Prices Nobody Corrects

The efficient market hypothesis comes in a defensible version and an indefensible one, and the argument about it usually involves one person defending each.

The defensible version says prices incorporate publicly available information, so consistently beating the market is very hard. This is well supported. The great majority of actively managed funds underperform a simple index over long periods, and past outperformance predicts future outperformance poorly. If you have a reliable way of identifying mispriced assets, the evidence says you probably do not.

The indefensible version says prices equal fundamental values. That is a much stronger claim, it is not required by the first, and the evidence against it is substantial.

Robert Shiller produced the sharpest piece of it. If a share price is the present value of the dividends it will eventually pay, then prices should be less volatile than the dividend stream, because a price is an expectation averaged over many future years, and averaging smooths. Shiller compared actual historical prices with the prices that perfect foresight about dividends would have implied.

Actual prices moved several times more than the fundamentals could justify. Something other than information about future payouts is doing most of the moving, and that excess volatility finding is the empirical foundation of behavioural finance. Shiller and Eugene Fama shared the 2013 Nobel for work pointing in substantially opposite directions, which is a reasonable summary of where the subject stands.

The obvious objection is that if prices are wrong, rational investors should trade against them and make money, correcting the price in the process. This is the mechanism that is supposed to make markets efficient.

Shleifer and Vishny explained why it fails exactly when it is needed, and called it the limits of arbitrage.

Suppose you correctly identify a bubble. Betting against it means losing money for as long as it continues inflating. You are managing other people's money; they see losses mounting and withdraw, forcing you to close the position at the worst possible moment. Keynes had the line long before: the market can remain irrational longer than you can remain solvent.

So the correcting mechanism is weakest precisely when prices are furthest from value, because that is when the correcting trade has been losing money for longest. Several well-known investors who identified the housing bubble early lost most of their clients before being proved right.

Hyman Minsky, ignored for most of his career and rediscovered enthusiastically in 2008, described the process that gets you there.

His sequence has three stages of borrowing. Hedge finance: income covers interest and principal. Speculative finance: income covers interest, and the principal must be rolled over. Ponzi finance: income covers neither, and the borrower depends entirely on the asset appreciating.

The mechanism is that a long period of stability moves the system through these stages. If asset prices have risen for years, lending against them looks safe, so lenders relax terms; borrowers who take more leverage do better, so the cautious ones look foolish and are competed away; and each individual decision is reasonable given recent experience. The system becomes progressively more fragile during exactly the calm that makes it look safe.

Stability is destabilising. That is the whole hypothesis, and it explains why crises follow good times rather than bad ones.

Which brings up the variable that determines whether a crash matters, and it is not the size of the fall.

The dot-com crash destroyed more paper wealth than the housing crash and produced a mild recession. Housing fell less and produced a global financial crisis. The difference is leverage. Tech shares were largely held unleveraged, by people who could absorb the loss. Houses were bought with borrowed money, so a 30% fall wiped out equity entirely, triggered defaults, and transmitted the losses into the banks that held the debt, and from there into everything.

A price fall is a loss for the holder. A leveraged price fall is a loss for the holder's creditors, and their creditors, which is how a sectoral problem becomes a systemic one.

Which leaves the question everybody wants answered and nobody can: whether you can tell at the time. The honest position is that you frequently can identify the dynamic (surveys during housing and crypto booms find buyers who explicitly say they do not think the asset is worth the price and expect someone else to pay more) and you cannot identify the timing. Knowing that a market is running on the expectation of a greater fool tells you how it ends and not when, and the gap between those two is where fortunes are made and lost.