Actionable insights — AIA Newsletter September 2026

1. Settle a policy question by elimination, not by forecast — enumerate every outcome and kill each on a specific mechanism

1. State the problem as a closed set of outcomes, not as a prediction. For a debt load there are exactly three: grow/cut your way out, default, or inflate it away. If your list is not exhaustive the conclusion is worthless, so start by proving the set is complete.
2. Kill each branch on a named mechanism, not on sentiment. "They won't cut" is an opinion; "the growth is in Social Security and Medicare transfer payments and the retiring cohort is expanding, and no voting bloc will accept reform" is a mechanism you can check.
3. Kill the branches whose consequences are unacceptable to the decision-maker, which is a different test from unlikely. Default is not improbable arithmetic — it is politically impossible because it ends the issuer.
4. Whatever survives is the working assumption. Then go find the historical instance of it and read what actually happened, including the parts that were unpleasant for people holding the surviving asset.
5. Separate certainty from timing explicitly. An elimination argument tells you the destination and nothing about the date, so say so out loud before anyone sizes a position on it.
6. Sanity-check yourself against the doomer failure mode: if your conclusion is the same one you would have reached from temperament alone, tighten the mechanism tests rather than the rhetoric.

• Any branch you eliminated coming back to life — a genuine entitlement reform, or a growth rate that outruns the debt — because that invalidates the whole chain, not one leg of it; the explicit adoption of a long-rate target or a large-scale Treasury purchase program restarted for reasons other than a crisis, which is yield-curve control arriving under another name; and the 10-year yield sitting persistently below realized inflation, which is the observable signature of the surviving branch actually running.

2. Audit the escape hatch before you rely on it — map the regulatory levers that make the "obvious" avoidance impossible

1. Whenever you conclude "the authorities will do X to savers," immediately state the obvious individual defence against X — here, simply refusing to own the bonds — and then test whether it actually survives.
2. Ask who owns the asset in size. Individuals are not the marginal buyer of government debt; banks, pension funds and insurers are. A policy only has to compel the institutions holding your money, not you.
3. Enumerate the levers that compel those institutions without any announcement: liquidity requirements, capital treatment favouring sovereigns, pension allocation minimums, insurance solvency rules, tax preferences, restrictions on foreign investment, directed lending, pressure to invest domestically.
4. Look for the rehearsal — political language reframing private savings as a public resource. The rhetoric reliably precedes the rule, and it is visible years ahead.
5. Conclude at the level the coercion operates: if the exposure arrives through vehicles you do not control, the defence must be ownership of things outside those vehicles, not a decision about what to buy inside them.

• The levers moving one at a time, which is how it always looks — a change to bank liquidity ratios, a pension "domestic investment" consultation, a tweak to insurance capital charges; the EU securitization and savings-union proposals converting from speeches into directives, since Europe is running ahead of the US here and is the observable pilot; and the tell that matters most for a US reader — any proposal touching retirement-plan default allocations, because that is where the un-opt-out-able money sits.

3. Build a book by economic regime, then break the template only where you can name why the hedge fails

1. Start from the admission that sets the design: you cannot predict the next regime. That is not humility, it is the specification — it means the portfolio must be pre-positioned for all of them rather than tilted toward the one you expect.
2. Enumerate the regimes and assign one asset to each job: prosperity, deflation/falling rates, recession and tight money, inflation and monetary disorder. The test of a real allocation is that some sleeve is always losing.
3. Check that your existing portfolio is not fake diversification — hundreds of positions that all need the same conditions (rising asset prices, easy credit) is a single bet with many tickers.
4. Now allow yourself exactly one kind of deviation: a sleeve whose hedge you can argue is broken, with the mechanism stated. Not "I don't like bonds" but "the specific policy I have concluded is coming — yield-curve control plus persistent inflation — is the thing that disables this sleeve."
5. Replace, don't just delete. A regime left unhedged is a hole; find another asset that does the job under the policy you expect.
6. Define the rebalancing rule up front — annually, or on tolerance bands (a sleeve reaching ~35% or falling to ~15%) — so the discipline is mechanical rather than a judgement call in a panic.
7. State the mandate so it cannot be confused with your other books: this one is for preservation of accumulated wealth, and building it is allowed to take issues, not one write-up.

