Japan 1989 (bubble & lost decades)

Definition

At the end of 1989 the Nikkei peaked near 39,000 with Japanese equities around 40%+ of world market capitalization and Tokyo land priced at multiples that made the Imperial Palace grounds notionally worth more than California. The subsequent unwind ran DECADES: the index bottomed ~80% lower in 2009 and did not reclaim its 1989 peak until 2024. It is the KB’s most important counterexample to “markets always come back on a human timescale” — an entire developed market where buy-and-hold from the peak spent 34 years underwater nominally.

How it works / structure

• The bubble mechanics (BOJ retrospective): loose post-Plaza-Accord monetary policy, bank credit expanding against LAND collateral whose price the credit itself was inflating (philosophy-reflexivity collateral loop in textbook form), cross-shareholding networks recycling paper gains into more buying, and equity valuations (Nikkei P/E above 60) rationalized by land holdings priced in the same loop.
• The unwind’s anatomy: the BOJ tightened deliberately into the bubble (1989-90); land and equities deflated together, converting bank capital into fiction; the documented policy lesson (Okina et al) — the damage came less from the crash than from the DECADE of unrecognized bad loans and zombie lending that followed; deflation entrenched; rates hit zero and stayed (the original ZIRP/QE laboratory — macro-fed-balance-sheet’s precedents were built here).
• The investor arithmetic (engine-relevant): from the peak, nominal index recovery took 34 years (dividends and currency soften but do not rescue the lesson); from the mid-1990s onward, valuation- disciplined entries did fine — the catastrophe was PRICE-INDIFFERENT entry at the top of a collateral-loop bubble, not “Japan.”
• The allocation lesson: 1989’s world-index weight meant passive global investors held ~40% Japan at the worst moment — cap-weighting concentrates into whatever is most expensive (style-passive-indexing concentration-drift caveat, at its historical maximum).

When it applies

Cited against recovery-assumption sizing (drawdown budgets that assume V-shapes — 2020 — must also carry the L-shape, risk-max-drawdown-budget severity calibration); for cap-weight concentration audits (any market or sector approaching dominant index weight); for collateral-loop diagnostics (credit growing against self-inflated collateral is the checkable fingerprint); for policy-response regime analysis (Japan is the original case study for what zero-bound economies do).

Risk profile & failure modes

• Timescale complacency (the central lesson): “stocks always recover” was true in Japan only at a 34-year horizon — recovery assumptions are REGIME-conditional, and no retail plan survives three underwater decades.
• Valuation dismissal at index level: index P/E 60 was rationalized then exactly as later bubbles rationalized theirs — aggregate-level reverse-DCF arithmetic applies to markets, not just stocks.
• The zombie-decade extension: policy that defers loss recognition extends stagnation (the documented Japanese lesson) — post-bubble economies carry DURATION risk beyond the crash itself.
• Misuse: “Japan proves equities fail” over-reads a peak-entry story; disciplined entries throughout the aftermath earned normal returns — the lesson is price, concentration, and timescale, not futility.

Evidence & limits

The BOJ’s own retrospective (Okina-Shirakawa-Shiratsuka 2001) documents the bubble and policy anatomy; the market record is public history. Counterfactuals (earlier loss recognition, different policy) remain debated — the KB cites the observed path and its mechanics.

Falsifiable-thesis examples

Illustrations only, not signals:

• “Any single market exceeding 35% of world index weight underperforms the ex-that-market world index over the following decade (concentration check)” — falsified by the historical cohort (n is small; labeled as such).
• “Credit-to-GDP growing >20 points in 5 years with land/asset collateral concentration precedes banking-system stress within a decade (collateral-loop fingerprint)” — falsified by the BIS-series cohort.

Cross-references

• The loop machinery: philosophy-reflexivity; the policy laboratory: macro-fed-balance-sheet
• The allocation caveat: style-passive-indexing, ext-international-equities
• The recovery-assumption discipline: risk-max-drawdown-budget, bias-recency
• The sibling bubbles: episode-dotcom-2000, episode-nifty-fifty-1972