Cross-Shareholding in Two Idioms: Chaebol and Keiretsu

Two architectures of mutual holding

Korean and Japanese cross-shareholding share the surface — minority equity positions held by listed firms in counterparties — but differ on every architectural feature that matters.

Korea: chaebol pyramid (and historical circular). The chaebol group structure is family-controlled at the apex. A founder family holds equity directly in a holding company or top-tier subsidiary, which holds equity in second-tier subsidiaries, which hold equity in third-tier subsidiaries. Circular structures (third tier holding equity back into the apex or sister firms) were substantially curtailed by the 2014 amendment to the Korean Monopoly Regulation and Fair Trade Act (passed by the National Assembly at the end of 2013, effective in 2014), which prohibited the formation or strengthening of new circular cross-shareholdings by domestic affiliates of large business groups.^[2] Existing circulars were not banned but were subject to disclosure and unwind pressure; the 2025 amendment to the Korean Commercial Code added further governance reforms aimed at chaebol structures. The economic substance has remained: family control of all listed entities through minority equity, with cross-shareholding as the instrument that levers minority shareholding into majority control. Korea's chaebol cross-shareholding is vertical, family-anchored, and instrumentally deployed for control concentration.

Japan: keiretsu lateral and bilateral. The horizontal keiretsu (Mitsubishi, Sumitomo, Mitsui, Mizuho, Fuyo, DKB-Sanwa) are not family-controlled. There is no apex shareholder. Instead, member firms — typically a trading house, a major bank, an industrial firm, a real estate developer, and several manufacturers — hold minority equity in each other in a roughly symmetric pattern.^[3] The vertical keiretsu (Toyota Group, Honda Group) follow a different logic: an apex industrial firm holds equity in component-supplier subsidiaries, and those subsidiaries hold equity back into the apex — but the structure is bilateral, not pyramidal, and most affiliated firms remain separately listed with their own boards. Japan's keiretsu cross-shareholding is horizontal-or-bilateral, sectorally-anchored, and instrumentally deployed for stable supplier relationships, M&A defense, and political-economic coordination.

The same word names two different mechanisms. The framework, applied to each, reads each through the axis where the underlying governance risk actually concentrates.

Two distribution shapes from the same framework

Note 1 established Japan's archetype distribution: 41% Celestial, 46% Chameleon, 87% concentrated in two labels.^[4] Korea's distribution, derived from the same framework architecture applied to the 2,662 listed firms in KOSPI + KOSDAQ:^[5]

Korea is a Chameleon monoculture. Japan is bipolar.

The same framework, with the same axis architecture (T-axis / B-axis / R-axis), the same indicator stack adapted for market-specific conditions, the same archetype classifier producing the same five labels — applied to two markets and returning two fundamentally different shapes.

The shape difference reflects two things. First, the maturity of the disclosure floor. Japan's T-axis has saturated through a decade of CG Code revisions, pushing nearly every firm above the disclosure adequacy threshold. Korea's T-axis has not saturated — filing-timeliness gaps and disclosure-completeness shortfalls remain meaningful at the universe level. Second, the depth of structural reform. Japan's audit-governance structure (Note 5) shows partial movement toward majority-independent committee structures — 5% Three Committees, 46% Audit-and-Supervisory, 40% Kansayaku-with-voluntary, 8% Kansayaku-alone — a four-tier mix that, taken together, lifts most firms above the structural floor. Korea's chaebol governance structure has resisted equivalent reforms; the family-controlled board remains dominant.

Where the framework reads each mechanism

The framework architecture is uniform across the eight markets. The axes — T (Transparency), B (Balance of Power), R (Risk-of-Conflict) — carry the same indicators in skeleton, with market-specific adaptation. The cross-shareholding mechanism shows up on different axes in the two markets, and this is the heart of the comparison.

