Explainer · Japan & Korea allocators

Insurer capital charges, explained

How Japan's ESR and Korea's K-ICS decide what a life insurer can buy from a fund manager

Written 3 September 2026 · All explainers

How to read the markers. [D] — verified from a source retrieved for this note, listed under Sources. [I] — industry-understood; not formally disclosed by a named party. [S] — our own inference or judgment. Every figure is either sourced or marked as an estimate.

The short version

When a Japanese or Korean life insurer looks at a private credit, infrastructure or private equity fund, the first number it reads is not the yield. It is the capital charge: the share of the asset’s value the regulator makes the insurer hold as its own capital against it. At a 49% charge, every ₩100 invested locks up ₩49 of capital that could otherwise back new business. At 20%, ₩20. The yield has to pay for that capital before it pays for anything else.

Both countries set those charges through an economic-value solvency regime built on the same template, the IAIS Insurance Capital Standard (ICS). Korea’s K-ICS has applied since 1 January 2023, with a ten-year phase-in. Japan’s ESR (economic value-based solvency ratio, often called J-ICS) was promulgated on 23 July 2025 and applied for the first time at the year-end of 31 March 2026. Both measure required capital as the loss in a one-in-200 year, and both put the same headline figures on the main buckets: 35% for listed developed-market equity, 49% for unlisted or “other” equity, 25% for real estate. The bucket an asset lands in, not its return, decides whether an insurer can afford to own it — and a fund the insurer cannot see through lands in the 49% bucket by default.

They diverge where it matters most to a fund manager. Korea charges qualifying infrastructure equity 20% and long-held equity 20%; in April 2026 it extended the 20% to renewables and AI data centres, cut qualifying venture investments from 49% to 35%, and changed how unfunded fund commitments are counted. Japan carries the ICS infrastructure charge of 27% and has declined any policy discount for long-term or infrastructure investment.

1. What the ratio measures

A solvency ratio compares two things. Available capital (Japan’s “qualifying capital”) is what the insurer actually has: assets minus what it owes policyholders and others. Required capital is what the regulator says it must hold given its risks. The ratio is available divided by required; 200% means twice the requirement. [D — Milliman; FSC 2022]

What changed is how the two sides are valued. Under Japan’s old solvency margin ratio (SMR), in force since 1996 with a 200% threshold, insurance liabilities were largely carried at book value. Under the ESR both sides are at economic value: the liability is the present value of expected payouts, discounted at current market rates, plus a margin for uncertainty (the “MOCE”, margin over current estimate). Korea made the same move in 2023 when K-ICS replaced its risk-based capital (RBC) regime alongside IFRS 17. [D — JCIA; Skadden; Korean Re]

So the balance sheet now moves with interest rates. A life insurer holding forty-year guarantees has a liability that grows when yields fall, and it wants long-dated assets that grow in step. This is why market risk — interest rate, spread, equity, real estate, currency — was 57% of Japanese life insurers’ required capital in the FSA’s 2024 field test, with equity the largest item within it. [D — Milliman, Figure 41]

Japan’s ESR floor is 100%, with intervention steps at 100%, 70% and 35%. Korea’s statutory floor is 100%; its supervisory comfort level was lowered from 150% to 130% in June 2025. Nippon Life reported a regulatory ESR of 195% consolidated and 204% non-consolidated at 31 March 2026; the Korean industry average was 216.1% at the same date, after transitional measures. [D — Nippon Life; Businesskorea/FSS; Kim & Chang]

The economic-value balance sheet and the solvency ratio Left: assets at market value. Middle: liabilities at their current estimate plus a margin over current estimate, with available capital as the difference at the top. Right: required capital as a stacked bar of risk modules (market, life insurance, credit, catastrophe, operational) reduced by diversification. The headline ratio is available capital divided by required capital, shown as an illustrative 200 per cent. Economic-value regime · Japan ESR and Korea K-ICS Available capital is a difference; required capital is a sum of stresses Assets at market value bonds · loans equities · funds real estate infrastructure cash · other Liabilities and capital liabilities at economic value Available capital assets − liabilities margin (MOCE) current estimate of policy payouts, at market rates other liabilities = Required capital 99.5% one-year stress, by module market risk life insurance credit cat · operational less diversification gross net compare ESR / K-ICS ratio available ÷ required 200% illustrative floor 100% in both; Korea guides 130% A new asset adds its stress to the required-capital bar; a long-dated bond that moves with the liabilities can shrink it. Ratio = available capital ÷ required capital · proportions illustrative
Figure 1. The economic-value balance sheet. Assets at market value on the left; liabilities at their current estimate plus a margin on the right; available capital is the difference. Required capital is the sum of risk modules less diversification. The ratio is the headline. Proportions are illustrative, not any insurer's figures.

