2601 — China Pacific Insurance (Group) Co., Ltd. Class H

Deep-value analysis — 2601 · China Pacific Insurance (Group) Co., Ltd. (H shares, HKEX)

28 August 2026 · reference price … · H shares: 2601.HK · A shares: 601601.SS · GDR: LSE “CPIC”

Identification warning — read before any figure. The research brief auto-generated this morning (research-2601-20260828.md) describes First Steamship Co., Ltd. (2601.TW), a small-cap Taiwanese dry-bulk shipping company. That is the WRONG company. The tracker (Sumar, row 277) lists 2601 with Exchange = HKEX, name “China Pacific Insurance (Group) Co., Ltd. Class H”, price 30.28 — and today’s verified market price (…) confirms the match. This is the same numeric-ticker-collision class as 1318 / 2145 / 2555 / 3600 in _BVB_ALIASES. I added the alias "2601": ("2601.HK", ...) to deep_data_pack.py and regenerated the data pack, which had come out completely empty. Nothing from the research brief was used in this report — no prices, no events, no “analyst-coverage gap” (which is false: CPIC is covered by 14 analysts). All figures below come from the 2026 interim report (published 27.08.2026, reviewed by Ernst & Young Hua Ming), the FY2025 results presentation (26.03.2026), and market prices.

Executive summary

The thesis. CPIC is China’s third-largest life insurer and third-largest property insurer by gross premiums, with 186.8 million customers, RMB 4,081 bn of assets under management, and an actuarial embedded value (EV) of RMB 636.9 bn at 30 June 2026. The H shares trade at … i.e. RMB 25.85 — 0.39x embedded value per share (RMB 66.21), 0.78x net book value (RMB 33.18), and 6.6x trailing-four-quarter recurring operating profit. The company isn’t declining: attributable operating profit (OPAT) grew 6.2% in H1 2026, new business value (NBV) 12.7%, the NBV margin went from 15.0% to 17.5%, and the property division’s combined ratio fell to 95.0%, the best half-year in the recent cycle. The thesis isn’t “a company recovering”, it’s “a company performing normally, priced by the market as if its actuarial value were inflated by 60%”.

The counter-thesis, equally important. The discount has an identifiable reason, and the reason is quantifiable: embedded value is built on a long-term investment-return assumption of 4.0% per year, while the actual realized net return was 3.4% in 2025 (versus 3.8% in 2024) and ~3.0% annualized in H1 2026. The company’s published sensitivity is brutal: −50 basis points on the return assumption cuts the value of business in force by 27.2% (from RMB 255.1 to 185.7 bn). The market isn’t saying “CPIC is worth 0.39x EV”; it’s saying “the 4.0% assumption is wrong.” The second reason, less forgivable: 32% of 2025’s accounting net profit isn’t operational — RMB 17.1 bn comes from short-term investment volatility on a Chinese equity market that had an exceptional year. The third: it’s a state-linked company, ~41% held by Shanghai state vehicles and the Baowu steel group, with no buyback program and a dividend that still hasn’t returned to its 2020 peak.

Estimated value. Five independent, triangulated models: a DCF on real distributable cash flow (…), a pessimistic DCF with return deterioration (…), Greenwald earnings power value (…), embedded value re-adjusted to a realistic return assumption (…), and the stock’s own four-year trading multiples (…). The resulting range is …-61, with a median of …, i.e. a median margin of safety of … versus the current price. A Monte Carlo simulation over 20,000 scenarios gives a median of … and an …% probability of undervaluation, but with a pessimistic tail at … — the range is wide, and it’s wide for a single reason, the investment-return assumption.

The verdict. A moderate-size buy, not a maximum-conviction position. Three arguments for: the dividend flow is real and covered (4.45% yield, paid from 30% of operating profit, with 255% composite solvency against a 100% regulatory minimum), book value grows steadily (9.5% CAGR since the new accounting standards were adopted), and the consensus of the 14 analysts covering the stock (average target …) converges within 3% of my base DCF — a coincidence that validates the method, not one that makes it redundant. Three arguments for moderation: on its own historical multiples the stock isn’t cheap, it’s near its high (0.78x book value versus a 2022-2025 average of 0.60x; 6.6x OPAT versus an average of 4.78x), the stock has already fallen 21% from its 2026 peak of … with no operational bad news, and the group’s core capital fell over six months (442.4 from 450.6 bn RMB) while the minimum capital requirement rose 5.8% — composite solvency lost 18 percentage points in a single half-year, for a reason (falling interest rates) that hasn’t ended.

Suggested position size: 2… of the portfolio, not 5…. The …% probability of undervaluation comes from a distribution with a real left tail, and the risk isn’t that the business breaks — it’s that the actuarial assumption producing the RMB 66.21/share figure proves optimistic by 100 basis points, in which case “fair” value is somewhere around RMB 53 (…), not … Even so, today’s price stays below half of that.