Actionable insights — Our Shopping List: Part 3

1. Size the feared exposure, then compare it to the discount it produced

1. Name the fear precisely — not "sentiment", but the specific balance-sheet item the market is worried about.
2. Find that item as a percentage of assets, revenue or earnings.
3. Measure the drawdown the fear has caused.
4. Compare the two. If the whole company has fallen as though the exposure were total, the mispricing is arithmetic rather than a judgement call.
5. Then check the quality of the exposure itself, so you are not just arguing about size when the small piece is genuinely bad.

• The remaining 79% being correlated with the feared 21% — an asset manager's fundraising slows when its credit book sours, even if the book is small.
• Exposure percentages measured on AUM rather than on earnings, which can understate the profit contribution.

2. Check whether an asset manager's capital can actually leave

1. For any fee-earning business, split assets under management into redeemable, locked-up and permanent.
2. Find the lockup lengths and the share of AUM behind them.
3. Identify permanent sources — insurance balance sheets, listed vehicles, evergreen funds.
4. The fee base that survives a 40% drawdown is the one worth capitalising; the rest is cyclical.
5. Then read fundraising as the growth line and lockups as the durability line, separately.

• Insurance-funded permanent capital bringing its own risk — the liabilities are permanent too.
• Performance fees (carry) being counted as recurring. Only the management fee on locked capital really is.

3. Chart price against revenue over five years to isolate multiple compression

1. Pick a window long enough to span a re-rating — five years works.
2. Plot the share price and a fundamental line (revenue, EPS or free cash flow per share) rebased to 100.
3. The divergence between them is the change in multiple, with no valuation model required.
4. If fundamentals doubled while the price rose 20%, you have measured a de-rating exactly, and the question becomes whether the starting multiple was the anomaly.
5. Apply the same chart to the whole portfolio to tell a stock-picking problem from a style drawdown.

• The starting point being the distortion. A stock that "got quite expensive starting in 2019" was re-rating up then; five years of nothing may just be the round trip.
• Revenue doubling through acquisition rather than organically — per-share figures avoid this.

4. Look for pricing power in components that are a rounding error to the buyer

1. Work out what the product costs as a share of the customer's total spend on the end item.
2. If it is tiny, and the item cannot function without it, price increases are unlikely to be contested.
3. Confirm the customer has no legal alternative supplier — sole-source certification, patent, or regulatory approval.
4. Check the realised increase rate over several years; a consistent 5-6% is the signature.
5. Ask what would make the customer start caring: a big enough cumulative increase, a certified alternative, or a regulator.

• Customers organising. Airlines have complained about aftermarket pricing before, and regulators occasionally listen.
• The cumulative effect: 5-6% a year doubles the price in twelve years, which is when "barely notice" stops being true.

5. Rank switching costs by what has to be re-learned, not by what has to be re-bought

1. Ask what a customer would have to relearn to switch — not what they would have to repurchase.
2. Check where the learning happens. Skills acquired during professional training are the stickiest, because the decision was made before the customer had a budget.
3. Add the risk dimension: in medicine, aviation and security, switching carries a safety cost, not just an inconvenience one.
4. Look for a renewal mechanism — each new cohort of trainees resets the lock-in for another career.

• Hospital purchasing consolidating and overriding individual surgeon preference.
• A hardware head start (Mako's three years) being confused with a permanent one — measure it in installations, which is what is done here.

6. For a small-cap serial acquirer, the moat is deal flow, not product

1. Establish the typical target size and ask who else is bidding for it.
2. If the targets are too small to move a large acquirer's needle, the buyer faces no competition and pays low multiples.
3. Check the operating model: decentralised, owner-retained management is what makes sellers choose you over a private equity auction.
4. Check the funding: interest coverage and leverage decide whether the machine keeps running through a downturn.
5. Check alignment: a controlling family or founder is the usual reason the discipline survives.

• Size killing the model. The same argument used against Constellation in favour of Topicus applies to any acquirer that outgrows its target pool.
• Entry multiples: quality serial acquirers rarely get cheap in absolute terms. The target here is the stock's own six-year floor (20x), not a bargain.

7. When buying a holding company, ask what the discount does not protect you from

1. Split the underlying portfolio into listed and unlisted.
2. The listed share moves with the market daily; the discount does not cushion it, it only sets your entry.
3. The unlisted share is marked by the manager, so it is smoother — and less reliable.
4. Set the entry as a discount level (price-to-book or price-to-NAV), and hold the expectation of volatility separately.

• A holding-company discount that never closes. Family control is what makes the capital allocation good and the discount permanent, at the same time.
• Family control cutting both ways — no activist can force a re-rating either.

8. Apply your own balance-sheet standard to every candidate, including the ones you admire

1. Write down the balance-sheet metrics you require — interest coverage, net debt to EBITDA, debt to equity.
2. Report them for every candidate, not only the ones that pass.
3. When a business model is explicitly debt-funded, say what the leverage is and what happens to it in a downturn.
4. If you make an exception, say that you are making one and why.

• Track records that were produced by leverage in a falling-rate era being extrapolated into a different one — 27.4% a year since 2006 spans exactly that period.
• The asymmetry in your own write-ups: which candidates get a balance-sheet paragraph and which get a compounding statistic instead.