The Anatomy of the Bear Ch. 3: Valuation as a Statement About Regime

Investment Background

This chapter has to open by saying what it is not going to do, because the temptation to do it is enormous and the library already has the answer.

It is not going to teach you how to calculate a valuation ratio. stocks-for-the-long-run-siegel chapter 5 owns the cyclically adjusted price-to-earnings ratio — what it is, why a ten-year average is used, what starting levels have historically implied for the following decade, and, crucially, that it has explanatory power over ten years and almost none over one or two. That chapter also states the caveat that matters most: the absolute level drifts over time, so comparing today's reading directly against 1930's is problematic. All of that is settled and is not restated here.

This chapter is about a different use of the same numbers. Not "what return does this valuation imply" — that is Siegel's question and Ilmanen's. But "what does this valuation, in combination with these other conditions, tell me about which regime I am standing in." Those are different questions and the second one is the one nobody in the library has asked.

The Wall Street Translation

Why Valuation Alone Is Useless and Valuation Plus Conditions Is Not

A single valuation reading tells you almost nothing about timing, and Siegel already proved it.

The market was expensive in 1996 and got much more expensive for four more years. It was cheap in 1931 and got sixty percent cheaper. Anybody who has used a valuation level as a trade trigger has discovered this personally and expensively.

Napier's argument is that this is the wrong test to apply to the instrument. A valuation reading is not a forecast of the next move. It is a description of a state — the same way a thermometer reading of forty degrees does not predict when a fever will break, but does tell you unambiguously that you have one.

And states have durations. The reason valuation matters over ten to twenty years and not over one is that the regime, not the price, is the slow-moving object. A regime does not end because valuation reached a threshold. It ends when the conditions producing it end — and those conditions are observable separately.

The Co-Incidence Condition

This is the operational core of the book, and it is a claim about CONJUNCTION rather than about any single measure.

At each of the four bottoms, several conditions held simultaneously. No one of them was sufficient. All of them together had occurred four times in a century.

Condition	What it looked like
Equities below replacement cost	The market pricing established, functioning businesses below the cost of rebuilding them from scratch
The reaction function had changed	Bad news arriving and no longer producing sustained declines, over months
Consensus had become structural	Not "stocks are risky" but "stocks are finished as an asset class"
The monetary condition had turned	Deflation ending, or the policy stance reversing after a long period in one direction
Coverage had thinned	The market had left the front page; there was nothing to say about it

The word doing the work is "simultaneously". Cheap valuation occurred repeatedly without a bottom. Terrible sentiment occurred repeatedly without a bottom. A changed reaction function occurred repeatedly without a bottom — 1930 again. The conjunction is what was rare, and the conjunction is what the four dates share.

Where Replacement Cost Comes In, and Why This Book Will Not Teach It

The replacement-cost comparison — the market value of companies against what it would cost to rebuild their assets — is the measure Napier leans on hardest, and this book deliberately does not teach its construction.

Two reasons, and the second one is the important one.

First, the arithmetic is genuinely contested. Reasonable people disagree about how to value intangible assets, how to treat an economy whose output is increasingly not made of factories, and whether the measure means the same thing in 2020 as in 1932. A chapter teaching a contested calculation as if it were settled would be doing the reader a disservice.

Second, and more importantly: a retail retirement investor should not be making an allocation decision on the strength of a number they computed themselves from a contested methodology. That is the mechanics trap. The useful thing to take from the measure is the QUESTION it asks — is the market pricing productive assets below what it would cost to recreate them — and the observation that the answer has been unambiguously yes four times in a century.

Where a reader genuinely wants a forward return estimate, the library already routes cleanly. stocks-for-the-long-run-siegel chapter 5 gives the CAPE-to-return relationship with its caveats. expected-returns-ilmanen gives the full decomposition of what you are being paid for. This chapter adds neither and defers to both.

What Changes Between Regimes: The Same Number, Different Meaning

The most important and least intuitive claim in the chapter is that an identical valuation reading means different things in different monetary regimes.