Korea: B-axis weakness. The chaebol cross-shareholding architecture, by enabling family control through minority equity, captures the board.^[6] Outside directors at chaebol-affiliated firms tend to be technical compliance with the regulatory minimum, not substantive checks on family decisions. Audit committees lack teeth because committee members are nominated within the chaebol governance ecosystem. Parent-subsidiary structures concentrate power upward. The framework's B-axis indicators — independent-director ratio, audit committee independence, parent-sub structure — collectively register the chaebol architecture as B-axis weakness. Korea's modal Chameleon is B-weak, and the board-architecture problem traces directly to the chaebol cross-shareholding system.

Japan: R-axis weakness. The keiretsu cross-shareholding architecture has a measurable B-axis footprint — the framework's B-01 indicator (Note 2) reads cross-shareholding intensity directly through share-of-net-assets ratios — but the dominant Chameleon-pathology weakness shows up on the R-axis (Risk-of-Conflict). Capital efficiency profiles, related-party transactions, and large-shareholding patterns all read the keiretsu architecture from different angles. Japan's modal Chameleon is R-weak, and the conflict-of-interest problem traces directly to the keiretsu cross-shareholding system.

The same word, one syllable difference: B-weak in Korea, R-weak in Japan. The framework reads each mechanism through the axis where the underlying governance risk actually concentrates.

This is one of the framework's empirical findings worth pausing on. A board-centric governance assessment — the model used by the largest commercial governance ratings — would read Korea's chaebol pathology accurately and miss Japan's keiretsu pathology by structural mismatch. A conflict-centric assessment would do the inverse. The cross-market comparison is what reveals the mismatch.

Two reform trajectories

Korea and Japan have both undertaken cross-shareholding reform in the past decade, but the trajectories differ in pace and scope.

Japan's decade of code revision. Japan has compressed cross-shareholding through coordinated regulatory and market pressure: the Stewardship Code was introduced in 2014 and revised in 2017, 2020, and 2025 (Version 3.0);^[7] the Corporate Governance Code was introduced in 2015 and revised in 2018 and 2021;^[8] the 2022 TSE Prime market reform required ≥1/3 independent directors and enhanced disclosure;^[9] and the 2023 FSA Action Programme for Accelerating Corporate Governance Reform: From Form to Substance, alongside the TSE 2023 capital-cost initiative, intensified pressure on capital efficiency including cross-shareholding reduction.^[10] By ACGA's measure, the simple-average ratio of strategic shareholdings to net asset value across TOPIX 500 firms fell from 13.5% in fiscal year-end March 2015 to 8.4% in fiscal year-end March 2023 — substantial compression though the residual remains structurally significant, with the 6.1% above-ISS-line cohort that Note 2 traced concentrated in regional banks and mid-cap holdouts.^[11]

Korea's incremental reform path. The Korean reform record has been less coordinated. The 2014 Monopoly Regulation and Fair Trade Act revision (described above), the introduction of the Korean Stewardship Code in 2016, and FSC oversight on chaebol-related transactions have all moved the needle in specific dimensions.^[2] The August 2025 amendment to the Korean Commercial Code introduced expanded shareholder rights including mandatory cumulative voting and independent audit committee members — a meaningful step but, as commentary has noted, addressing certain governance concerns without fundamentally dismantling chaebol cross-shareholding and circular ownership architecture.^[12] The 88% Chameleon distribution suggests the floor has not been substantially raised; firms across the universe still register at least one axis below the adequacy threshold. The reform pressure has reached the regulatory framework but not the chaebol cross-shareholding architecture's core.

The trajectories explain the distribution shapes. Japan's bipolar split reflects a reformed-but-residual landscape. Korea's monoculture reflects a structure that has absorbed regulatory pressure without fully compressing.

Two KS centers

Both markets have a Kill Switch population — firms the framework reads as so structurally compromised that all standard archetype labels are bypassed. The KS centers in the two markets reveal where each market's pathology actually lives.

Japan's KS center: regional banks. Of the 115 KS firms in Japan Prime, the largest sectoral concentration sits in regional and second-tier banks — 18 of 69 listed banks register above the ISS-recommended threshold of 20% strategic shareholdings to net asset value, the structural condition the framework's KS-1 trigger reads.^[11] The three Megabank holding companies — MUFG, SMFG, Mizuho FG — are not among the 18; they have publicly committed to and reported substantial cross-shareholding reduction over the past decade. The regional banks remain. Japan's KS center is a sectoral concentration that the secular reform pressure has not fully reached.

