Prices are a wonderful tool for compressing and transmitting information, so in general, more accurate prices are a good thing.¹ But sometimes, they feel entirely too efficient. Stock volatility and stock-based comp injects a little bit of manic depression into every tech workforce. Levered institutions can go into a death spiral when the value the market places on their liabilities becomes an input into future lender's decisions about whether or not to cut them off. If a developing world country goes through an economic hiccup, the forex market can bloodlessly turn this into a currency crisis that throws millions of people out of work.

That doesn't sound so good.

Consider a country with the usual developing country mix that leads to uncertain access to capital markets: unstable politics, dubious public finances, and a legal system that does a better job of punishing speech than protecting rights. When one such country, the UK, introduced an unusually free-spending budget in 2022, yields on government debt spiked and the pound crashed. But one reason for that was that British pensions had reduced their ownership of bonds, because of the return drag, and had replaced that exposure with bond futures, which they had to quickly liquidate when prices dropped. So the reaction to the budget was really a reaction to some longer-term problems: pensions taking on more risk in order to reduce their immediate funding needs, a spike in inflation that broke the two decade-plus streak of bonds moving inversely with stocks, and a leverage-driven liquidation.

But in another sense, this was just a vigorous marking-to-market of various losses. Replacing direct ownership of bonds with a portfolio of riskier assets plus levered exposure to bonds creates this kind of risk; treating the bond/equity relationship as a universal one rather than a feature of low-inflation environments is also a way to get paid a fairly steady return in exchange for inflation risk. And underestimating how fast asset prices can whipsaw also leads to more risk-taking than the alternative. In this case, market volatility was actually showing that the market wasn't efficient enough; more effort spent on portfolio construction would have led to higher bond yields in advance of the 2022 mini-budget, and higher interest costs would have made those tax cuts look less affordable.

A surprisingly adjacent case is when markets are efficient at pricing an asset in reference to supply and demand, as opposed to its cash flows. This shows up most starkly in a corner, when some shareholder or group of shareholders has managed to buy more than 100% of the outstanding shares of a business, or a short squeeze, when prices spike because traders worry about getting cornered. For a brief period in October 2008, Volkswagen was the most valuable company on earth, for the simple reason that Lower Saxony owned 20%, Volkswagen owned 42.6%, they had options to buy another 31.5%. That left 5.9% for the rest of the market to fight over, and in October of 2008 they were particularly unexcited about taking big nonlinear risks. One reason this squeeze was so damaging was that Porsche's buying coincided with a big drop in the market, which made Volkswagen look more expensive. So if you were a long/short trader, it looked like something sensible to short: a cyclical company that was clearly going to suffer in a downturn, with a big enough market cap that there was plenty of liquidity.

The traders who made that bet joined the very big club of market participants who, one way or another, got smoked in Q3 '08 through Q1 '09. But, like the British pensions, the reason they were in a position to lose a lot was that they were rounding a risk from "low" to "zero." In many contexts, this has a small effect on behavior—if there's a 0.1% chance that you'll get hit by a car if you cross the street without looking both ways, it happens one time in a thousand. But when you're participating in a market, you're always providing liquidity to someone who disagrees with you, and you're sending a signal about the state of the world. If you bought British bonds, you were both betting that inflation in the UK would be low and telling the government that it was okay to borrow more. If you shorted Volkswagen because it was fundamentally overpriced, you were offering a great deal to someone who was betting more on the mechanics of the market.

And you can't really have a working market without those complicated mechanics. Market-makers need to be able to short in order to provide liquidity to both sides of the market; that's a much worse business if everyone needs to have enough capital to hold inventory and needs to take the directional risk on how the price of that inventory fluctuates. Options are a natural feature of markets, because there can be efficiencies in the level of prices and inefficiencies in their volatility, which options straightforwardly capture. So if you're going to have all of the tools to accurately price things, you're also going to have a system where sometimes a surprising interaction between these tools produces a crazy result. It's a bit like the presence of bugs as a so-far-inevitable result of turning manual processes into software: computers make fewer little errors like transposing digits, but also don't notice illogical things like the fact that they're sending a utility bill for an amount higher than M2 or something. But the fact that this weirdness is ultimately tolerable, and that the most complex markets are also widely regarded as the most efficient, is a sign that meta-efficiency creates certain kinds of inefficiency that you only see when markets are doing a very good job at the basics.

We’ve covered the nuances of market efficiency from single stocks to macro shifts:

• If you want more background on why bonds used to hedge stocks so well, consider that modern financial history begins in 1998, with the Asian Financial Crisis and its aftermath.

• There’s more than one efficient market paradox.

• One way to understand sudden policy-driven swings is to look at them from the inside, by reading about the banker on the other side of Soros’ legendary pound short ($).

• Even if you’re right about the fundamentals, you can still lose money shorting shares of worthless companies ($).

• A fun efficient markets parable: once upon a time, the news that a company had $450m more cash than investors thought sent the stock down.

Join the discussion!

1  It's possible for society to invest too much talent in making prices efficient, but that's a separate concern. And that same information transmission effect means that this might be the single part of the economy where it's hardest to definitively say we collectively invest too much or too little in it. Usually it's easier to tell a top-down story, like "these people could be curing cancer, but they're trading index options instead," rather than trying to figure out what the socially-optimal spread for at-the-money S&P options expiring in a month is. In principle, it should be easier to defend a conclusion like "we grow too many almonds" or "we produce too much red dye," but if you try to figure out the optimal amount of those to produce, you'll find that it's fractally complicated. And those products play a much narrower and more directly-substitutable role in the economy! If you do find a case where you can confidently say we do too much or too little of something, it usually comes from looking at the consequence of some cognitive bias (I think the world would be a better place if we made fewer cigarettes) or some legal distortion (we'd have fewer personal vehicles if drivers fully internalized the cost of accidents). Turn that argument to the financial sector, and suddenly you have even more to argue about! Does the tax structure subsidize investment research because long-term capital gains are taxed at less than personal income? Or should you treat a capital gain as a change in the net present value of a stream of future dividends, and thus the full tax burden is the corporate tax rate plus the tax rate on dividends plus the tax rate on the gain itself? If a company's expected pretax earnings rise by $1.98, that's $1.56 after corporate taxes. If they plan to pay it as a dividend, that's $1.25 after-tax to the recipient. And if the price of the asset rises by $1.25 and the owner sells, they pay 20% of that gain and are back to a dollar. Making all of those aggressive assumptions at once—like assuming that companies are fully taxed on their economic profits and that prices are all set in reference to what top-bracket taxpayers would owe, is obviously going a bit too far. And it doesn't account for the fact that corporate structures need to have some kind of benefit to be worth adopting in the first place, and that all the tax barriers between earning a profit within a company and enjoying spending money as an owner of that company could be thought of as fees-for-service. Maybe the US government is right up there with cloud computing in jacking up the price of egress. Anyway, long story short, nobody has any idea whether there are too many or too few people in finance. We will never know. It's probably a harder version of the problem of "what is the optimal bid/ask spread to quote for every asset in existence?" which, as we've established, has absorbed the complete attention of many smart people. One potential failure mode of prediction markets describes where efficiency may go too far: prediction market prices/spreads are more accurate and “efficient” if people with inside information of the outcomes of those events and/or control the outcomes of those events participate in the markets. The markets are not perfectly efficient (otherwise those inside or proxy betters would make no money), but marginally more efficient than if they hadn’t known and participated. The price signal is efficient, but the mechanism becomes socially destructive. Insiders are also a tax on the research of smart outsiders, because they reduce market-makers’ willingness to trade. You always want markets to be efficient within some bounds.