill is now about a third of net income.
At roughly ₩12.1tn of market value — that figure implies about ₩184,000 a share and is not a live quote — against ₩14.0tn of June equity and FY2025 net income of ₩1,790.6bn, the stock is somewhere near 0.86 times book and under seven times earnings. I do not know how much of that ₩14.0tn is attributable to controlling shareholders versus minorities, which now matters more than it did with 202 subsidiaries in the group.
The K-ICS ratio in the third-quarter disclosure, and specifically whether the Fortegra acquisition and the June capital issuance moved it in opposite directions by similar amounts.
Second, the accumulated OCI balance in the FY2026 annual report. If it has returned to positive territory, the FY2023-24 book value destruction was a rates episode and is over. If it has not, something structural in the asset-liability match is still unresolved.
Third, interest expense in the second half. Another 30% year-on-year increase would mean the funding cost is growing faster than the investment book that services it, and that is a squeeze no amount of accounting explanation makes go away.
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.