From Main Bank to Mizuno: Japan's pre-reform governance architecture

TL;DR - Postwar Japanese governance rested on five reinforcing institutions: keiretsu, main bank, cross-shareholdings, insider boards with kansayaku auditors, and lifetime employment. - At the 1988 peak, over 50% of TSE market capitalisation sat in cross-shareholdings; by 2017 it had fallen below 10%. Bank ownership collapsed from 20%+ to under 5% over the same period. - The disassembly produced a monitoring vacuum: the main bank no longer disciplined management, but no replacement node existed. That vacuum is the problem the Stewardship Code (2014), Corporate Governance Code (2015) and GPIF stewardship framework under Hiromichi Mizuno were built to fill.

The five-pillar system

Through the 1970s and 1980s, Japanese corporate governance ran on five mutually reinforcing institutions. Each individually was unusual; in combination they produced an unusually insulated managerial class.

Keiretsu — horizontal industrial groups (Mitsubishi, Mitsui, Sumitomo, Fuyo, Sanwa, Dai-Ichi Kangyo) and vertical supplier networks. Each horizontal keiretsu was anchored by a city bank and a sogo shosha (general trading company), with manufacturers, insurers and real-estate companies clustered around them.
Main-bank system — a long-term relationship lender that also held equity, monitored management quarterly, and intervened in distress. The framework was formalised in Masahiko Aoki's writings during the 1980s and was, for a time, the export-grade theory of how Japanese capitalism delivered superior coordination.
Cross-shareholdings (mochiai) — reciprocal equity stakes between group firms, designed to lock in friendly voting blocs and deter hostile takeovers.
Insider boards with kansayaku monitoring — large boards (often 20–40 members) composed almost entirely of internally promoted "salaryman" executives, with a parallel kansayaku (statutory auditor) board supplying the only formal monitoring role.
Lifetime employment (shūshin koyō) and seniority pay — making the firm a community-of-employees rather than a community-of-shareholders.

graph TD
    MB["Main Bank<br/>(equity holder + lender + monitor)"]
    TC["Trading Company<br/>(sogo shosha)"]
    M1["Manufacturer A"]
    M2["Manufacturer B"]
    M3["Manufacturer C"]
    INS["Group Insurer<br/>(equity holder)"]
    RE["Group Real-Estate Co"]

    MB -- equity + loans --> M1
    MB -- equity + loans --> M2
    MB -- equity + loans --> M3
    MB -- equity + loans --> TC
    MB -- equity + loans --> INS
    MB -- equity + loans --> RE

    TC -- equity + supply --> M1
    TC -- equity + supply --> M2
    TC -- equity + supply --> M3

    INS -- equity --> MB
    INS -- equity --> M1
    INS -- equity --> M2
    INS -- equity --> M3
    INS -- equity --> TC

    M1 -- equity --> MB
    M1 -- equity --> M2
    M2 -- equity --> M3
    M3 -- equity --> M1

    RE -- equity --> MB
    RE -- equity --> INS

Stylised horizontal-keiretsu cross-shareholding pattern, 1980s. Arrows represent reciprocal equity stakes. The main bank sits at the centre because it combined three functions — equity holder, primary lender, and the de facto board monitor — that no Western governance system bundled in one institution.

Why the system worked (and for whom)

The five-pillar system was not designed for shareholder returns. It was designed for coordination — between firms in a supply chain, between management and labour, between capital and operating decisions. By those metrics it delivered. Group firms could share information faster than arm's-length counterparties; the main bank could intervene early in distress without triggering bankruptcy; lifetime employment let firms invest in firm-specific human capital because workers would not arbitrage their skills to a competitor.

The cost was a fiduciary structure tilted away from outside shareholders. Surplus was distributed to lenders (in stable interest), to employees (in seniority pay and job security), to suppliers (in long-term contracts) — and only what was left went to dividends or retained for shareholder return on equity. The combined effect was an extraordinarily insulated managerial class with low ROE pressure and a tradition of treating dispersed shareholders as the least pressing stakeholder constituency.

Sidebar — The kansayaku, in one paragraph. The kansayaku (statutory auditor) system was the formal monitoring mechanism Japanese law required of large companies. Unlike outside directors, kansayaku could not vote on board resolutions, hire or fire executives, or set strategy. They could examine accounts, attend board meetings, and report to shareholders. In practice, kansayaku were often retired senior employees of the same company or of its main bank — embedded in the same social hierarchy they were supposed to police. The 2001 Commercial Code reform required that at least half of kansayaku be "outside" auditors and extended their term to four years, but the structural limitation — no vote, no veto, no hiring power over directors — remained.