2. How a charge is built

Both regimes define required capital as the fall in available capital in a 99.5% one-year Value at Risk event — a stress that should occur once in 200 years. Japan’s ESR states this explicitly, following the ICS target; Korea raised its confidence level from the RBC regime’s 99% to 99.5% with K-ICS. [D — Milliman; IAIS Level 1 text; FSC 2022; Korean Re]

For each risk the regulator prescribes a stress, a specific bad-year move: listed developed-market shares fall 35%; real estate falls 25%; a corporate bond carries a factor that rises with lower rating and longer maturity. The charge is what that stress would take off the insurer’s net assets, so “a 35% charge” means “assume equities lose 35% in a year and hold that loss as capital now”. [D — IAIS Level 2 text; Milliman; KIRI]

Japan’s ESR has six modules: life insurance, non-life insurance, catastrophe, market, credit and operational risk. K-ICS has five: life/long-term insurance, general P&C, market (including interest rate), credit and operational. A correlation matrix combines them, so the total is less than the sum — diversification. In Japan’s 2024 field test diversification cut the life industry’s gross requirement by about 20%, and the tax effect by a further 17%. [D — Skadden; Korean Re; Milliman]

Two consequences follow. The stress applies to the net position, so an asset that moves opposite to the liabilities can reduce the charge — the appeal of long-dated fixed income. And because of aggregation, the marginal charge on a new asset depends on what the insurer already holds: the bucket figure is a ceiling; the effective charge is lower and differs by insurer. [I]

3. The buckets, Japan and Korea side by side

The table lists only charges stated in the regulators’ documents, the IAIS texts, or a named actuarial or asset-management publication. “Not retrieved” means the number exists but none of the sources listed states it; “not published” means no such distinct charge is known.

Asset bucket Japan ESR (from 31 Mar 2026) Korea K-ICS (from 1 Jan 2023) ICS reference (IAIS, Dec 2024) Source
Listed equity, developed markets (incl. domestic) 35% 35% 35% (before dampener) Milliman Fig. 33; KIRI Table II-8; IAIS L2-226
Listed equity, emerging markets 48% 48% 48% same
Infrastructure equity (qualifying) 27% developed / 37% emerging 20% 27% / 37% Milliman Fig. 33; KIRI; J.P. Morgan AM
Long-held equity (≥10-year documented holding) no separate bucket 20% none FSA 2024 paper (declined); KIRI
Unlisted / “other” equity, incl. PE and any fund not looked through 49% 49% 49% Milliman; KIRI; IAIS L2-226(d)
Qualifying venture investment (Korea, 2026 rule change) — 35% (was 49%) — FSC 16 Apr 2026
Hybrid debt / preference shares 4%–35% by rating 4%–49% by rating 4%–35% by rating Milliman; KIRI; IAIS Table 17
Real estate 25% 25% (20% for mandatory-hold property) 25% Milliman; KIRI; IAIS L2-229
Corporate bond, credit factor by rating and maturity follows ICS structure; J-ICS table not retrieved not retrieved ~10-year: AAA–AA 2.1%, A 3.2%, BBB 5.6%, BB 9.8%, unrated 12.7% Milliman Fig. 40; IAIS Table 23
Unrated infrastructure debt not published as a separate figure discount exists; figure not published 75% of the corporate unrated factor (9.5% vs 12.7% at 10 years) IAIS calibration doc; J.P. Morgan AM
Government bonds (own currency) interest-rate module; credit factor not retrieved 0% credit factor for central government — KIRI

The equity buckets are identical at the headline. Both regimes copied the ICS stresses of 35%, 48% and 49%. Japan dropped the ICS “neutral adjusted dampener” (a ±10% adjustment that softens the stress after a crash), so its 35% is a flat 35%. [D — Milliman; FSA 2024 paper]

Infrastructure is where they part. Korea’s 20% is 15 points below its listed-equity charge and 7 below the ICS figure Japan carries, and Korea’s 20% long-holding bucket has no Japanese equivalent. The FSA considered a “reduction of risk factors for infrastructure and long-term investment” in its 2022, 2023 and 2024 papers and each time kept the field-test specification, citing regulatory arbitrage and the risk of blunting insurers’ own risk management. [D — KIRI; FSA 05_1 and 07_1 papers]

