Join our community of smart investors

Bubbles: what we know, what we don't

Bubbles are common, but they needn't be very dangerous for investors or the economy
December 5, 2017

The huge rise in the price of Bitcoin – up from $900 to over $11,000 so far this year – has prompted talk of there being a bubble in it. But what is a bubble, and what do we know about them? Here are some things.

Bubbles are common

You might think we know this from history. From Tulipmania in 1636 through the South Sea bubble in 1720 and railway shares in the 1840s to tech stocks and credit derivatives, we have many examples of assets becoming overpriced and then crashing.

But it’s not as simple as that. Take one stock. It rose from under $2.50 in 1997 to over $70 two years later only to fall back below $6 in 2001. This looks like a classic bubble and burst tech stock. It was. But it is now priced at over $1,100. The stock is Amazon. In retrospect, it was actually underpriced even at the peak of the tech bubble.

So, was Amazon in a bubble in 1999 or not? Viewed from 2001, the answer was yes. Viewed from today, the answer is no.

A bubble exists when an asset’s price is far above its fundamental value. But in real time we cannot tell for sure what fundamental value is, because it depends upon future growth. In Amazon’s case, that growth has vindicated 1999’s valuation, even though it looked bubbly at the time and in the following months.

And here’s the thing. It wasn’t entirely unreasonable for investors in 1999 to have supposed that there were more companies that would have grown a lot, as Amazon did. In retrospect, such a supposition was wrong. But it wasn’t obviously stupid. High prices can occur if investors attach some probability to a future boom, and they can crash as that probability is revised down. It would, though, be harsh to call such episodes bubbles. In a world where the future is unknowable, mispricings are inevitable.

Some people have been talking about a 'bond bubble' for years – and yet bond prices have continued to rise. This warns us that it’s far too easy to call a bubble, when in fact the fundamentals might actually justify high prices.

You might object that while all this applies to shares, it’s not true of Bitcoin whose fundamental value is nothing.

Maybe. But those pieces of polymer and paper in your wallet also have a nugatory fundamental value. They are 'worth' £10 and £20 only because other people believe they are worth that. Bitcoin is 'worth' over $11,000 because other people believe that’s what it’s worth. What’s the difference? It’s that years of experience justifies our supposition that others will accept these scraps of paper as being 'worth' £20 whereas the basis for assuming Bitcoin to be worth $11,000 is not so rooted in longstanding habit. Cash benefits from being familiar, whereas Bitcoin is freighted with uncertainty. Is this really a strong enough basis for calling one a bubble and not the other?

Given that history isn’t so clear, why do I say that bubbles are common? It’s because we have laboratory evidence. Researchers have created artificial securities which – unlike real world ones – have a known fundamental value because they are designed to have certain dividends and a certain future terminal value. When subjects trade these artificial assets under laboratory conditions, they are often grossly overpriced. “Markets for longer-lived assets have a strong tendency to generate price bubbles and crashes,” says Tilburg University’s Charles Noussair in a review of experimental markets.

It’s this evidence, more than real world experience, which suggests that bubbles exist.

 

Bubbles are hard-wired into our minds

Researchers at Caltech have connected traders in artificial assets to fMRI scanners to monitor their brain activity. They’ve found that price patterns are correlated with activity in the nucleus accumbens, a part of the brain associated with processing rewards. People who buy when activity in this region is high tend to lose money because they buy when prices are high. People who don’t buy tend to do better. Success, it seems, goes to those traders who can resist the lesson of positive reinforcement.

What’s more, they also show that buying during bubbles is correlated with activity in the ventromedial prefrontal cortex, part of the brain associated with processing other people’s intentions. This, they say, suggests that bubbles are caused in part by “maladaptive attempts to forecast the intentions of other players”. Mistaken buying happens when people believe that others will buy in the future.

For me, this evidence should reduce our fears of robot investing. It suggests that removing the human factor might help to stabilise markets. Yes, algorithmic trading can occasionally give us flash crashes. But it might avoid the big long-lasting bubbles and bursts caused by humans.