The slow disassembly, 1990–2014

The Nikkei peaked on 29 December 1989. The five-pillar system began unwinding almost immediately, in three overlapping waves.

Wave 1 — bank impairment (1990–2003). Falling land and equity prices destroyed the capital base of the main banks. Forced disposals followed: between 1981 and 1999 intra-keiretsu ownership of group-firm shares fell from ~25% to ~20%; by 1999 no group had a majority-holding bloc. Bank ownership of listed equities collapsed from above 20% in the 1980s to under 5% by 2014. The main bank's capacity to monitor went away with its capital.

Wave 2 — accounting and regulatory reform (1997–2008). Mark-to-market accounting for cross-held equities (introduced from FY2001) made the cost of holding non-strategic equity visible on the balance sheet. The 2001 Commercial Code amendment strengthened the kansayaku board. The 2002 amendment (effective April 2003) introduced the optional "company with committees" structure — a US-style board with nomination, audit and compensation committees, each majority-independent. Uptake was tepid (about 70 adopters by end-2016 out of ~3,800 listed companies). J-SOX — internal-control-over-financial-reporting requirements under the Financial Instruments and Exchange Act (promulgated 14 June 2006) — came into force from fiscal years beginning on or after 1 April 2008.

Wave 3 — foreign demand (2003–2014). Foreign ownership of TSE-listed equities rose from negligible levels in the 1980s to ~30% by 2014. Foreign investors voted their shares, attended AGMs, and asked the questions that domestic banks had stopped asking when they stopped owning the equity. Cross-held shares fell below 10% of total TSE holdings for the first time in 2017.

The numbers tell a coherent story:

Indicator	1988 peak	2014 (pre-Stewardship Code)
Cross-shareholdings as % of TSE market cap	> 50%	~15% (and falling)
Bank ownership of listed equities	> 20%	< 5%
Foreign ownership of TSE-listed equities	negligible	~30%
TOPIX 500 firms with at least one independent director	n/a	~22%
Listed companies "conscious of cost of capital"	n/a	~40%
Listed companies disclosing cost of capital	n/a	< 10%

The monitoring vacuum

The disassembly created a problem no one had designed for. The main bank had been the de facto monitor of Japanese management. With banks no longer holding the equity, the monitoring function was vacant. Outside directors barely existed (22% of TOPIX 500 had even one as of 2013). Domestic institutional investors held equity but did not engage. Foreign investors engaged but had no institutional voice inside Japan.

The Ito Review's diagnostic of typical pre-reform CEO tenure — 4–6 years, fixed by seniority rather than performance — captures the consequence. Boards staffed by internally promoted lifers, with kansayaku auditors who could not vote them out, accountable to a shareholder base half of which was cross-held by friendly counterparties, faced no credible mechanism for removal except scandal. The Review called the seniority-tenure norm "a fundamental impediment to long-term value creation."

"Japanese corporate governance was, until the early 2010s, a system in which monitoring was provided by an institution — the main bank — that had largely exited the role, and in which no other actor had been authorised to replace it." — paraphrasing Curtis Milhaupt, On the (Fleeting) Existence of the Main Bank System and Other Japanese Economic Institutions (Columbia Law School)

That vacancy is what the post-2013 reform wave was designed to fill. The Stewardship Code (Feb 2014) authorised institutional investors to engage. The Corporate Governance Code (June 2015) authorised — required, with comply-or-explain — outside directors on listed boards. The Ito Review (Aug 2014) supplied the diagnostic vocabulary. GPIF, under Hiromichi Mizuno from January 2015, became the enforcement node.

From main bank to Mizuno

The shift can be drawn as a single migration: the monitoring function moved from the main bank, through a fifteen-year vacuum, to GPIF and its external asset managers.

Main bank (1950s–1990s). Quarterly meetings with management. Equity stake of 3–5% per investee. De facto veto over major capital decisions. No formal mandate; monitoring as a byproduct of being lender and equity holder and trading-group anchor simultaneously.

Vacuum (1990–2014). Banks held less equity, lent less, and increasingly preferred to exit distressed positions rather than restructure them. Outside directors not yet mandatory. Foreign investors voting from offshore custodians. Domestic institutional investors rarely engaging.