Korea's KS center: chaebol-architecture intersection. Korea's KS firings concentrate where the chaebol B-axis weakness intersects with R-axis indicators — typically related-party transaction patterns, holding-company R-axis weakness (the 115-firm holding-company cohort that Korea Foundation Series Note 3 mapped), and treasury-stock disposition patterns (the 1:1,066 disposition-to-cancellation pathology that Korea Note 5 traced).^[5] Korea's KS center is a control-architecture concentration distributed across the chaebol group structure rather than concentrated in a single sector.

The two KS centers reflect the two governance pathologies. Japan: a sectoral residual after a decade of broad reform. Korea: a structural residual within an unreformed control architecture.

What the comparison teaches

Three readings.

First, the framework's cross-market application is not a uniform translation. The same axis architecture and same indicator stack produce different findings in each market because the underlying mechanisms map onto different axes. The eight-market production reading is not "the same governance assessment applied uniformly." It is "the same framework, reading different mechanisms through the axes where they actually concentrate."

Second, the cross-shareholding word should not be assumed to mean the same thing across markets. Korean cross-shareholding is family-control infrastructure read on the B-axis. Japanese cross-shareholding is mutual-holding infrastructure read on the R-axis. An analyst sourcing across both markets needs market-specific structural literacy — neither the Korean nor the Japanese version is a default governance pathology that the other can be reduced to.

Third, the framework's signature cross-market finding — that the dominant Chameleon weakness varies across markets — disciplines any global governance reading.^[13] Japan's R-weak dominance and Korea's B-weak dominance jointly suggest that no single axis is the universal weakness signal. The implication is direct for the commercial governance-ratings industry. The leading systems — ISS QualityScore, MSCI's governance pillar, Sustainalytics governance score — lead with board-centric metrics and apply broadly uniform weighting across markets. The framework's eight-market evidence suggests this approach systematically under-weights what actually moves the Japanese governance distribution. Cross-market work requires market-specific axis weighting at the empirical level, even with uniform framework architecture at the methodological level.

The comparison closes a loop. Note 1 mapped Japan's bipolar archetype distribution. Note 2 traced cross-shareholding as the dominant mechanism behind that distribution. Note 5 traced audit-governance structure as the second mechanism. Note 6 places both findings in cross-market context: Japan's mechanisms differ from Korea's, the framework reads each through the appropriate axis, and the comparison is the cleanest natural experiment in the eight-market stack.

The bipolar Japan distribution and the monoculture Korea distribution are not in tension. They are two outcomes of the same framework reading two structurally different governance systems. The comparison is what makes the framework's cross-market design defensible.

Apex G-Score™ Japan Foundation Series. Production cohort: TSE Prime, 1,576 issuers (1,556 rated; 20 NR). The "FY2025" label denotes a rolling-window cohort — each firm's most recent yuho filed within 2025-03-11 to 2026-04-15. Fiscal year coverage by 決算月 (fiscal year-end month): 3月 (March) ~76.5% (reporting period 2024-04 to 2025-03); 12月 (December) ~13.7% (calendar year 2025); 9月 (September) ~5.1% (2024-10 to 2025-09); 6月 (June) ~4.6% (2024-07 to 2025-06). The December-end sub-cohort shares the calendar-year reporting window of the Korea cohort. Framework: Apex G-Score v2 (T 0.30 / B 0.30 / R 0.40); archetype classifier: unified-v1. Sample firm-level scores in this article are limited to the IP-guardrail-v2.0 disclosed set (Toyota Motor; Samsung Electronics and Reliance Industries appear in cross-market comparisons only). All other firm-level scores remain unpublished. Multi-year indicators where cited are noted explicitly with source vintages. This article is research output, not investment advice; readers should consult their own advisors before making investment decisions.