• Which asset he nominates to take over the deflation job the bond sleeve is being cut back from — that is the real test of whether this is a designed portfolio or a hard-asset tilt with a framework attached; whether the promised rebalancing discipline actually gets applied to a winner (selling the sleeve that worked is where these portfolios fail in practice); and mandate drift — a 3x-10x-shaped name appearing in a book explicitly built for preservation.

4. Run the implementability screen on any advice before adopting it — solve for your reader's constraints, not the ideal case

1. Take the standard prescription for the problem — usually written for someone far wealthier than the person reading it — and state it in full and fairly before touching it. Credit the source; do not straw-man it.
2. Score it against the reader's real constraints: capital, access to competent advisors, time, appetite for administration, and tolerance for counterparty and fraud risk in unregulated corners.
3. Name the capital threshold at which it flips from sensible to absurd, concretely. A range that includes a real person is more useful than "high net worth."
4. Flag the adverse-selection problem in the service industry that has grown up around the advice — the smaller the client, the worse the counterparties available.
5. Keep the objective and discard the implementation. If the goal is not depending on one government's currency and rules, ask what achieves that inside a normal brokerage account.
6. Prefer the liquid proxy on an explicit second ground: it can be sold quickly. When the risk you are hedging is disorder, exit speed is part of the hedge.

• The liquid proxy failing in exactly the scenario it was chosen for — a REIT is a security in a brokerage account inside the same jurisdiction whose rules you are hedging, so it inherits the capital-controls risk raw land does not; whether he ever names the threshold above which physical ownership becomes worth the headache; and the general trap this screen exists to catch — adopting a framework for its authority while quietly lacking the means to execute it.

5. Value an asset play with the P/E switched off — price the acreage, then let normalized earnings be the second question

1. Decide up front which kind of security you are looking at. If the value sits in an irreplaceable, slowly-appreciating asset and the earnings are cyclical, say explicitly that the multiple is not the tool — otherwise you will reject the stock at the bottom of its cycle, which is the only time you want it.
2. Confirm the earnings weakness is cyclical, not structural. Depressed end-demand with the asset base intact qualifies; a shrinking market for the output does not.
3. Build the asset value from observable per-unit comps — actual transactions, ideally the company's own — and use a deliberately wide band rather than a point estimate.
4. Multiply out to a gross figure and then say what it isn't. A gross asset value is not an equity NAV: debt, taxes, corporate costs and illiquidity all sit between the two. Refusing to launder gross into NAV is what keeps the estimate honest.
5. Anchor on the trough earnings number anyway, as the floor case — knowing what the business makes at the bottom tells you what you are being paid to wait.
6. Check the capital-return policy, because on a long hold the distribution framework, not the multiple, is what converts patience into cash.

• The gap between the gross timberland range and the actual enterprise value closing — if the market re-rates before the cycle turns, the asset argument was the whole return; per-acre transaction prices in the company's regions, which is the one input the whole valuation rests on and is published by others; the falsifier for "cyclical not structural" — a durable shift down in US housing starts or a substitution away from wood products; and the trap this method invites, using "asset play" as a permanent excuse for earnings that never normalize.

6. Judge capital allocation by the multiple spread between what management buys and what it sells

1. Collect every disclosed acquisition and disposal over a comparable period, with size, price and price per unit.
2. Ignore the headline prices. A high price per acre may be excellent and a low one terrible — the per-unit price tells you about the asset's quality, not about the decision.
3. Convert each side into a multiple of the cash flow the asset produces, then compare. Buying at a lower multiple than you are selling at is value creation by arithmetic, independent of any forecast.
4. State the assumption the whole exercise rests on: the multiples come from management's own forward estimates, so the conclusion is conditional on those estimates being honest.
5. Look for the pattern over several transactions rather than one. A single good trade is luck; consistent spread is a process — and it is the cheapest available evidence about a management team you will never meet.

