Reflexivity

A price is supposed to be a measurement. It is meant to reflect what something is worth, the way a thermometer reflects a temperature, without altering it. Most of the time it does. Sometimes the measurement is wired into the thing it measures, and then the reading and the reality start moving each other.

Where the wiring is

A company's share price is an opinion about its future earnings. It is also the price of the shares it hands over to buy other companies, the pay of everybody holding options, and part of what a lender looks at before quoting a rate. A higher price makes all three of those easier, which raises future earnings, which was what the price was an opinion about.

That is the whole idea. The measurement is an input to the thing measured, so the loop can run away from the fundamentals in either direction and be right about itself on the way.

A company that gets more valuable by being valuable

Below is a roll-up: a company trading at 12 times earnings that buys smaller companies trading at 10 times, paying in its own shares. Because its shares are dearer, it issues few of them, so the earnings it buys are spread over barely more shares and earnings per share jumps. The market pays a higher multiple for the faster growth. The higher multiple makes the next deal cheaper still.

Then switch the feedback off and run the same businesses again. Same acquisitions, same organic growth, same pool of companies running out on the same schedule. No boom, and nothing to give back.

The pattern is George Soros's account of the conglomerate boom of the late 1960s, set out in The Alchemy of Finance. The figures above come from the model on this page, which is unit tested, including a check that earnings per share rises in every year of the run that ends 70% down.

ConventionThe textbook rule is that an all-stock deal is accretive when the acquirer's multiple is above the target's. That rule ignores growth, and building this model showed the cost of that: at equal multiples the deal is neutral to the level of earnings per share and mildly dilutive to the growth rate, because you paid for earnings that had not grown with you that year. It is a small effect and it points at the large one. A roll-up sold on its growth rate is being valued on the one thing the accretion test does not look at.

Where else the loop shows up

• Bank runs. A bank is solvent if depositors stay and insolvent if they leave. The belief is an input to the fact, which is why deposit insurance is cheap: it changes the belief, so the event it insures against mostly stops happening.
• Housing. Rising prices raise the value of the collateral, so lenders lend more, so buyers can bid more, so prices rise. Falling prices run the same loop backwards and turn borrowers who were fine into borrowers who owe more than the house.
• Currencies. A currency falling raises the cost of servicing debt in foreign currency, which weakens the borrowers, which is a reason to sell the currency.
• Reputation and hiring. A company that looks like it is winning attracts better people, who make it more likely to win. The signal moves the thing it signals, and it works in both directions.

What to do with this

When a price has moved a long way, ask whether it can reach back and change the thing it is pricing. Name the channel: acquisition currency, cost of borrowing, ability to hire, willingness of customers to sign. If there is no channel, you are looking at a normal mispricing that will correct. If there is one, the correction can be a long way off and the fundamentals may have moved by the time it arrives.

And be suspicious of any story where the evidence for the story is the price. A company is valuable because its stock is up, a currency is sound because it is strong, a founder is brilliant because the round was oversubscribed. Those are loops presenting themselves as conclusions, and the way to tell is to ask what the evidence would look like if the price went the other way.