For credit, a rating counts for more than seniority. Under the ICS table both regimes descend from, a ten-year BBB exposure carries 5.6% and the same exposure unrated 12.7%; unrated infrastructure debt gets 75% of that, or 9.5%. Japan admits seven rating agencies (R&I, JCR, Moody’s, S&P, Fitch, DBRS, AM Best) and does not allow insurers’ internal ratings. [D — IAIS L2 Tables 23 and 26; IAIS calibration; Milliman; FSA 07_1]

Capital charges by asset bucket: Japan ESR and Korea K-ICS Horizontal bar chart. Six asset buckets, each with a Japan bar and a Korea bar showing the capital charge as a percentage of asset value. Listed developed-market equity 35 and 35. Listed emerging-market equity 48 and 48. Infrastructure equity 27 for Japan and 20 for Korea. Unlisted or other equity, including funds not looked through, 49 and 49. Real estate 25 and 25. Long-held equity: no Japan bucket, Korea 20. Axis from 0 to 50 per cent. Capital charge by asset bucket · % of asset value Same template, one difference that matters: infrastructure and long-held equity Japan ESR (from 31 Mar 2026) Korea K-ICS (from 1 Jan 2023) 0% 10% 20% 30% 40% 50% capital charge, % of asset value (stress: fall in value held as capital) Listed equity, developed incl. domestic shares 35% 35% Listed equity, emerging 48% 48% Infrastructure equity qualifying · Japan developed-market figure 27% 20% · extended to renewables and AI, Apr 2026 Unlisted / other equity PE · any fund not looked through 49% 49% Real estate 25% 25% Long-held equity ≥10-year documented holding Japan: no separate bucket 20%
Figure 2. Capital charges by asset bucket, Japan ESR and Korea K-ICS, as a percentage of the asset's value. Sources: Milliman (Feb 2026) for Japan; Korea Insurance Research Institute (2024) and the FSC's 16 April 2026 release for Korea. Japan has no long-held equity bucket, so that bar is absent.

4. Funds, look-through and unfunded commitments

A fund commitment is not itself a bucket. The regulator wants to charge the assets inside the fund, and the rule is look-through: identify each underlying holding and charge it by bucket. The ICS sets the ladder both countries follow: full look-through if possible; partial look-through on the Basel III pattern if not; and “when no look-through is possible, the full investment is considered as unlisted equity” — the 49% bucket. [D — IAIS L2-1 to L2-3]

Japan adds a middle rung, the mandate approach: if full look-through is impossible, assume the fund fills its documented limits with the highest-charge assets first, then the next highest. Leverage inside the fund must be counted in every case. [D — Milliman section 2.1.3; FSA 07_1] Korea uses the same ladder; a Korean valuation firm selling fund-decomposition services puts it in one line: full look-through gives the lowest charge, “asset reconstruction” (undecomposed holdings as unrated credit at 12.7%) is higher, and no look-through — 49% on the whole fund — is highest. [D — Korea Asset Pricing, industry source]

Unfunded commitments are the trap. A blind-pool fund draws capital over years, but the charge can apply to the commitment, not the drawn amount. Korea’s FSC said so on 16 April 2026: where drawdown timing was unset, the risk factor had been applied to “the entire remaining commitment”, producing charges that were “excessively high”. The fix applies the charge to an expected drawdown computed with credit conversion factors (CCFs), the estimated share of a commitment that will be called. The same release stops treating a fund as leveraged when its borrowing is documented as liquidity-only and under one year. [D — FSC 16 Apr 2026; KIRI weekly] Japan’s treatment of unfunded commitments was not found in any retrieved source; see Caveats.

5. Korea’s moving parts

Phase-in. K-ICS came with optional transitional measures for up to ten years: gradual recognition of the fall in available capital from market valuation, of the newly added insurance risks (longevity, lapse, expense, catastrophe), and of the higher equity and interest-rate charges, the last two starting at 60% recognition. Nineteen of 53 insurers applied in 2023 (12 life, 6 non-life, one reinsurer); they must disclose ratios before and after, and face dividend limits. [D — KIRI; AC Actuarial] The industry’s 216.1% at March 2026 is the after-measures number. [D — Businesskorea/FSS]

The 130% line and the 2027 core-capital rule. In June 2025 the FSC lowered the supervisory recommendation from 150% to 130%, the level to keep after redeeming subordinated debt. From 2027 a second ratio applies: core capital (paid-in capital and retained earnings, excluding hybrid and subordinated instruments) must be at least 50% of required capital, 80% recommended, with a transition to 2035. Six insurers were below 50% at June 2026. [D — Kim & Chang; FSC Mar 2025; Seoul Economic Daily; Insurance Business Asia]