• The spread narrowing or inverting in later disclosures — buying at higher multiples than you sell is the same evidence running the other way; whether the forward estimates underpinning those multiples are ever revisited after the fact, since the qualifier is doing real work; and the disclosure itself disappearing, because managements that stop publishing per-transaction economics usually have a reason.

7. Inventory the optionality on a single asset — and separate what already earns from what merely might

1. Take one physical asset and ask what else the same unit can produce simultaneously, without displacing the primary use. Layers that stack rather than compete are the ones worth counting.
2. List them concretely and, where possible, with the contracted acreage or project count — an option with a signed agreement attached is a different animal from one described in a slide.
3. Split the list into already earning and speculative, and say which is which. Cash flowing today with lumpy timing is not the same claim as a use case that needs a market to exist first.
4. Identify each layer's binding constraint, because the constraint is what determines how much of the asset actually qualifies. Distance to a grid connection, distance to a city, proximity to a suitable geological formation — these turn "10 million acres of optionality" into a much smaller real number.
5. Note where the accounting flatters the economics: assets carried at decades-old historical cost convert at very high reported margins, which makes the earnings lumpy and easy to misread as a one-off.
6. Finally, discount all of it against what actually drives the price near term, and say so — otherwise the optionality inventory becomes a reason to ignore the tape indefinitely.

• The CCS exploration agreement converting into an executed injection lease with disclosed economics — that is the moment the least-appreciated option becomes a number; data-center and industrial land sales appearing as a named line rather than inside "real estate," since that is the layer with the largest per-acre uplift; and the standard failure — an optionality stack that stays permanently one announcement away while the shares track lumber anyway.

8. When a resource company misses a stated goal, sell first and ask questions later — and keep the commodity view separate from the company

1. Write down, in advance, the specific milestone a pre-revenue resource company must hit — a financing package closed, a JV partner signed, a permit granted, a production date. The thesis is the milestone, not the ore body.
2. Treat the first miss as the exit, not the second. The rule exists because the second, third and fourth misses feel like sunk-cost decisions and the first one does not.
3. Watch the funding path rather than the deposit. In a pre-revenue miner the question is never whether the metal is there; it is who ends up owning it. Repeated equity issuance answers that question against you.
4. Quantify the dilution as ownership, not as price. The relevant number is your share of the same mine after each raise.
5. Keep the commodity thesis and the company thesis on separate ledgers. Exiting a name for a balance-sheet reason implies nothing about the metal, and conflating them is how people sell the sector at the bottom.
6. When you break your own rule, log the breach as a rule, not as a bad-luck story — it is the only way the next instance gets caught.

• Share-count growth quarter over quarter in every pre-revenue holding — the number that makes this rule mechanical rather than a judgement call; an ATM facility or shelf registration being filed, which is the announcement before the dilution; financing announced as "advanced discussions" for a second consecutive reporting period; and whether the rule actually gets applied next time to a name he likes, since it has now failed once precisely because he liked the mine.

9. Read the customer's spending decisions as the supplier's leading indicator — and audit the index before you trade the headline

1. For a services or equipment supplier, stop forecasting its revenue directly. Forecast its customers' capital budgets, which are decided first and disclosed publicly.
2. Trace the cash chain with its lags: commodity price rises → producers' cash flow improves → exploration and development budgets are approved → contracts are awarded → the supplier's revenue prints. Each arrow takes quarters, and the share price usually moves at the arrow before the one you are watching.
3. Prefer suppliers whose binding constraint is a scarce input (skilled crews, specialised rigs) rather than capital — a constraint competitors cannot buy their way past is what stops the upcycle competing away margins.
4. On the demand side, when a headline number supports your position, audit its construction before citing it: who builds it, what are the constituents, how liquid are they. Publish the flaw alongside the number.
5. Then say which part you are actually trading — the underlying trend, or the number. If you cannot explain the number, trade the trend and disown the number.

• Producers' announced exploration and development budgets for the next fiscal year — the input to the whole chain, published months before any supplier's revenue; day rates and utilization rather than reported revenue, since they turn first; whether the labour constraint becomes a margin problem instead of a moat (wage inflation outrunning pricing); and, on the frontier side, a Uzbek index reconstitution or disclosure of constituents, which would finally make the headline number usable.