GPIF and the new node (2014–present). GPIF accepted the Stewardship Code on 30 May 2014, twelve weeks after the Code's publication. Hiromichi Mizuno became CIO in January 2015 and served until March 2020. Because GPIF is statutorily barred from holding Japanese equities directly, it influences governance entirely through its ~25 external asset managers — making Mizuno's manager-monitoring framework the most powerful private-sector enforcement mechanism in the Code's history. By 2024 GPIF AUM exceeded ¥240 trillion; the additional ¥109 trillion accumulated since 2014 dwarfs every other Japanese institutional pool.

The two endpoints look superficially different — a city banker meeting an industrial CEO in the 1980s, an offshore-domiciled asset manager voting GPIF's beneficiary capital in the 2020s. Structurally they perform the same function: they are the institution outside management that has standing to ask, and is expected to ask, why are you not doing better with our capital?

The legacy that has not yet been removed

Three pre-reform structures remain operationally relevant to IR in 2026:

Residual cross-shareholdings. Despite falling below 10% of TSE holdings in 2017, cross-held equities remain concentrated at certain insurance companies, megabanks, and trading houses. The 2021 CG Code revision and the March 2023 TSE request both require explicit per-stock disclosure of cross-shareholding rationale.
Kansayaku companies. As of 2024 a majority of listed companies still use the kansayaku (Audit & Supervisory Board) structure rather than either the audit-committee or three-committees structure. The Code accommodates all three, but engagement letters from foreign investors increasingly ask why a board has not migrated to one of the latter two.
Lifetime-employment expectations. While the formal contract has eroded, the cultural expectation of inside-promotion and seniority shapes who joins a board. External-director searches still confront a shallow pool because retiring executives from one industry rarely move to another industry's board mid-career.

Each of these is a legitimate engagement topic and will reappear in Themes 4 and 5 of this curriculum.

What this means for IR

Map your own shareholder register to the 1988 → 2024 arc. If domestic financial-institution holdings are above 20%, you are above the current TOPIX median and should expect engagement questions about residual cross-shareholdings.
Know your board's monitoring history. Foreign investors will ask when your company last replaced a director for performance reasons. If the answer is "never," prepare a substantive reply that does not rely on cultural deflection.
Translate kansayaku for English-language audiences. Use "Audit & Supervisory Board" (the formal English term) and explain in one sentence that members do not vote on board resolutions. Misunderstanding on this point is the single most common foreign-investor question on Japanese board structure.
Disclose cross-shareholding rationale per stock, not in aggregate. The 2021 CG Code revision requires this; the engagement standard expects it. "We hold the stake to maintain a stable business relationship" without further per-stock detail is treated as non-disclosure.
Track your foreign-ownership ratio over the long run. A rising ratio is the simplest single index of the unfinished migration from the main-bank governance equilibrium to the post-Stewardship Code equilibrium. If foreign ownership is rising and domestic-institutional ownership is falling, the questions your IR team faces will tilt accordingly.

Sources & further reading

Curtis J. Milhaupt, On the (Fleeting) Existence of the Main Bank System and Other Japanese Economic Institutions (Columbia Law School / SSRN): https://papers.ssrn.com/sol3/papers.cfm?abstract_id=290283
Hideki Kanda, Japan's Audit & Supervisory Board Member System (Japan Audit & Supervisory Board Members Association, Aug 2021): https://www.kansa.or.jp/wp-content/uploads/2021/10/Japans-Audit-Supervisory-Board-Member-System.pdf
Federal Reserve Bank of San Francisco, Japan's Cross-Shareholding Legacy (Aug 2009): https://www.frbsf.org/wp-content/uploads/August-2009-Japans-Cross-Shareholding-Legacy-the-Financial-Impact-on-Banks-august-09-FINAL.pdf
Nomura Institute of Capital Markets Research, Corporate Governance and Reform of Japan's Commercial Code (2002): https://www.nicmr.com/nicmr/english/report/repo/2002/2002sum01.pdf
FSA, Financial Instruments and Exchange Act overview: https://www.fsa.go.jp/en/policy/fiel/index.html
METI, Ito Review of Competitiveness and Incentives for Sustainable Growth — Final Report (Aug 2014): https://www.meti.go.jp/policy/economy/keiei_innovation/kigyoukaikei/ito_review__released_august2014_en.pdf
GPIF, Adoption of Japan's Stewardship Code: https://www.gpif.go.jp/en/investment/pdf/adoption_Japans_stewardship_code.pdf

Previous in this theme: 1.1 Why 8% Was the Number That Changed Japan

Next in this theme: 1.3 Abenomics' Third Arrow: how governance became growth policy