2628 — China Life Insurance Company Limited

Deep-value: 2628.HK — China Life Insurance Company Limited

Analysis of 28 August 2026. Reference price: … (28.08.2026 close, H shares). Rate: … = 0.8569 CNY. All operational figures in CNY millions unless otherwise specified.

Primary sources used: the FY2025 annual report (summary, Chinese CAS standards, approved by the board 25.03.2026, audited by Ernst & Young Hua Ming + Ernst & Young, unqualified opinion under both standards) — this is the source of all FY2025/FY2024/FY2023 figures; the H1 2026 interim results published on HKEXnews on 27.08.2026; the internal data pack (yfinance, quarterly balance sheet) for the 5-quarter series; tracker Pregatire_investitii_21.xlsx, column ZI.

Note on the SEC filing: China Life deregistered from the SEC (Form 15F-12B filed 13.11.2023, CIK 0001268896), following the ADR delisting from the NYSE. There’s no 20-F or 6-K after 2023, and edgar_10k_downloader.py would only produce three-year-old filings. I replaced the SEC filing with HKEX/SSE filings, which are the primary source anyway for an A+H issuer.

Executive summary

The thesis, in one sentence: you’re buying China’s largest life insurer at 0.45x embedded value and 7.2x normalized profit, with a 3.8% dividend growing at an accelerating pace and zero dilution for nineteen years — but you pay for it by accepting that reported profit is a function of the Shanghai stock exchange, not of underwriting, that the recurring interest margin is visibly closing, and that capital allocation belongs to the Ministry of Finance, not to you.

The surface figures are spectacular and misleading. In H1 2026 the company reported attributable net profit of RMB 134,489 mn, … y/y — the best half-year in its history, the first in the Chinese insurance industry to exceed RMB 100 billion in a half-year. Operating revenue rose 81.5%, to 434,307. On this profit, the H shares trade at a TTM P/E of 4.97x and a P/B of 1.08x. An automated value screen would flag them as obviously cheap.

They aren’t obviously cheap, because the denominator isn’t stable. The quarterly breakdown of attributable profit shows the real amplitude: Q1 2025 +28,800 · Q2 2025 +12,130 · Q3 2025 +126,874 · Q4 2025 −13,726 · Q1 2026 +19,500 · Q2 2026 +114,989 (Q3 and Q4 2025 and Q2 2026 are arithmetically derived from the reported semiannual and annual totals; the Q4 2025 loss is explicitly confirmed by president Li Mingguang, who attributed it to “capital-market structural adjustment”). Between two consecutive quarters, profit amplitude is RMB 140 billion. That isn’t an earnings stream, it’s the mark-to-market of a stock-and-fund portfolio that exceeded RMB 1,040,000 mn in June 2026.

The recurring part of the engine is moving in the opposite direction. Net investment income — the coupon, the dividend, the rent, i.e. the money actually coming in — has fallen in absolute terms: 195,674 in 2024, 193,795 in 2025. On an asset base that grew 12.3%, the net yield fell from 3.26% to 2.76%, and in H1 2026 to roughly 2.71% annualized. The reported total return, 6.09% for 2025 and 5.58% for H1 2026, is the gap between this 2.7% and market gains. The guaranteed cost of historical liabilities sits in the 2.5… band. The margin is closing before our eyes.

Estimated value. Triangulating five models gives an extremely wide range — from … (the five-year median P/EV applied to current embedded value) to … (an optimistic DCF). The center of gravity of the models I consider applicable to a life insurer — a conservative DCF and Greenwald earnings power — sits in the …-46 band. The median of the five models is a margin of safety of ….

A Monte Carlo simulation over 20,000 scenarios gives a median of … and an undervaluation probability of …%. I don’t treat this probability as a truth. It measures assumption uncertainty inside the DCF model; it doesn’t measure the uncertainty of whether a flow-based DCF is the right framework for a life insurer. It isn’t.

Verdict: BUY, but a small position — not a conviction one. The buy argument is that, at … the market is capitalizing in perpetuity owner earnings of roughly RMB 45,300 mn (calculated backward, at r = 9%, g1 = 4%, gt = 2%) — less than 2023’s profit, which was the worst year of the last decade. Equivalent: the market demands a discount rate of ** … in yuan** to hold these shares. The argument for a small position is that the only model calibrated on the stock’s actual behavior over the last five years — historical multiples — is also the only one with a negative margin, at …, and what has prevented the discount from closing hasn’t changed: the state holds 68.37%, has never bought back a share, and the payout ratio has fallen from 17.2% to 15.7% in exactly the year profit exploded.