Expectations and confidence

A café owner will hire another waiter only if she expects more customers. Her customers will eat out more only if they expect to keep their jobs. Their jobs depend on firms like hers expecting good sales. Everyone is guessing about everyone else, and the guesses help decide what happens.

Your best number depended on what you expected everyone else to pick.

If others pick at random, the target is about 33. One step further gives 22, and reasoning all the way gives 0. In the economist Rosemarie Nagel’s 1995 experiments, first-time players guessed about 37 on average, so the target was about 24. Almost half the guesses were close to 33 or 22: one or two steps ahead. Newspaper and magazine games in the UK, Spain and Germany, reported in 2002, found the same steps. More of those readers than Nagel’s players went all the way to 0. Evidence: robust.

In the game, if everyone expects low guesses, everyone guesses low, and the average really is low. A self-fulfilling expectation is a belief that comes true because people act on it. The economy is full of them, starting with confidence.

Imagine a town hearing that job cuts may be coming. Nobody has lost a job yet, but families put off buying cars and holidays. Shops sell less, so they hire fewer staff, and some cut jobs. Now the fear looks justified, so families cut back further.

Firms do the same. One that expects weak sales puts off a new factory, cutting its investment (spending on buildings and machines), so builders earn less. Pessimism → less spending → fewer jobs → more pessimism. Optimism runs the loop the other way.

Prices work the same way. If workers expect prices to rise 6% next year, they ask for 6% pay rises, and firms raise prices to cover them. Households may also buy big items sooner, before prices rise. The expected inflation can turn into actual inflation.

So central banks need to be believed. Many aim for about 2% inflation a year. A credible central bank is one people trust to bring inflation back to target, because it has done so year after year. People then set pay and prices around 2%, and a jump in inflation fades. Their expectations are anchored: they barely move when prices jump for a while.

Bank runs are the sharpest case. In a bank run, savers rush to take their money out at once (see the lesson on money and banks). A bank lends or invests most deposits, so it can’t repay everyone together. If your savings aren’t fully insured and you expect others to withdraw, your best move is to go first.

Northern Rock, UK, 2007. This bank borrowed heavily from investors in financial markets, and in August 2007 those markets all but froze. On 14 September, the Bank of England announced emergency support. That showed how much trouble the bank was in, and savers queued to withdraw. The queues didn’t cause the bank’s trouble, but once they began, they fed on themselves. At the time, deposit insurance (which pays savers back if their bank fails) covered only the first 2,000 pounds in full. In October 2007, Mervyn King, head of the Bank of England, said that once the queues started, other savers “faced every incentive to join them”.

Silicon Valley Bank, United States, 2023. On 8 March, it reported a 1.8 billion dollar loss on investments it had sold. Investors who fund start-up firms urged companies on social media to move their money out. On 9 March, customers took out 42 billion dollars in one day. At the end of 2022, 88% of its deposits were uninsured, so their owners could lose that money if the bank failed. Regulators closed it on 10 March.

Both banks had real problems, but each withdrawal gave the next customer more reason to withdraw.