The Bubble Triangle

When we started researching historical bubbles, we were surprised by how much the different episodes had in common. They always seemed to occur when there was abundant money or credit, they always seemed to occur in assets that had recently become much easier to buy and sell, and they always involved investors speculating on future price rises. Each bubble resulted from either a political initiative or a technological innovation.

Putting these elements together, we came up with the bubble triangle: a causal framework based on the ‘fire triangle' of oxygen, fuel, and heat. When each element is present, a fire can be started by a simple spark.

In the bubble triangle, the fuel is money and credit. If there is no money, borrowed or otherwise, to invest in a bubble, the bubble cannot grow.

The oxygen is marketability. If it is too difficult to buy and sell an asset, it’s no fun for momentum traders to speculate in. Highly illiquid assets may be overpriced, but they don’t typically experience a speculative bubble.

The heat is speculation. At the core of the bubble is investors buying an asset because they think its price will rise, rather than because they think it is inherently valuable. Just as a fire creates its own heat, early price rises attract more speculators, driving prices even higher.

The spark can come from one of two places. The first is politics. Governments can decide to create, or inadvertently create, a bubble for many reasons, the most common of which is to enrich the coalition of voters or power brokers they need to stay in power.

The second is technology. New technology can lead to widespread excitement and high profits for early adopters, which in turn often leads to a sharp rise in associated asset prices. When this attracts speculative traders, a bubble starts to form.

This framework isn’t perfect. The various elements often overlap - the spark might overlap with the fuel, for example, if politicians generate a bubble by making credit easier to obtain. The elements can’t be reduced to simple metrics, making the framework difficult to formally test. It is, however, an excellent starting point for understanding the causes of bubbles.