How an Economy Grows and Why It Crashes
第 6 章 · 22 分钟

Crisis: malinvestment and the cost of liquidation

The strongest passage in the book, and the least falsifiable.

概念地图
Malinvestment
When the rate signal is distorted, projects longer than real saving can sustain get started.
深色为本章概念,浅色为其他章
陪练模式
不想只是读?让它带你走一遍。
导师把这一章拆成小步,每一步先问你,再讲;你用自己的话答,它先认下对的部分再纠偏。用你自己的 AI,对话只存在这台设备上。
对话只存在本机
Locus01

延伸阅读

Mises and Hayek · 1920s–30s · the monetary theory of the cycle
Explained booms as long projects over-started under a distorted rate signal, and slumps as their liquidation. They argued against attributing cycles to exogenous shocks or animal spirits; the cost is that 'over' is relative to an unobservable benchmark, and the hole has never been filled.
Foundational
Fisher · 1933 · debt deflation
Showed that liquidation reinforces itself: selling depresses prices, lower prices raise real debt, which forces more selling. The most direct counterexample to the 'let it clear' position.
Counterexample
Minsky · mid to late twentieth century · the financial instability hypothesis
Locates the source of instability in the calm itself: a long quiet stretch gradually shifts financing structures toward fragile forms. Unlike the Austrian line, it needs no central bank in the leading role.
Revision
Schiff · 2010 · the chapters on bubbles and bailouts
Writes the crisis as the inevitable convergence of the earlier mechanisms and argues against rescue. The book gives no ex ante criterion for what counts as malinvestment; the third crack lands there.
Subject of this course
机制02

When the gap surfaces

The gap left by chapter 2 — a detour length the signal supports but the physical stock cannot — is invisible while projects are starting. It surfaces at the moment the resources needed to continue stop arriving. That is malinvestment: not 'too much investment' but investment whose time structure does not match the time structure of the physical supply. Everything the book says about slumps follows from this one line.

机制A distorted rate signal, through projects started that cannot be finished, produces clustered liquidation when resources run out.
可迁移性测试
Move it to a long voyage: leave port on a faulty fuel gauge and you discover halfway that the fuel will not reach, at which point you can neither continue nor return, only divert to the nearest port. The isomorphism breaks in that a ship's fuel is precisely measurable while an economy's available saving has no gauge at all.
机制03

Liquidation feeding itself

Once liquidation begins, selling depresses prices, lower prices raise the real burden of nominal debt, and more holders are forced to sell. Fisher called this debt deflation. It runs against the previous mechanism: that one says malinvestment must be cleared, this one says clearing itself destroys more than the malinvestment. The book acknowledges only the first, and that is its most substantive disagreement with its opponents.

机制Asset sales, through depressed prices raising real debt burdens, force still more sales.
可迁移性测试
Move it to a margin call: a shortfall triggers forced closure, forced closure moves the price, and the move puts more accounts short. The isomorphism breaks in that an exchange can raise margin requirements or halt trading, while an economy has no corresponding master switch.
Derivation04

From 'malinvestment exists' to 'do not rescue'

What is being derived is the book's policy conclusion, with each step tagged — the controversy sits almost entirely on one of them.

There exists a correct capital structure set by time preference
Chosen, and the heaviest in the chapter. It assumes 'correct' is determinate in principle, though unobservable.
Everything liquidated belongs to the malinvested set
Chosen: it assumes liquidation is precise. This is exactly where Fisher's line objects.
Liquidation itself does not change physical capacity
Chosen: it assumes dismantled plants and dispersed teams can be reassembled at no cost.
推导 · 0 / 4
Worked derivation05

Precision, not size, decides the conclusion

Suppose malinvested projects are twenty per cent of the stock. Compare a collateral-damage rate of zero with one of fifty per cent — what happens to the book's conclusion?
Compute what gets cleared in each case, then what capacity remains.
裂缝06

Cracks in this framework

争议地形07
本书主张
A crisis is the settlement of earlier malinvestment. Rescue and stimulus only postpone the reckoning and keep the wrong structure alive longer, at a higher final cost.
另一种看法
The Fisher and Bagehot line holds that liquidation is self-reinforcing and imprecise, and that lending in a panic to solvent but illiquid parties can break the spiral without preserving the malinvestment.
分歧扎在
The disagreement is rooted in whether liquidation is precise. The book assumes what is cleared deserved to be cleared; the other side assumes collateral damage. Neither disputes that malinvestment should eventually be cleared.
什么证据能裁决
What would settle it is whether the targets can be told apart: group institutions in a crisis by an ex ante solvency measure, then compare three-year survival and output contribution between solvent institutions that received liquidity support and those that did not. No difference would mean support saved nobody who could have survived, and the book is closer to the facts.
The evidence points to liquidation being imprecise: across several crises, institutions later judged sound also failed in the panic. What is missing is a solvency measure usable at the time — existing indicators tend to distort during the same crisis, which is why 'whom to rescue' remains hard in the moment.
Falsification08

When the gap surfaces, as a preregistrable test

The mechanism under test: a distorted rate signal, through projects started that cannot be finished, produces clustered liquidation.

Data source: firm-level fixed investment and project completion records, with policy and market rate series
Only project-level records show 'started and not finished'; aggregate investment cannot.
Sample: all industries with completion records, real estate included
Dropping any industry changes the direction of the result; sealed off in advance.
Window: quarters 4 to 16 after the rate divergence begins
Covers the typical interval from starting a long project to its funding tightening.
Threshold: for projects begun during the divergence, the abandonment rate of long-cycle projects minus that of short-cycle ones is significantly positive
The long/short boundary is fixed in advance at the median planned duration.
Failure condition: the difference is insignificant, wrong-signed, or holds only after splitting the sample by whether a crisis followed
The third counts as failure.
推导 · 0 / 3
接口09
Which slot it hangs on
挂在哪个槽位The conclusion you already hold is probably 'let a burst bubble fall all the way' — a judgement about markets repairing themselves. This chapter goes at its hidden premise.
Does this chapter (a) replace 'let it fall all the way', or (b) constrain it by requiring an estimate of the collateral-damage rate first?
慢变量Register three slow variables: debt-to-assets in the area you follow, the share of long-cycle projects in new investment, and collateral prices. Look at the second first — when it climbs fastest, this mechanism is most active.
小结10
本章小结
01Malinvestment is a mismatch of time structures, not an excess of total investment.
02'A departure must be corrected' is an identity, not a discovery about crises.
03'Do not rescue' holds only if liquidation is precise and spares healthy capacity.
04Fisher's spiral and Bagehot's rule each attack one of those two assumptions.
05The correct capital structure is unobservable, which makes the account self-consistent under any outcome.
提取练习 · 合上书,先自己答一遍。
?Is 'a structure that has departed must be adjusted back' (a) an identity or (b) an empirical claim?
?Does Bagehot's lender-of-last-resort rule attack (a) assumption two (precision) or (b) assumption one (a correct structure exists)?
Forced choice11

C · Locate the crack

In which situation does this framework give a confident and wrong answer?
二选一
Forced choice12

A · Identify the mechanism

Which of these is a mechanism?
二选一
读到这里,把它记为已读。