A price-to-earnings ratio of fifteen in a world of two percent inflation and a two percent policy rate is a different object from a ratio of fifteen in a world of nine percent inflation and a fifteen percent policy rate. In the second case, a government bond is paying fifteen percent and the equity is competing against that. In the first, the bond is paying two percent and the equity is competing against almost nothing.

This is exactly why Siegel's caveat about the drifting absolute level exists, and it is why a mechanical valuation threshold has never worked as a trigger. The threshold is not a constant. It moves with the regime. Which produces the chapter's central and slightly circular-sounding but genuinely operational conclusion: you cannot read the regime off the valuation, because the valuation only means something once you know the regime. The regime is read from the monetary and policy conditions, and the valuation is then interpreted inside it.

Division of Labor With the Rest of the Library

Book	Owns
stocks-for-the-long-run-siegel ch05	CAPE and the starting-valuation-to-forward-return relationship, with its caveats — including that it is not a timing signal
expected-returns-ilmanen	Valuation as an input to expected return, and the decomposition of premia
mastering-the-market-cycle ch03, ch06	Cycle location from credit spreads, sentiment and valuation percentiles — the three-to-seven-year instrument
security-analysis, margin-of-safety	Valuing an individual business, and the margin of safety on a single security
capital-returns	The capital cycle — supply response and capex at sector level
This book	Valuation as a diagnosis of REGIME, and the conjunction that identified four bottoms

The boundary with mastering-the-market-cycle needs restating here because both books use valuation percentiles. Marks reads a percentile to answer "how hot is it now", and adjusts posture over the next few years. This chapter reads the same percentile to answer "what sample am I in", over the next fifteen. Marks's instrument will move from cold to hot and back several times inside one regime. Neither reading invalidates the other, and a plan needs both for different decisions.

Executable Trading Rules

1. Never use a valuation level as a trade trigger. Siegel chapter 5 states the evidence and it is unambiguous: every approach treating CAPE as a timing signal has performed poorly. Nothing in this chapter changes that, and a reader who converts a regime diagnosis into an entry rule has misread the whole book.

2. Use valuation to set the planning assumption, not the allocation. When long-run valuation sits at a historical extreme, lower or raise the real return you assume in your retirement projection. This is the change Siegel chapter 5 endorses and it is the only one this book endorses too.

3. Require conjunction before treating anything as a regime signal. One condition means nothing. The four bottoms were identified by five conditions holding at once, and every one of those conditions individually occurred many times without a bottom following.

4. Read the monetary condition before you interpret the valuation number. The same ratio means different things at a two percent policy rate and a fifteen percent one. A threshold copied from another era is a threshold fitted to a regime you are not in.

5. Do not compute a contested measure yourself and then bet on it. If you want a forward-return number, take it from stocks-for-the-long-run-siegel chapter 5 or expected-returns-ilmanen, use it as a planning input, and leave the allocation alone.

Relevance to a Retirement Portfolio

The retirement use of this chapter is narrow, specific, and entirely about the assumption rather than the allocation.

A retirement projection contains an assumed real return, and that number is doing more work than any other input in the model. retirement-decumulation-mechanics chapter 1 explains that the four percent rule is itself the output of such projections run across historical starting points. What this chapter adds is that starting valuation is the single most informative thing you know on the day you retire about which part of that distribution you are likely to draw from.

The correct response to a poor reading is not to delay retirement or to sit in cash. It is to widen the adjustment mechanism — the guardrails in retirement-decumulation-mechanics chapter 4, the buffer in chapter 3. A plan that can flex its spending by ten or fifteen percent in a bad decade survives regimes that a rigid one does not, and it does so without requiring anybody to forecast anything.

And the standard recommendation is unchanged: a core of low-cost, globally diversified index funds, no leverage, a cash buffer covering essential spending, and withdrawal rules containing an adjustment mechanism. This chapter does not propose valuation-based allocation shifts. It proposes valuation-informed spending flexibility, which is a much weaker claim and a far better-supported one.

Chapter 4 asks the question this chapter has been avoiding: if the conditions were observable, why did nobody act on them?