What broke during Japan's asset price bubble collapse (1989-1992) — the Nikkei's crash and the decades-long real estate unwind that followed — and what warning signs were visible before the peak? Separate the immediate trigger from the structural cause, and explain the mechanism rather than only the sequence of events.

The mechanics were monetary before they were financial. After the 1985 Plaza Accord strengthened the yen and squeezed exporters, the Bank of Japan cut its discount rate to 2.5% by February 1987 and held it there for over two years, flooding cheap credit into an economy with few other outlets — it went straight into land and equities. Banks lent freely against real estate collateral under the assumption codified as tochi shinwa, the "land myth," that Japanese land prices simply didn't fall; keiretsu cross-shareholding let banks and their affiliated industrial firms hold each other's stock, inflating each other's reported capital and collateral value in a closed loop, while corporations practiced zaitech — borrowing cheap to speculate in equities and property rather than reinvest in operations. The Nikkei 225 rode this to a close of 38,915.87 on December 29, 1989, its all-time high. The BOJ started unwinding it in May 1989, hiking the discount rate five times to 6% by August 1990 under governor Yasushi Mieno, who'd taken office that July explicitly to deflate the bubble. The Nikkei broke almost immediately: it lost roughly 60% of its value within about two and a half years, falling under 15,000 by mid-1992. Real estate values, which had reached genuinely absurd extremes — the grounds of the Imperial Palace in Tokyo were, by contemporary estimates, worth more than all the real estate in the state of California, and Japan's total land value reportedly ran to roughly four times that of the entire United States despite a fraction of the land area — cracked more slowly than stocks but never really stopped falling for another decade-plus. What followed was less a crash than a forty-year unwind. Banks sat on non-performing loans collateralized at bubble-era land prices and, rather than write them down, kept extending credit to insolvent borrowers ("evergreening"), producing zombie banks and zombie companies that suppressed growth without triggering visible failure. The BOJ cut to zero rates by 1999 and launched the first modern quantitative easing in 2001. CPI stayed flat-to-negative for years. The "Lost Decade" of the 1990s stretched into what's now routinely called the Lost Two Decades. The Nikkei kept grinding lower in real terms, bottoming near 7,055 in March 2009, and didn't reclaim its December 1989 closing high until February 22, 2024 — a 34-year, two-month round trip, the longest of any major developed-market index on record. The warning signs were legible in real time, not just in hindsight: Nikkei stocks were trading at price-to-earnings ratios north of 60-70x against a historical/US norm near 15x; land-to-GDP ratios were multiples of historical Japanese norms; loan growth to the real estate sector was far outpacing GDP growth; and lending was increasingly secured by appreciating collateral value rather than borrower cash flow — the BOJ's own 1989 tightening cycle was itself an admission that policymakers saw the imbalance. The structural lesson is that collateral-based lending concentrated through cross-shareholding doesn't fail in a single event — it fails slowly, for a decade, as marks refuse to catch down to reality and bank balance sheets absorb the gap in silence rather than in one visible shock.

The chronology holds up well - Mieno's tightening cycle, the December 1989 peak, and the roughly 60% drawdown by 1992 are correctly sequenced, and the Imperial Palace-versus-California land value comparison is the right shorthand for how untethered valuations got. Where I'd push back is on treating BOJ rate hikes as the primary trigger: the sharper blow to property specifically was the Ministry of Finance's April 1990 "window guidance" directive capping real-estate lending growth at banks, a quantity restriction on credit that hit land speculation more directly than the discount rate alone. The piece also underweights why the hangover lasted so long - it's not just zombie-loan evergreening as a bank choice, but that Japanese banks weren't required to mark cross-shareholdings to market until fair-value accounting reforms arrived around 2001, so regulators let insolvency go unrecognized for over a decade rather than banks merely opting for forbearance. That regulatory and accounting permissiveness, more than the bubble's size itself, is a big part of why it took until February 2024 for the Nikkei to reclaim 38,915.

The signals were plain arithmetic, not hindsight: the Nikkei traded near 60x earnings at its December 1989 peak against roughly 15x in the US, and Japanese land was appraised at about four times the value of all land in the United States — on a landmass smaller than California. What made it foreseeable rather than merely expensive was that the zaitech loop was visibly circular: corporations booked speculative financial gains as earnings, and banks lent against cross-held shares and land whose collateral value was the very thing being inflated, so any interruption in credit had to unwind both sides at once. Mieno's tightening from 2.5% to 6% between May 1989 and August 1990 was the tacit admission — the central bank saying out loud that asset prices had detached from the real economy, and continuing to say it after the market broke. The winners were those who took that literally: Julian Robertson's Tiger was short Japanese equities into the decline, and the Nikkei put warrants Japanese insurers wrote and Western banks repackaged for US buyers around the turn of 1990 paid off enormously, precisely because the domestic institutions selling that downside believed the floor was policy-guaranteed. That's the part I'd underline: the losing side wasn't uninformed, it was institutionally incapable of pricing a decline it had spent a decade underwriting — which is exactly why evergreening followed so naturally, and why the round trip took until February 2024.