 

Bubbles are partly forecastable and might be traded successfully

This isn’t because we can reliably identify peaks in prices. We can’t. Researchers at Harvard University, however, have found some predictors of the probability of a crash. These include: increased price volatility; a greater volume and number of new share issues; and especially big price rises for newer shares. Of course, these indicators don’t help much with the question of whether Bitcoin is near the peak of a bubble. But they do suggest there isn’t a bubble in equities generally (although this is not to say that these cannot fall).

What’s more, there’s a rule that helps us manage bubbles. It’s the 10-month rule proposed by Meb Faber at Cambria Investment Management, which says we should buy when prices are above their 10-month average, and sell when they are below it. This doesn’t get you out that the top of the market. But it does save you from big losses as prices slide. Following this rule for tech stocks would have gotten us out of them in November 2000. We’d have suffered a 50 per cent loss. But we’d have saved ourselves the 90 per cent loss that followed. And we'd have ridden a lot of the bubble on the upside before then.

 

Bubbles remind us that textbook theories are too simple

One contributor to them is the fact that investors cannot easily short-sell overpriced assets. Because bubbly assets are volatile there’s a good chance they will rise even further. When they do, short-sellers face a margin call – the demand that they deposit more cash as collateral against further losses; the financial commentator Daniel Davies calls this “one of the most frightening things in the financial world”. The fear of this stops people short-selling.

Elementary textbooks underplay this. Efficient market theory says that information is quickly embedded into prices. It does not ask: what is the mechanism whereby this happens? Sometimes such a mechanism is absent; the fear of margin calls (or any other obstacle to short-selling) means that some information doesn’t enter the market. Market micro-structure – the question of who participates in markets, how and to what extent – matters more than elementary theory would have us believe.

 

Bubbles aren’t entirely a bad thing

In his book Pop! Daniel Gross argues that the low cost of capital and high expected returns that are associated with bubbles generate investment that we wouldn’t get if expectations were more realistic. The railway bubble of the 1840s gave us a railway network we wouldn’t otherwise have had (until Dr Beeching cut it), just as the tech bubble encouraged the spread of broadband. These were good legacies.

Also, the booms associated with bubbles provide useful learning and experience. We now have more and better software writers because young people entered a booming profession in the 1990s and learned quickly.

The same might be true of Bitcoin. Its rise to prominence is attracting attention and investment into blockchains – ways of storing data securely – which might well have useful spin-offs. And it is encouraging the Bank of England and other central banks to think about issuing their own digital currencies which might help improve the conduct of monetary policy in future – for example by allowing central banks to simply give people cash to boost demand in a recession.

What’s more, bubbles are driven in part by those most widespread of cognitive biases, overconfidence and wishful thinking. These aren’t wholly bad things. Without them, we would have very few entrepreneurs or artists. A world without the psychological impulses that give us bubbles would probably be a duller and poorer one.

 

The damage done by bursting bubbles varies hugely

The bursting of the tech bubble in 2001 led to the mildest of recessions in the US and to only a moderate slowdown in the UK. The bursting of the bubble in credit derivatives in 2007-08, by contrast, led to the worst crisis and recession since the 1930s even though the actual losses involved were much smaller.

Why the difference? It’s because losses on credit derivatives were concentrated in a few organisations which were key hubs in economic networks and which were mostly following similar strategies and so didn’t lay off risk with each other. Such losses therefore led to a systemic failure.

Losses on tech stocks, on the other hand, were dispersed among countless individuals, none of whom were central parts of any important network.

The world economy was thus much better able to survive the bursting of the tech bubble than the bursting of the credit bubble. Risk was better spread. It’s not the loss of wealth that matters when a bubble bursts, but how those losses are distributed.

Herein, perhaps, lies a reason to be relaxed about the Bitcoin bubble, if this is what it is. Jamie Dimon, CEO of JPMorgan, might have been wrong in October to call Bitcoin investors “stupid” – it has doubled in price since then – but his comment conveys an important fact. This is that banks are not speculating much if anything upon it. Instead, Bitcoin’s holders are probably much more like holders of tech stocks in 2000; they are retail investors. Their losses – should they occur – should be sufficiently dispersed so as not to do lasting economic harm.

Will they occur? My hunch is that they will, not least because Bitcoin has so many competitors. Hunches, though, have a fundamental value of close to zero. There’s always a bubble in opinions and guesswork.