By December 1989, the Imperial Palace grounds in Tokyo were famously said to be worth more than all the real estate in California. The Nikkei 225 had hit 38,957 — a level it wouldn't see again for 34 years. Japanese companies were buying Rockefeller Center, Pebble Beach, and Columbia Pictures. Banks were lending against land values that assumed 3% annual appreciation forever. The entire Japanese financial system had become a single hammer — the land-collateral model — and every problem looked like a nail that more lending could fix. Mr. Market was in his most manic phase. He was offering to sell Tokyo office space at 100x earnings and buy rural golf course memberships for $1 million each. The rational track of two-track analysis would have noted that Japanese P/E ratios had reached 60x ...
Popular framing: A speculative bubble popped and Japanese policymakers were too slow to fix it.
Structural analysis: When the land-collateral lending model becomes the single hammer of an entire financial system, every bank balance sheet, regulator's playbook, and firm's strategy is reflexively coupled to the same asset. The lollapalooza of recency bias, career risk, and herding inflates prices past where rational tracks could matter; once the shock hits, the same monoculture means restructuring would cascade through every institution, so policymakers extend rather than resolve — hysteresis locks the economy into a low-growth equilibrium.
The gap matters because the policy-error framing implies Japan's stagnation was contingent and reversible through better technocratic decisions, while the structural framing implies it was the mathematically predictable consequence of allowing a compounding credit-collateral loop to run unchecked. If the structural view is correct, the relevant intervention window was 1986-1988 (before the loop became self-reinforcing), not 1990, and the lesson for other bubble episodes is fundamentally different: you cannot manage your way out of a lollapalooza at peak — you must prevent the conditions that allow one to form.