April 2026 easing. At its fifth “productive finance” meeting on 16 April 2026 the FSC announced: policy-programme investments (the National Growth Fund) cut from 49% to 20% or less, stacking with the long-holding treatment to reach 16%; qualifying venture investments cut from 49% to 35%, applied at fund level without decomposition; qualifying infrastructure widened from roads and ports to renewable energy and AI facilities; the guaranteed portion of a partially government-guaranteed infrastructure loan treated as risk-free; and the fund changes in section 4. The FSC estimated up to ₩24.2 trillion of extra lending capacity. [D — FSC 16 Apr 2026]

Co-insurance. Korea permitted co-insurance in 2020 so that primary insurers could cede whole blocks of high-guarantee liabilities, “including the investment portion”, to a reinsurer ahead of K-ICS and IFRS 17. Korean Re’s ₩700 billion block from Samsung Life in October 2023 is the reference deal; reinsuring away interest-rate risk frees required capital for higher-charge assets. [D — FSC Jan 2020; Skadden; Reinsurance News]

6. What this means for a fund manager’s pitch

The number an insurer’s investment committee computes is capital-adjusted yield: the return on the capital the asset ties up. Two assets each returning 8%: at a 49% charge the insurer earns 8 on 49 of capital, about 16%; at 20%, 8 on 20, or 40%. A senior infrastructure loan yielding 6% at a 9.5% charge earns 63% on its capital. That is why unrated, equity-like returns compete poorly on an insurer’s balance sheet even at a higher headline IRR. [S — our arithmetic on the retrieved charges; diversification and tax lower every figure]

Four things follow, all about the bucket rather than the return.

  1. A rating moves the bucket. Under the ICS credit table a BBB rating, public or private, roughly halves the ten-year charge against an unrated exposure (5.6% against 12.7%). Both countries key credit charges to rating and maturity; Japan names its seven agencies and excludes internal ratings. [D]
  2. Look-through is worth building for. A reporting pack that lets the insurer decompose the fund into rated loans, real estate and listed holdings takes it out of the 49% default. Korea relaxes this only for named policy categories; for everything else, decomposition is the route. [D]
  3. Infrastructure has its own door, differently sized. Korea’s 20%, now covering renewables and AI data centres, is the most favourable bucket open to a foreign manager; Japan’s 27% is the ICS figure. For infrastructure debt, the ICS discounts the unrated factor by a quarter; Korea’s own discount is larger but unpublished. [D]
  4. Drawdown structure is now a capital question in Korea. With CCF-based charging, a documented drawdown schedule reduces the investor’s charge from day one. [D — FSC; S for the inference]

None of this makes an asset better. It changes who can hold it, and at what price.

Caveats

  • Japan’s exact credit-factor table was not retrieved. Milliman confirms the J-ICS charge is “net exposure × risk factor according to the rating and maturity” and that J-ICS follows the ICS structure; the ICS Table 23 values are shown as the reference. The FSA’s 23 July 2025 notices (告示) contain the definitive table and were not opened.
  • Japan’s infrastructure-equity criteria are assumed to follow the ICS Annex 3 definitions; Milliman lists the 27%/37% buckets under “J-ICS and ICS” but the qualifying criteria were not checked against the Japanese notices.
  • Japan’s treatment of unfunded fund commitments was not found in any retrieved source. It is listed as unknown, not as absent.
  • Korea’s charges are quoted from the Korea Insurance Research Institute’s 2024 report and the FSC’s 2026 press release, not from the FSS’s 300-page K-ICS manual, which did not load. The two secondary sources agree with each other and with the ICS values.
  • Korea’s April 2026 changes were announced for institutionalisation “by the first half of 2026”, with further detail promised in Q3 2026. Whether the enforcement rules have been amended as of 3 September 2026 was not verified.
  • The old SMR’s confidence level is not stated in a retrieved source, so the note does not give one; only the 200% threshold and the book-value basis are sourced.
  • Insurer ratios are point-in-time and preliminary: Nippon Life’s 195%/204% are marked preliminary in its own release; the Korean 216.1% is after transitional measures.
  • The capital-adjusted-yield examples in section 6 ignore diversification, tax effects, the volatility component of equity risk, and asset-concentration charges. They rank buckets correctly; they do not price them.
  • Solvency II figures (39%/49%/30%/36%/22%) are quoted from Milliman’s comparison table as a labelled comparator only and are not verified against EU texts.

Sources [D]

All retrieved and confirmed to load on 3 September 2026.

Japan — regulator

Japan — actuarial, legal, trade and insurer sources

IAIS — the template

Korea — regulator

Korea — research, legal, trade and industry sources

Cross-jurisdiction